1 2017-11-09 Jim Wilson <jimw@sifive.com>
3 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
4 (scan_prog_file): Likewise.
6 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8 * bb-reorder.c (max_entry_frequency): Remove.
9 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
10 connect_traces, push_to_next_round_p): Remove prototypes.
11 (find_traces_1_round): Use counts only.
12 (push_to_next_round_p): Likewise.
13 (find_traces): Likewise.
14 (rotate_loop): Likewise.
15 (find_traces_1_round): Likewise.
16 (connect_traces): Likewise.
17 (edge_order): Likewise.
19 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
21 * config/arm/arm.c (output_return_instruction): Add comments to
22 indicate requirement for cmse_nonsecure_entry return to account
23 for the size of clearing instruction output here.
24 (thumb_exit): Likewise.
25 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
26 return in hardfloat mode.
28 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
30 * config/rs6000/rs6000.c (machine_function): Add a bool,
31 "toc_is_wrapped_separately".
32 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
33 if it wasn't explicitly set or unset, we are optimizing for speed, and
34 doing separate shrink-wrapping.
35 (rs6000_get_separate_components): Enable the TOC component if
36 saving the TOC register in the prologue.
37 (rs6000_components_for_bb): Handle the TOC component.
38 (rs6000_emit_prologue_components): Store the TOC register where needed.
39 (rs6000_set_handled_components): Mark TOC as handled, if handled.
40 (rs6000_emit_prologue): Don't save the TOC if that is already done.
42 2017-11-09 Martin Jambor <mjambor@suse.cz>
44 * ipa-param-manipulation.c: New file.
45 * ipa-param-manipulation.h: Likewise.
46 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
47 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
48 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
49 (ipa_parm_adjustment): Likewise.
50 (ipa_parm_adjustment_vec): Likewise.
51 (ipa_get_vector_of_formal_parms): Moved declaration to
52 ipa-param-manipulation.h.
53 (ipa_get_vector_of_formal_parm_types): Likewise.
54 (ipa_modify_formal_parameters): Likewise.
55 (ipa_modify_call_arguments): Likewise.
56 (ipa_combine_adjustments): Likewise.
57 (ipa_dump_param_adjustments): Likewise.
58 (ipa_modify_expr): Likewise.
59 (ipa_get_adjustment_candidate): Likewise.
60 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
61 ipa-param-manipulation.c.
62 (ipa_get_vector_of_formal_parm_types): Likewise.
63 (ipa_modify_formal_parameters): Likewise.
64 (ipa_modify_call_arguments): Likewise.
65 (ipa_modify_expr): Likewise.
66 (get_ssa_base_param): Likewise.
67 (ipa_get_adjustment_candidate): Likewise.
68 (index_in_adjustments_multiple_times_p): Likewise.
69 (ipa_combine_adjustments): Likewise.
70 (ipa_dump_param_adjustments): Likewise.
71 * tree-sra.c: Also include ipa-param-manipulation.h
72 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
75 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
76 Alan Hayward <alan.hayward@arm.com>
77 David Sherwood <david.sherwood@arm.com>
79 * doc/sourcebuild.texi (vect_masked_store): Document.
81 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
82 Alan Hayward <alan.hayward@arm.com>
83 David Sherwood <david.sherwood@arm.com>
85 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
87 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
88 Alan Hayward <alan.hayward@arm.com>
89 David Sherwood <david.sherwood@arm.com>
91 * doc/sourcebuild.texi (vect_variable_length): Document.
93 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
94 Alan Hayward <alan.hayward@arm.com>
95 David Sherwood <david.sherwood@arm.com>
97 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
99 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
100 Alan Hayward <alan.hayward@arm.com>
101 David Sherwood <david.sherwood@arm.com>
103 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
105 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
106 Alan Hayward <alan.hayward@arm.com>
107 David Sherwood <david.sherwood@arm.com>
109 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
110 previously undocumented selectors.
111 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
113 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
115 * doc/rtl.texi (const_vector): Say that elements can be
117 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
118 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
119 (gen_vec_duplicate): Use it instead of CONSTANT_P.
120 * optabs.c (expand_vector_broadcast): Likewise.
122 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
123 Alan Hayward <alan.hayward@arm.com>
124 David Sherwood <david.sherwood@arm.com>
126 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
127 scaled index even if the unscaled address was invalid.
128 Don't increase the complexity of using a scale in that case.
130 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
131 Alan Hayward <alan.hayward@arm.com>
132 David Sherwood <david.sherwood@arm.com>
134 * doc/rtl.texi: Rewrite the subreg rules so that they partition
135 the inner register into REGMODE_NATURAL_SIZE bytes rather than
136 UNITS_PER_WORD bytes.
137 * emit-rtl.c (validate_subreg): Divide subregs into blocks
138 based on REGMODE_NATURAL_SIZE of the inner mode.
139 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
140 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
141 * expmed.c (lowpart_bit_field_p): Divide the value up into
142 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
143 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
144 whether something is likely to occupy more than one register.
146 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
149 * ipa-inline-transform.c (update_noncloned_frequencies): Use
150 profile_count::adjust_for_ipa_scaling.
151 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
152 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
154 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
156 2017-11-09 Jakub Jelinek <jakub@redhat.com>
158 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
159 (split_group): Add total_orig and total_new arguments, estimate the
160 number of statements related to the store group without store merging
161 and with store merging.
162 (imm_store_chain_info::output_merged_store): Adjust split_group
163 callers, punt if estimated number of statements with store merging
164 is not smaller than estimated number of statements without it.
166 (handled_load): Remove has_single_use checks.
167 (pass_store_merging::process_store): Likewise.
169 2017-11-09 Richard Biener <rguenther@suse.de>
171 PR tree-optimization/82902
172 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
174 2017-11-09 Martin Liska <mliska@suse.cz>
177 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
180 2017-11-09 Martin Liska <mliska@suse.cz>
182 PR tree-optimization/82669
183 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
185 2017-11-09 Martin Liska <mliska@suse.cz>
187 PR gcov-profile/48463
188 * coverage.c (coverage_begin_function): Output also end locus
189 of a function and information whether the function is
191 * gcov-dump.c (tag_function): Parse and print the information.
192 * gcov.c (INCLUDE_MAP): Add include.
193 (INCLUDE_SET): Likewise.
194 (struct line_info): Move earlier in the source file because
195 of vector<line_info> in function_info structure.
196 (line_info::line_info): Likewise.
197 (line_info::has_block): Likewise.
198 (struct source_info): Add new member index.
199 (source_info::get_functions_at_location): New function.
200 (function_info::group_line_p): New function.
201 (output_intermediate_line): New function.
202 (output_intermediate_file): Use the mentioned function.
203 (struct function_start): New.
204 (struct function_start_pair_hash): Likewise.
205 (process_file): Add code that identifies group functions.
206 Assign lines either to global or function scope.
207 (generate_results): Skip artificial functions.
208 (find_source): Assign index for each source file.
209 (read_graph_file): Read new flag artificial and end_line.
210 (add_line_counts): Assign it either to global of function scope.
211 (accumulate_line_counts): Isolate core of the function to
212 accumulate_line_info and call it for both function and global
214 (accumulate_line_info): New function.
215 (output_line_beginning): Fix GNU coding style.
216 (print_source_line): New function.
217 (output_line_details): Likewise.
218 (output_function_details): Likewise.
219 (output_lines): Iterate both source (global) scope and function
221 (struct function_line_start_cmp): New class.
222 * doc/gcov.texi: Reflect changes in documentation.
224 2017-11-09 Jakub Jelinek <jakub@redhat.com>
227 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
228 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
229 and similarly for not instead of neg.
231 2017-11-08 Andi Kleen <ak@linux.intel.com>
233 * config/i386/i386.opt: Add -mforce-indirect-call.
234 * config/i386/predicates.md: Check for flag_force_indirect_call.
235 * doc/invoke.texi: Document -mforce-indirect-call
237 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
239 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
241 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
242 config/riscv/riscv.c (predict.h): New include.
243 (riscv_slow_unaligned_access_p): No longer static.
244 (riscv_block_move_straight): Add require.
245 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
247 2017-11-08 Jakub Jelinek <jakub@redhat.com>
250 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
251 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
252 nonimmediate_operand predicate for operand 1 instead of
255 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
257 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
259 * config/aarch64/constraints.md (Uml): New constraint.
260 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
263 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
265 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
266 of two vec_duplicates into a vec_concat.
268 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
270 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
271 Simplify vec_merge of vec_duplicate and vec_concat.
272 * config/aarch64/constraints.md (Utq): New constraint.
273 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
276 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
278 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
279 Simplify vec_merge of vec_duplicate and const_vector.
280 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
282 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
283 mode iterator. Update predicate on operand 1 to
284 handle non-const_vec constants. Delete constraints.
285 (*aarch64_combinez_be<mode>): Likewise for operand 2.
287 2017-11-08 Jakub Jelinek <jakub@redhat.com>
289 PR tree-optimization/78821
290 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
292 (store_operand_info::store_operand_info): Initialize it to false.
293 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
294 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
295 store in the group, and if chain_info is non-NULL, to ignore altogether
297 (compatible_load_p): Fail if bit_not_p does not match.
298 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
299 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
300 (pass_store_merging::process_store): Adjust
301 terminate_all_aliasing_chains calls to pass NULL in all current spots,
302 call terminate_all_aliasing_chains newly when adding a store into
303 a chain with non-NULL chain_info.
305 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
307 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
309 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
311 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
313 (aarch64_layout_frame): Initialise emit_frame_chain.
314 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
315 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
317 2017-11-08 Martin Liska <mliska@suse.cz>
319 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
320 of gimple_call_internal_p.
322 2017-11-07 Tom de Vries <tom@codesourcery.com>
324 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
326 2017-11-07 Tom de Vries <tom@codesourcery.com>
328 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
331 2017-11-08 Martin Liska <mliska@suse.cz>
334 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
336 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
338 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
340 2017-11-07 Jakub Jelinek <jakub@redhat.com>
343 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
344 (*cmp<mode>_ccz_1): New insn with $k alternative.
347 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
348 RTX_COMM_COMPARE as commutative as well.
349 (ix86_binary_operator_ok): Formatting fix.
350 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
351 *<code><mode>3<mask_name><round_saeonly_name>,
352 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
353 *<s>mul<mode>3_highpart<mask_name>,
354 *vec_widen_umult_even_v16si<mask_name>,
355 *vec_widen_umult_even_v8si<mask_name>,
356 *vec_widen_umult_even_v4si<mask_name>,
357 *vec_widen_smult_even_v16si<mask_name>,
358 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
359 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
360 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
361 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
362 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
363 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
364 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
365 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
366 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
367 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
368 ix86_binary_operator_ok. Formatting fixes.
369 (*<plusminus_insn><mode>3<mask_name><round_name>,
370 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
373 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
375 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
376 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
377 eq and ne if TARGET_ISEL.
378 (cmp): New code_iterator.
379 (UNS, UNSU_, UNSIK): New code_attrs.
380 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
381 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
383 ("eq<mode>3"): ... this.
384 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
386 ("ne<mode>3"): ... this.
388 2017-11-07 Julia Koval <julia.koval@intel.com>
391 * common/config/i386/i386-common.c
392 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
393 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
394 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
395 * config/i386/i386.opt: Ditto.
396 * config/i386/i386.c (ix86_target_string): Ditto.
397 (ix86_option_override_internal): Ditto.
398 (ix86_init_mpx_builtins): Move MPX to args2.
399 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
400 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
401 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
402 __builtin_ia32_vgf2p8affineinvqb_v32qi,
403 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
404 __builtin_ia32_vgf2p8affineinvqb_v16qi,
405 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
407 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
410 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
411 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
412 (zero-extendsidi peephole2): Remove peephole.
414 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
417 * stor-layout.c: Include attribs.h.
418 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
419 explicit lookup of "aligned" attribute.
421 2017-11-07 Andrew Waterman <andrew@sifive.com>
423 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
424 (riscv_expand_block_move): Likewise.
425 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
427 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
428 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
429 * config/riscv/riscv.c (riscv_block_move_straight): New function.
430 (riscv_adjust_block_mem): Likewise.
431 (riscv_block_move_loop): Likewise.
432 (riscv_expand_block_move): Likewise.
433 gcc/config/riscv/riscv.md (movmemsi): New pattern.
435 2017-11-07 Michael Clark <michaeljclark@mac.com>
437 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
438 (MUSL_DYNAMIC_LINKER): Likewise.
440 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
442 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
443 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
446 2017-11-07 Richard Biener <rguenther@suse.de>
448 * match.pd: Fix build.
450 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
451 Jackson Woodruff <jackson.woodruff@arm.com>
453 PR tree-optimization/71026
454 * match.pd: Canonicalize negate in division.
456 2017-11-07 Sudakshina Das <sudi.das@arm.com>
459 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
461 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
463 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
464 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
466 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
468 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
469 non-scalar integral types.
470 * match.pd (negate_expr_p): Handle MINUS_EXPR.
471 (-(A-B), -(~A)): New transformations.
473 2017-11-07 Tom de Vries <tom@codesourcery.com>
475 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
476 semicolon after "do {} while (0)".
477 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
478 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
479 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
480 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
481 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
483 2017-11-07 Tom de Vries <tom@codesourcery.com>
485 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
486 after "do {} while (0)".
487 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
488 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
489 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
490 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
491 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
493 2017-11-07 Tom de Vries <tom@codesourcery.com>
496 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
498 (arm_rtx_costs_internal): Add missing semicolon after
499 HANDLE_NARROW_SHIFT_ARITH call.
501 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
503 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
504 disable isel if it was not set explicitly.
506 2017-11-06 James Bowman <james.bowman@ftdichip.com>
508 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
510 (add_type_attribute) likewise.
512 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
514 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
515 of struct ix86_frame.
516 (ix86_initial_elimination_offset): Likewise.
517 (ix86_expand_split_stack_prologue): Likewise.
519 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
521 * tree-vrp.h (enum value_range_type): Update stale comment.
523 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
524 Alan Hayward <alan.hayward@arm.com>
525 David Sherwood <david.sherwood@arm.com>
527 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
528 (aarch64_expand_vec_perm_const): Take the number of units too.
529 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
530 (aarch64_expand_vec_perm_const): Likewise.
531 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
532 (vec_perm<mode>): Update accordingly.
534 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
535 Alan Hayward <alan.hayward@arm.com>
536 David Sherwood <david.sherwood@arm.com>
538 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
539 Take the number of units too.
540 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
541 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
542 but check for a vector mode before rather than after the call.
543 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
544 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
545 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
546 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
547 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
548 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
549 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
550 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
551 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
552 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
553 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
554 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
555 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
556 (aarch64_sqdmull2_n<mode>): Update accordingly.
558 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
559 Alan Hayward <alan.hayward@arm.com>
560 David Sherwood <david.sherwood@arm.com>
562 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
563 the number of units too.
564 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
565 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
566 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
567 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
568 (vec_store_lanesxi<mode>): Update accordingly.
570 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
571 Alan Hayward <alan.hayward@arm.com>
572 David Sherwood <david.sherwood@arm.com>
574 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
575 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
576 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
577 of units rather than the mode.
578 * config/aarch64/iterators.md (nunits): New mode attribute.
579 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
580 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
581 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
582 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
583 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
584 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
585 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
586 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
587 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
588 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
589 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
590 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
591 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
592 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
593 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
594 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
595 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
596 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
597 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
598 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
599 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
600 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
601 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
602 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
603 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
604 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
605 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
606 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
607 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
608 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
609 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
610 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
611 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
612 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
613 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
614 (aarch64_simd_vec_setv2di): Likewise.
616 2017-11-06 Carl Love <cel@us.ibm.com>
618 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
620 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
621 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
623 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
625 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
626 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
627 Add power 8 macro expansions.
628 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
629 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
630 power 8 instructions. (VSX_XXBR): Add iterator.
632 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
634 * config/arm/arm.md (predicable_short_it): Change default to "no",
635 improve documentation, remove uses that are identical to the default.
636 (enabled_for_depr_it): Rename to enabled_for_short_it.
637 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
638 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
639 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
640 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
641 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
643 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
646 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
647 float128 helper macros, which are no longer used after deleting
648 the old 'q' built-in functions, and moving the round to odd
649 built-in functions to being special built-in functions.
650 (BU_FLOAT128_2): Likewise.
651 (BU_FLOAT128_1_HW): Likewise.
652 (BU_FLOAT128_2_HW): Likewise.
653 (BU_FLOAT128_3_HW): Likewise.
654 (FABSQ): Delete old 'q' built-in functions.
655 (COPYSIGNQ): Likewise.
656 (SQRTF128_ODD): Move round to odd built-in functions to be
657 special built-in functions, so that we can handle
658 -mabi=ieeelongdouble.
659 (TRUNCF128_ODD): Likewise.
660 (ADDF128_ODD): Likewise.
661 (SUBF128_ODD): Likewise.
662 (MULF128_ODD): Likewise.
663 (DIVF128_ODD): Likewise.
664 (FMAF128_ODD): Likewise.
665 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
666 built-in names to 'f128'.
667 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
668 old 'q' built-in functions, as the machine independent code for
669 'f128' built-in functions handles this.
670 (rs6000_expand_builtin): Add expansion for float128 round to odd
671 functions, keying off on -mabi=ieeelongdouble of whether to use
672 the KFmode or TFmode variant.
673 (rs6000_init_builtins): Initialize the _Float128 round to odd
675 * doc/extend.texi (PowerPC Built-in Functions): Document the old
676 _Float128 'q' built-in functions are now mapped into the new
677 'f128' built-in functions.
679 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
681 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
682 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
684 2017-11-06 David Malcolm <dmalcolm@redhat.com>
687 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
688 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
689 * toplev.c: Include "ipa-fnsummary.h".
690 (toplev::finalize): Call ipa_fnsummary_c_finalize.
692 2017-11-06 Jakub Jelinek <jakub@redhat.com>
694 PR tree-optimization/82838
695 * gimple-ssa-store-merging.c
696 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
697 on a separate gimple_seq which is then appended to seq.
699 2017-11-06 Jeff Law <law@redhat.com>
702 * config/i386/i386.c (PROBE_INTERVAL): Remove.
703 (get_probe_interval): New functions.
704 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
705 (ix86_adjust_stack_and_probe): Likewise.
706 (output_adjust_stack_and_probe): Likewise.
707 (ix86_emit_probe_stack_range): Likewise.
708 (ix86_expand_prologue): Likewise.
710 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
712 PR tree-optimization/82816
713 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
714 if the modes of the two types are the same.
715 (convert_plusminus_to_widen): Likewise.
717 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
719 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
721 (usadv16qi): New define_expand.
722 (usadv8hi): New define_expand.
724 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
727 * ipa-inline-transform.c (update_noncloned_frequencies): Always
729 (inline_transform): Likewise.
730 * predict.c (counts_to_freqs): Remove useless conditional.
731 * profile-count.h (profile_count::apply_scale): Move sanity check.
732 * tree-inline.c (copy_bb): Always scale.
733 (copy_cfg_body): Likewise.
735 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
738 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
740 (*cmp_ite0): Add enabled_for_depr_it attribute.
741 (*cmp_ite1): Likewise.
743 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
745 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
748 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
750 * tree-vrp.c (vrp_int_const_binop): Return true on success and
751 return the value by pointer.
752 (extract_range_from_multiplicative_op_1): Update accordingly.
753 Return as soon as an operation fails.
755 2017-11-05 Tom de Vries <tom@codesourcery.com>
758 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
759 (DEF_SANITIZER_BUILTIN): ... here.
760 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
761 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
763 2017-11-05 Tom de Vries <tom@codesourcery.com>
766 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
768 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
769 ASM_OUTPUT_BEFORE_CASE_LABEL call.
770 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
772 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
773 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
774 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
776 2017-11-05 Tom de Vries <tom@codesourcery.com>
779 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
782 2017-11-04 Michael Clark <michaeljclark@mac.com>
784 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
785 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
787 (*addsi3_extended): Likewise.
788 (*addsi3_extended2): Likewise.
789 (<optab>si3): Likewise.
790 (<optab>di3): Likewise.
791 (<optab><mode>3): Likewise.
792 (<*optabe>si3_internal): Likewise.
793 (zero_extendqi<SUPERQI:mode>2): Likewise.
794 (*add<mode>hi3): Likewise.
795 (*xor<mode>hi3): Likewise.
796 (<optab>di3): Likewise.
797 (*<optab>si3_extend): Likewise.
798 (*sge<u>_<X:mode><GPR:mode>): Likewise.
799 (*slt<u>_<X:mode><GPR:mode>): Likewise.
800 (*sle<u>_<X:mode><GPR:mode>): Likewise.
802 2017-11-04 Andrew Waterman <andrew@sifive.com>
804 * config/riscv/riscv.c (riscv_option_override): Conditionally set
805 TARGET_STRICT_ALIGN based upon -mtune argument.
807 2017-11-04 Andrew Waterman <andrew@sifive.com>
809 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
811 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
813 config/i386/i386.c (choose_basereg): Use optional scratch
814 register and add assertion.
815 (x86_emit_outlined_ms2sysv_save): Use scratch register when
816 needed, and don't allocate stack.
817 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
818 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
819 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
821 2017-11-03 Jeff Law <law@redhat.com>
823 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
824 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
825 to probe at the start of a noreturn function.
827 2017-11-03 Jakub Jelinek <jakub@redhat.com>
829 PR tree-optimization/78821
830 * gimple-ssa-store-merging.c: Update the file comment.
831 (MAX_STORE_ALIAS_CHECKS): Define.
832 (struct store_operand_info): New type.
833 (store_operand_info::store_operand_info): New constructor.
834 (struct store_immediate_info): Add rhs_code and ops data members.
835 (store_immediate_info::store_immediate_info): Add rhscode, op0r
836 and op1r arguments to the ctor, initialize corresponding data members.
837 (struct merged_store_group): Add load_align_base and load_align
839 (merged_store_group::merged_store_group): Initialize them.
840 (merged_store_group::do_merge): Update them.
841 (merged_store_group::apply_stores): Pick the constant for
842 encode_tree_to_bitpos from one of the two operands, or skip
843 encode_tree_to_bitpos if neither operand is a constant.
844 (class pass_store_merging): Add process_store method decl. Remove
845 bool argument from terminate_all_aliasing_chains method decl.
846 (pass_store_merging::terminate_all_aliasing_chains): Remove
847 var_offset_p argument and corresponding handling.
848 (stmts_may_clobber_ref_p): New function.
849 (compatible_load_p): New function.
850 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
851 if there is overlap and rhs_code is not INTEGER_CST. For
852 non-overlapping stores terminate group if rhs is not mergeable.
853 (get_alias_type_for_stmts): Change first argument from
854 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
855 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
856 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
858 (get_location_for_stmts): Change first argument from
859 auto_vec<gimple *> & to vec<gimple *> &.
860 (struct split_store): Remove orig_stmts data member, add orig_stores.
861 (split_store::split_store): Create orig_stores rather than orig_stmts.
862 (find_constituent_stmts): Renamed to ...
863 (find_constituent_stores): ... this. Change second argument from
864 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
865 to info structures rather than the statements.
866 (split_group): Rename ALLOW_UNALIGNED argument to
867 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
868 it. Adjust find_constituent_stores caller.
869 (imm_store_chain_info::output_merged_store): Handle rhs_code other
870 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
871 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
872 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
873 (mem_valid_for_store_merging): New function.
874 (handled_load): New function.
875 (pass_store_merging::process_store): New method.
876 (pass_store_merging::execute): Use process_store method. Adjust
877 terminate_all_aliasing_chains caller.
879 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
881 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
882 Return true for more constants, symbols and label references.
883 (aarch64_valid_floating_const): Remove unused function.
885 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
888 * config/aarch64/aarch64.c (aarch64_layout_frame):
889 Undo forcing of LR at bottom of frame.
891 2017-11-03 Jeff Law <law@redhat.com>
894 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
895 for int_registers_saved.
897 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
898 extracted from tree-ssa-dom.c.
899 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
900 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
901 (record_equivalences_from_incoming_edge): Add additional argument
902 to single_pred_edge_ignoring_loop_edges call.
903 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
904 (uncprop_dom_walker::before_dom_children): Add additional argument
905 to single_pred_edge_ignoring_loop_edges call.
906 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
907 single_pred_edge_ignoring_loop_edges rather than open coding.
908 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
910 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
912 * match.pd (-(-A)): Rewrite.
914 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
916 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
917 (rs6000_emit_int_cmove): New declaration.
918 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
919 (rs6000_emit_sISEL): Delete.
920 (rs6000_emit_int_cmove): Make non-static.
921 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
922 instead of rs6000_emit_sISEL.
924 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
926 * asan.c (create_cond_insert_point): Maintain profile.
927 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
929 * basic-block.h (struct basic_block_def): Remove frequency.
930 (EDGE_FREQUENCY): Use to_frequency
931 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
933 (find_traces): Update to use to_frequency.
934 (find_traces_1_round): Likewise; use only IPA counts.
935 (bb_to_key): Likewise.
936 (connect_traces): Use IPA counts only.
937 (copy_bb_p): Update to use to_frequency.
938 (fix_up_crossing_landing_pad): Likewise.
939 (sanitize_hot_paths): Likewise.
940 * bt-load.c (basic_block_freq): Likewise.
941 * cfg.c (init_flow): Set count_max to uninitialized.
942 (check_bb_profile): Remove frequencies; check counts.
943 (dump_bb_info): Do not dump frequencies.
944 (update_bb_profile_for_threading): Update counts only.
945 (scale_bbs_frequencies_int): Likewise.
946 (MAX_SAFE_MULTIPLIER): Remove.
947 (scale_bbs_frequencies_gcov_type): Update counts only.
948 (scale_bbs_frequencies_profile_count): Update counts only.
949 (scale_bbs_frequencies): Update counts only.
950 * cfg.h (struct control_flow_graph): Add count-max.
951 (update_bb_profile_for_threading): Update prototype.
952 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
953 (find_many_sub_basic_blocks): Likewise.
954 * cfgcleanup.c (try_forward_edges): Likewise.
955 (try_crossjump_to_edge): Likewise.
956 * cfgexpand.c (expand_gimple_cond): Likewise.
957 (expand_gimple_tailcall): Likewise.
958 (construct_init_block): Likewise.
959 (construct_exit_block): Likewise.
960 * cfghooks.c (verify_flow_info): Check consistency of counts.
961 (dump_bb_for_graph): Do not dump frequencies.
962 (split_block_1): Do not update frequencies.
963 (split_edge): Do not update frequencies.
964 (make_forwarder_block): Do not update frequencies.
965 (duplicate_block): Do not update frequencies.
966 (account_profile_record): Do not update frequencies.
967 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
968 for global heuristics.
969 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
970 (expected_loop_iterations_unbounded): Use counts only.
971 * cfgloopmanip.c (scale_loop_profile): Simplify.
972 (create_empty_loop_on_edge): Simplify
974 (duplicate_loop_to_header_edge): Simplify
975 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
976 (update_br_prob_note): Take care of removing note when profile
978 (relink_block_chain): Do not dump frequency.
979 (rtl_account_profile_record): Use to_frequency.
980 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
981 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
982 (cgraph_update_edges_for_call_stmt_node): Likewise.
983 (cgraph_edge::verify_count_and_frequency): Update.
984 (cgraph_node::verify_node): Temporarily disable frequency verification.
985 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
987 (cgraph_edge::rebuild_edges): Convert to ipa counts.
988 * cgraphunit.c (init_lowered_empty_function): Do not initialize
990 (cgraph_node::expand_thunk): Update profile.
991 * except.c (dw2_build_landing_pads): Do not update frequency.
992 * final.c (compute_alignments): Use to_frequency.
993 (dump_basic_block_info): Do not dump frequency.
994 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
995 (dump_gimple_bb_header): Do not dump frequency.
996 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
998 * gimple-streamer-in.c (input_bb): Do not stream frequency.
999 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1000 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1001 (init_before_recovery): Do not update frequency.
1002 (sched_create_recovery_edges): Do not update frequency.
1003 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1004 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1005 (ipa_cp_c_finalize): Set max_count to uninitialized.
1006 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1007 (param_change_prob): Use counts.
1008 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1010 * ipa-split.c (consider_split): Use to_frequency.
1011 (split_function): Use to_frequency.
1012 * ira-build.c (loop_compare_func): Likewise.
1013 (mark_loops_for_removal): Likewise.
1014 (mark_all_loops_for_removal): Likewise.
1015 * loop-doloop.c (doloop_modify): Do not update frequency.
1016 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1018 * lto-streamer-in.c (input_function): Update count_max.
1019 * omp-expand.c (expand_omp_taskreg): Update count_max.
1020 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1021 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1022 (maybe_hot_count_p): Use ipa counts only.
1023 (maybe_hot_bb_p): Simplify.
1024 (maybe_hot_edge_p): Simplify.
1025 (probably_never_executed): Do not take frequency argument.
1026 (probably_never_executed_bb_p): Do not pass frequency.
1027 (probably_never_executed_edge_p): Likewise.
1028 (combine_predictions_for_bb): Check that profile is nonzero.
1029 (propagate_freq): Do not set frequency.
1030 (drop_profile): Simplify.
1031 (counts_to_freqs): Simplify.
1032 (expensive_function_p): Use to_frequency.
1033 (propagate_unlikely_bbs_forward): Simplify.
1034 (determine_unlikely_bbs): Simplify.
1035 (estimate_bb_frequencies): Add hack to silence graphite issues.
1036 (compute_function_frequency): Use ipa counts.
1037 (pass_profile::execute): Update.
1038 (rebuild_frequencies): Use counts only.
1039 (force_edge_cold): Use counts only.
1040 * profile-count.c (profile_count::dump): Dump new count types.
1041 (profile_count::differs_from_p): Check compatiblity.
1042 (profile_count::to_frequency): New function.
1043 (profile_count::to_cgraph_frequency): New function.
1044 * profile-count.h (struct function): Declare.
1045 (enum profile_quality): Add profile_guessed_local and
1046 profile_guessed_global0.
1047 (class profile_proability): Decrease number of bits to 29;
1048 update from_reg_br_prob_note and to_reg_br_prob_note.
1049 (class profile_count: Update comment; decrease number of bits
1050 to 61. Check compatibility.
1051 (profile_count::compatible_p): New private member function.
1052 (profile_count::ipa_p): New member function.
1053 (profile_count::operator<): Handle global zero correctly.
1054 (profile_count::operator>): Handle global zero correctly.
1055 (profile_count::operator<=): Handle global zero correctly.
1056 (profile_count::operator>=): Handle global zero correctly.
1057 (profile_count::nonzero_p): New member function.
1058 (profile_count::force_nonzero): New member function.
1059 (profile_count::max): New member function.
1060 (profile_count::apply_scale): Handle IPA scalling.
1061 (profile_count::guessed_local): New member function.
1062 (profile_count::global0): New member function.
1063 (profile_count::ipa): New member function.
1064 (profile_count::to_frequency): Declare.
1065 (profile_count::to_cgraph_frequency): Declare.
1066 * profile.c (OVERLAP_BASE): Delete.
1067 (compute_frequency_overlap): Delete.
1068 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1069 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1070 * sched-ebb.c (rank): Use counts only.
1071 * shrink-wrap.c (handle_simple_exit): Use counts only.
1072 (try_shrink_wrapping): Use counts only.
1073 (place_prologue_for_one_component): Use counts only.
1074 * tracer.c (find_best_predecessor): Use to_frequency.
1075 (find_trace): Use to_frequency.
1076 (tail_duplicate): Use to_frequency.
1077 * trans-mem.c (expand_transaction): Do not update frequency.
1078 * tree-call-cdce.c: Do not update frequency.
1079 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1080 (gimple_merge_blocks): Likewise.
1081 (gimple_split_edge): Likewise.
1082 (gimple_duplicate_sese_region): Likewise.
1083 (gimple_duplicate_sese_tail): Likewise.
1084 (move_sese_region_to_fn): Likewise.
1085 (gimple_account_profile_record): Likewise.
1086 (insert_cond_bb): Likewise.
1087 * tree-complex.c (expand_complex_div_wide): Likewise.
1088 * tree-eh.c (lower_resx): Update profile.
1089 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1091 (initialize_cfun): Do not initialize frequencies
1092 (freqs_to_counts): Delete.
1093 (copy_cfg_body): Ignore count parameter.
1094 (copy_body): Update.
1095 (expand_call_inline): Update count_max.
1096 (optimize_inline_calls): Update count_max.
1097 (tree_function_versioning): Update count_max.
1098 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1099 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1101 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1102 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1103 (try_peel_loop): Likewise.
1104 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1106 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1107 (tree_transform_and_unroll_loop): Do not use frequencies
1108 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1109 Use reliable prediction only.
1110 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1111 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1112 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1113 probability scaling.
1114 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1115 not update frequency
1116 (any_remaining_duplicated_blocks): Likewise.
1117 (update_profile): Likewise.
1118 (estimated_freqs_path): Delete.
1119 (freqs_to_counts_path): Delete.
1120 (clear_counts_path): Delete.
1121 (ssa_fix_duplicate_block_edges): Likewise.
1122 (duplicate_thread_path): Likewise.
1123 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1124 * tree-tailcall.c (decrease_profile): Do not update frequency.
1125 (eliminate_tail_call): Likewise.
1126 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1127 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1128 (optimize_mask_stores): Likewise.
1129 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1130 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1131 (ubsan_expand_ptr_ifn): Update profile.
1132 * value-prof.c (gimple_ic): Simplify.
1133 * value-prof.h (gimple_ic): Update prototype.
1134 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1135 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1137 (want_inline_self_recursive_call_p): Likewise.
1138 (resolve_noninline_speculation): Only cummulate defined counts.
1139 (inline_small_functions): Use nonzero_p.
1140 (ipa_inline): Do not access freed node.
1142 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1144 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1145 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1147 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1149 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1150 non-legitimate address.
1152 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1154 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1155 (*lt0_<mode>di, *lt0_<mode>si): New.
1157 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1159 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1160 TARGET_PAIRED_FLOAT.
1162 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1163 Jim Wilson <jim.wilson@linaro.org>
1165 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1166 * config/aarch64/aarch64-tune.md: Regenerated.
1167 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1168 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1170 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1172 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
1173 (arc_expand_prologue): Restore blink for millicode.
1174 * config/arc/linux.h (LINK_EH_SPEC): Defined.
1176 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
1179 * config/i386/i386.c (ix86_vector_duplicate_value): Use
1180 gen_vec_duplicate after forcing the scalar into a register.
1182 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
1184 * combine (try_combine): Print the insns input to try_combine to the
1187 2017-11-02 Steve Ellcey <sellcey@cavium.com>
1190 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
1191 Remove second argument from aarch64_process_target_attr call.
1192 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
1194 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
1196 (aarch64_handle_attr_arch): Remove second argument.
1197 (aarch64_handle_attr_cpu): Ditto.
1198 (aarch64_handle_attr_tune): Ditto.
1199 (aarch64_handle_attr_isa_flags): Ditto.
1200 (aarch64_process_one_target_attr): Ditto.
1201 (aarch64_process_target_attr): Ditto.
1202 (aarch64_option_valid_attribute_p): Remove second argument.
1203 on aarch64_process_target_attr call.
1205 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1207 * diagnostic.c: Include "selftest-diagnostic.h".
1208 (selftest::assert_location_text): New function.
1209 (selftest::test_diagnostic_get_location_text): New function.
1210 (selftest::diagnostic_c_tests): Call it.
1212 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1214 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
1215 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
1216 (class selftest::test_diagnostic_context): Move to...
1217 * selftest-diagnostic.c: New file.
1218 * selftest-diagnostic.h: New file.
1220 2017-11-02 James Bowman <james.bowman@ftdichip.com>
1222 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
1223 offset range for FT32B.
1224 * config/ft32/ft32.h: option "mcompress" enables relaxation.
1225 * config/ft32/ft32.md: Add TARGET_NOPM.
1226 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
1227 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
1229 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
1231 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
1233 2017-11-02 Jeff Law <law@redhat.com>
1235 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1236 virtual keyword on FINAL OVERRIDE members.
1238 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1239 virtuals together. Add virtual destructor.
1240 (substitute_and_fold_engine): Similarly.
1242 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1244 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1246 2017-11-02 Richard Biener <rguenther@suse.de>
1248 PR tree-optimization/82795
1249 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1251 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1253 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1255 (gcc_GAS_CHECK_FEATURE): Remove.
1256 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1257 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1258 * configure: Regenerate.
1260 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1262 * config/arc/arc.c (hwloop_optimize): Account for empty
1265 2017-11-02 Richard Biener <rguenther@suse.de>
1268 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1269 Truncate ARRAY_REF index and element size.
1271 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1273 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1275 2017-11-01 Jeff Law <law@redhat.com>
1277 * tree-ssa-ccp.c (ccp_folder): New class derived from
1278 substitute_and_fold_engine.
1279 (ccp_folder::get_value): New member function.
1280 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1281 (ccp_fold_stmt): Remove prototype.
1282 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1283 * tree-ssa-copy.c (copy_folder): New class derived from
1284 substitute_and_fold_engine.
1285 (copy_folder::get_value): Renamed from get_value.
1286 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1287 * tree-vrp.c (vrp_folder): New class derived from
1288 substitute_and_fold_engine.
1289 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1290 (vrp_folder::get_value): New member function.
1291 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1292 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1293 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1294 provide a class interface to folder/substitute routines.
1295 (ssa_prop_fold_stmt_fn): Remove typedef.
1296 (ssa_prop_get_value_fn): Likewise.
1297 (subsitute_and_fold): Remove prototype.
1298 (replace_uses_in): Likewise.
1299 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1300 Renamed from replace_uses_in. Call the virtual member function
1301 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1302 (substitute_and_fold_dom_walker): Remove initialization of
1303 data member entries for calbacks. Add substitute_and_fold_engine
1304 member and initialize it.
1305 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1306 member functions for get_value, replace_phi_args_in c
1307 replace_uses_in, and fold_stmt calls.
1308 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1309 substitute_and_fold. Remove assert. Update ctor call.
1311 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1312 (ssa_prop_visit_phi_fn): Likewise.
1313 (class ssa_propagation_engine): New class to provide an interface
1315 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1317 (ssa_prop_visit_phi): Likewise.
1318 (ssa_propagation_engine::simulate_stmt): Moved into class.
1319 Call visit_phi/visit_stmt from the class rather than via
1320 file scoped static variables.
1321 (ssa_propagation_engine::simulate_block): Moved into class.
1322 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1323 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1324 set file scoped statics for the visit_stmt/visit_phi callbacks.
1325 * tree-complex.c (complex_propagate): New class derived from
1326 ssa_propagation_engine.
1327 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1328 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1329 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1331 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1332 ssa_propagation_engine.
1333 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1334 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1335 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1336 * tree-ssa-copy.c (copy_prop): New class derived from
1337 ssa_propagation_engine.
1338 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1339 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1340 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1341 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1342 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1343 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1344 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1346 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1348 PR rtl-optimization/82778
1349 PR rtl-optimization/82597
1350 * compare-elim.c (struct comparison): Add in_a_setter field.
1351 (find_comparison_dom_walker::before_dom_children): Remove killed
1352 bitmap and df_simulate_find_defs call, instead walk the defs.
1353 Compute last_setter and initialize in_a_setter. Merge definitions
1354 with first initialization for a few variables.
1355 (try_validate_parallel): Use insn_invalid_p instead of
1356 recog_memoized. Return insn rather than just the pattern.
1357 (try_merge_compare): Fix up comment. Don't uselessly test if
1358 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1360 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1363 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1364 Alan Hayward <alan.hayward@arm.com>
1365 David Sherwood <david.sherwood@arm.com>
1367 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1368 aarch64_hard_regno_nregs to get the number of registers
1371 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1372 Alan Hayward <alan.hayward@arm.com>
1373 David Sherwood <david.sherwood@arm.com>
1375 * config/aarch64/constraints.md (Upl): Rename to...
1377 * config/aarch64/aarch64.md
1378 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1381 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1382 Alan Hayward <alan.hayward@arm.com>
1383 David Sherwood <david.sherwood@arm.com>
1385 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1386 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1389 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1390 Alan Hayward <alan.hayward@arm.com>
1391 David Sherwood <david.sherwood@arm.com>
1393 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1394 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1395 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1397 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1398 of a single element.
1399 * config/aarch64/iterators.md: Add a comment above the permute
1400 unspecs to say that they are generated directly by
1401 aarch64_expand_vec_perm_const.
1402 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1404 2017-11-01 Nathan Sidwell <nathan@acm.org>
1406 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1408 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1410 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1411 and medany code models, and describe what they do.
1413 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1415 Revert accidental duplicate:
1417 * combine.c (can_change_dest_mode): Reject changes in
1418 REGMODE_NATURAL_SIZE.
1420 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1422 PR rtl-optimization/64682
1423 PR rtl-optimization/69567
1424 PR rtl-optimization/69737
1425 PR rtl-optimization/82683
1426 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1427 register mentioned in the note, drop the note, unless it came from I3,
1428 in which case it should go to I3 again.
1430 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1432 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1433 and return false if not.
1434 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1436 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1438 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1439 range is known to be empty.
1441 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1442 Alan Hayward <alan.hayward@arm.com>
1443 David Sherwood <david.sherwood@arm.com>
1445 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1446 and CONST_VECTOR_NUNITS instead of computing the number of units from
1447 the byte sizes of the vector and element.
1448 (simplify_binary_operation_1): Likewise.
1449 (simplify_const_binary_operation): Likewise.
1450 (simplify_ternary_operation): Likewise.
1452 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1453 Alan Hayward <alan.hayward@arm.com>
1454 David Sherwood <david.sherwood@arm.com>
1456 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1457 (int_mem_offset): ...this new function.
1458 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1459 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1460 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1463 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1464 Alan Hayward <alan.hayward@arm.com>
1465 David Sherwood <david.sherwood@arm.com>
1467 * lower-subreg.c (interesting_mode_p): New function.
1468 (compute_costs, find_decomposable_subregs, decompose_register)
1469 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1470 (resolve_clobber, dump_choices): Use it.
1472 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1473 Alan Hayward <alan.hayward@arm.com>
1474 David Sherwood <david.sherwood@arm.com>
1476 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
1478 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1479 Alan Hayward <alan.hayward@arm.com>
1480 David Sherwood <david.sherwood@arm.com>
1482 * alias.c (find_base_value, find_base_term): Only process integer
1483 truncations. Check the precision rather than the size.
1485 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1486 Alan Hayward <alan.hayward@arm.com>
1487 David Sherwood <david.sherwood@arm.com>
1489 * machmode.h (is_narrower_int_mode): New function
1490 * optabs.c (expand_float, expand_fix): Use it.
1491 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1493 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1494 Alan Hayward <alan.hayward@arm.com>
1495 David Sherwood <david.sherwood@arm.com>
1497 * rtl.h (narrower_subreg_mode): New function.
1498 * ira-color.c (update_costs_from_allocno): Use it.
1500 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1501 Alan Hayward <alan.hayward@arm.com>
1502 David Sherwood <david.sherwood@arm.com>
1504 * optabs-query.h (convert_optab_p): New function, split out from...
1505 (convert_optab_handler): ...here.
1506 (widening_optab_handler): Delete.
1507 (find_widening_optab_handler): Remove permit_non_widening parameter.
1508 (find_widening_optab_handler_and_mode): Likewise. Provide an
1509 override that operates on mode class wrappers.
1510 * optabs-query.c (widening_optab_handler): Delete.
1511 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1512 parameter. Assert that the two modes are the same class and that
1513 the "from" mode is narrower than the "to" mode. Use
1514 convert_optab_handler instead of widening_optab_handler.
1515 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1516 instead of widening_optab_handler.
1517 * expr.c (expand_expr_real_2): Update calls to
1518 find_widening_optab_handler.
1519 * optabs.c (expand_widen_pattern_expr): Likewise.
1520 (expand_binop_directly): Take the insn_code as a parameter.
1521 (expand_binop): Only call find_widening_optab_handler for
1522 conversion optabs; use optab_handler otherwise. Update calls
1523 to find_widening_optab_handler and expand_binop_directly.
1524 Use convert_optab_handler instead of widening_optab_handler.
1525 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1526 find_widening_optab_handler and use scalar_mode rather than
1528 (convert_plusminus_to_widen): Likewise.
1530 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1531 Alan Hayward <alan.hayward@arm.com>
1532 David Sherwood <david.sherwood@arm.com>
1534 * machmode.h (fixed_size_mode): New class.
1535 * rtl.h (get_pool_mode): Return fixed_size_mode.
1536 * gengtype.c (main): Add fixed_size_mode.
1537 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1538 (get_raw_arg_mode): Likewise.
1539 * doc/tm.texi: Regenerate.
1540 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1541 * targhooks.c (default_get_reg_raw_mode): Likewise.
1542 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1543 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1544 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1545 (msp430_get_raw_result_mode): Likewise.
1546 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1547 * dbxout.c (dbxout_parms): Require fixed-size modes.
1548 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1549 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1550 * omp-low.c (lower_oacc_reductions): Likewise.
1551 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1552 (simplify_subreg): Update accordingly.
1553 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1554 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1555 that aren't fixed-size.
1556 (get_pool_mode): Return a fixed_size_mode.
1557 (output_constant_pool_2): Take a fixed_size_mode.
1559 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1560 Alan Hayward <alan.hayward@arm.com>
1561 David Sherwood <david.sherwood@arm.com>
1563 * doc/rtl.texi (vec_series): Document.
1564 (const): Say that the operand can be a vec_series.
1565 * rtl.def (VEC_SERIES): New rtx code.
1566 * rtl.h (const_vec_series_p_1): Declare.
1567 (const_vec_series_p): New function.
1568 * emit-rtl.h (gen_const_vec_series): Declare.
1569 (gen_vec_series): Likewise.
1570 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1571 (gen_vec_series): Likewise.
1572 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1573 * simplify-rtx.c (simplify_unary_operation): Handle negations
1575 (simplify_binary_operation_series): New function.
1576 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1577 (test_vector_ops_series): New function.
1578 (test_vector_ops): Call it.
1579 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1580 gen_const_vec_series.
1581 (altivec_lvsr): Likewise.
1582 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1584 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1585 Alan Hayward <alan.hayward@arm.com>
1586 David Sherwood <david.sherwood@arm.com>
1588 * doc/rtl.texi (const): Update description of address constants.
1589 Say that vector constants are allowed too.
1590 * common.md (E, F): Use CONSTANT_P instead of checking for
1592 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1593 checking for CONST_VECTOR.
1594 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1596 * expr.c (expand_expr_real_2): Check for vector modes instead
1597 of checking for CONST_VECTOR.
1598 * rtl.h (const_vec_p): New function.
1599 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1600 (unwrap_const_vec_duplicate): Handle them here too.
1602 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1603 David Malcolm <dmalcolm@redhat.com>
1604 Alan Hayward <alan.hayward@arm.com>
1605 David Sherwood <david.sherwood@arm.com>
1607 * rtl.h (vec_duplicate_p): New function.
1608 * selftest-rtl.c (assert_rtx_eq_at): New function.
1609 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1610 (assert_rtx_eq_at): Declare.
1611 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1612 * selftest-run-tests.c (selftest::run_tests): Call it.
1613 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1614 (simplify_unary_operation_1): Recursively handle vector duplicates.
1615 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1617 (simplify_subreg): Handle subregs of vector duplicates.
1618 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1619 (selftest::simplify_rtx_c_tests): New functions.
1621 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1622 Alan Hayward <alan.hayward@arm.com>
1623 David Sherwood <david.sherwood@arm.com>
1625 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1626 (gen_vec_duplicate): Likewise.
1627 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1629 (gen_const_vector): ...here.
1630 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1631 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1632 whose elements are all equal.
1633 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1634 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1635 (simplify_relational_operation): Likewise.
1636 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1638 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1639 (aarch64_expand_vector_init): Likewise.
1640 * config/arm/arm.c (neon_vdup_constant): Likewise.
1641 (neon_expand_vector_init): Likewise.
1642 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1643 (arm_block_set_unaligned_vect): Likewise.
1644 (arm_block_set_aligned_vect): Likewise.
1645 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1646 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1647 (expand_vec_perm_even_odd_pack): Likewise.
1648 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1649 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1650 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1651 gen_const_vec_duplicate.
1652 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1653 * config/mips/mips.c (mips_gen_const_int_vector): Use
1654 gen_const_vec_duplicate.
1655 (mips_expand_vector_init): Use CONST0_RTX.
1656 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1657 (define_split): Use gen_const_vec_duplicate.
1658 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1659 (define_split): Use gen_const_vec_duplicate.
1660 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1661 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1662 * config/spu/spu.c (spu_const): Likewise.
1664 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1665 Alan Hayward <alan.hayward@arm.com>
1666 David Sherwood <david.sherwood@arm.com>
1668 * combine.c (can_change_dest_mode): Reject changes in
1669 REGMODE_NATURAL_SIZE.
1671 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
1673 * configure.ac (--enable-libssp): New.
1674 (gcc_cv_libc_provides_ssp): Check for explicit setting before
1675 trying to determine target-specific default. Adjust indentation.
1676 * configure: Regenerated.
1677 * doc/install.texi (Configuration): Expand --disable-libssp
1680 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
1682 config/i386/i386.c (ix86_expand_epilogue): Correct stack
1685 2017-10-31 Martin Jambor <mjambor@suse.cz>
1688 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1690 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1692 * auto-profile.c (autofdo_source_profile::read): Use
1693 UNKNOWN_LOCATION rather than 0.
1694 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1695 (warning_at): ...this overload.
1696 (warning_at_rich_loc_n): Rename to...
1697 (warning_n): ...this overload.
1698 (error_at_rich_loc): Rename to...
1699 (error_at): ...this overload.
1700 (pedwarn_at_rich_loc): Rename to...
1701 (pedwarn): ...this overload.
1702 (permerror_at_rich_loc): Rename to...
1703 (permerror): ...this overload.
1704 (inform_at_rich_loc): Rename to...
1705 (inform): ...this overload.
1706 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1707 (diagnostic_n_impl_richloc): Rename to...
1708 (diagnostic_n_impl): ...this rich_location *-based decl.
1709 (inform_at_rich_loc): Rename to...
1710 (inform): ...this, and add an assertion.
1711 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1712 (warning_at_rich_loc): Rename to...
1713 (warning_at): ...this, and add an assertion.
1714 (warning_at_rich_loc_n): Rename to...
1715 (warning_n): ...this, and add an assertion.
1716 (warning_n): Update location_t-based implementation for removal of
1717 location_t-based diagnostic_n_impl.
1718 (pedwarn_at_rich_loc): Rename to...
1719 (pedwarn): ...this, and add an assertion.
1720 (permerror_at_rich_loc): Rename to...
1721 (permerror): ...this, and add an assertion.
1722 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1723 (error_at_rich_loc): Rename to...
1724 (error_at): ...this, and add an assertion.
1725 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1726 (driver::do_spec_on_infiles): Likewise.
1727 * substring-locations.c (format_warning_va): Update for renaming
1728 of inform_at_rich_loc.
1730 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1732 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1733 _Float<N>X built-in functions so that the variant without the
1734 "__builtin_" prefix is only enabled for the GNU C and Objective C
1735 languages when they are in non-strict ANSI/ISO mode.
1736 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1737 * target.def (floatn_builtin_p): Add a target hook to control
1738 whether _Float<N> and _Float<N>X built-in functions without the
1739 "__builtin_" prefix are enabled, and return true for C and
1740 Objective C in the default hook. Include langhooks.h in
1742 * targhooks.h (default_floatn_builtin_p): Likewise.
1743 * targhooks.c (default_floatn_builtin_p): Likewise.
1744 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1745 floatn_builtin_p target hook.
1746 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1748 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1749 Eric Botcazou <ebotcazou@adacore.com>
1751 PR rtl-optimization/81803
1752 * lra-constraints.c (curr_insn_transform): Also reload the whole
1753 register for a strict subreg no wider than a word if this is for
1754 a WORD_REGISTER_OPERATIONS target.
1756 2017-10-31 Jason Merrill <jason@redhat.com>
1758 * gdbinit.in: Skip over inlines from timevar.h.
1760 2017-10-31 Martin Liska <mliska@suse.cz>
1762 * doc/gcov.texi: Document new option.
1763 * gcov.c (print_usage): Likewise print it.
1764 (process_args): Support the argument.
1765 (format_count): New function.
1766 (format_gcov): Use the function.
1768 2017-10-31 Martin Liska <mliska@suse.cz>
1770 * gcov.c (struct name_map): do not use typedef.
1771 Define operator== and operator<.
1772 (name_search): Remove.
1773 (name_sort): Remove.
1774 (main): Do not allocate names.
1775 (process_file): Add vertical space.
1776 (generate_results): Use std::find.
1777 (release_structures): Do not release memory.
1778 (find_source): Use std::find.
1780 2017-10-31 Martin Liska <mliska@suse.cz>
1782 * gcov.c (struct line_info): Remove it's typedef.
1783 (line_info::line_info): Add proper ctor.
1784 (line_info::has_block): Do not use a typedef.
1785 (struct source_info): Do not use typedef.
1786 (circuit): Likewise.
1787 (get_cycles_count): Likewise.
1788 (output_intermediate_file): Iterate via vector iterator.
1789 (add_line_counts): Use std::vector methods.
1790 (accumulate_line_counts): Likewise.
1791 (output_lines): Likewise.
1793 2017-10-31 Martin Liska <mliska@suse.cz>
1795 * gcov.c (struct source_info): Remove typedef.
1796 (source_info::source_info): Add proper ctor.
1797 (accumulate_line_counts): Use struct, not it's typedef.
1798 (output_gcov_file): Likewise.
1799 (output_lines): Likewise.
1800 (main): Do not allocate an array.
1801 (output_intermediate_file): Use size of vector container.
1802 (process_file): Resize the vector.
1803 (generate_results): Do not preallocate, use newly added vector
1805 (release_structures): Do not release sources.
1806 (find_source): Use vector methods.
1807 (add_line_counts): Do not use typedef.
1809 2017-10-31 Martin Liska <mliska@suse.cz>
1811 * doc/gcov.texi: Document that.
1812 * gcov.c (add_line_counts): Mark lines with a non-executed
1814 (output_line_beginning): Handle such lines.
1815 (output_lines): Pass new argument.
1816 (output_intermediate_file): Print it in intermediate format.
1818 2017-10-31 Martin Liska <mliska@suse.cz>
1820 * color-macros.h: New file.
1821 * diagnostic-color.c: Factor out color related to macros to
1823 * doc/gcov.texi: Document -k option.
1824 * gcov.c (INCLUDE_STRING): Include string.h.
1825 (print_usage): Add -k option.
1826 (process_args): Parse it.
1827 (pad_count_string): New function.
1828 (output_line_beginning): Likewise.
1829 (DEFAULT_LINE_START): New macro.
1830 (output_lines): Support color output.
1832 2017-10-31 Martin Liska <mliska@suse.cz>
1834 PR gcov-profile/82633
1835 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1836 their interaction with GCOV infrastructure.
1837 * configure.ac: Add -fkeep-{inline,static}-functions to
1839 * configure: Regenerate.
1841 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1844 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1846 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1849 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1850 into a register if it does not fit into an immediate offset field.
1852 2017-10-31 Olivier Hainque <hainque@adacore.com>
1854 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1856 2017-10-31 Julia Koval <julia.koval@intel.com>
1858 * config.gcc: Add gfniintrin.h.
1859 * config/i386/gfniintrin.h: New.
1860 * config/i386/i386-builtin-types.def
1861 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1862 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1863 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1864 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1865 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1866 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1867 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1868 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1869 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1870 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1871 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1872 * config/i386/immintrin.h: Include gfniintrin.h.
1873 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1875 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1877 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1879 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1881 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1882 (arm_ashldi3_1bit): Remove pattern.
1883 (ashrdi3): Remove shift by 1 expansion.
1884 (arm_ashrdi3_1bit): Remove pattern.
1885 (lshrdi3): Remove shift by 1 expansion.
1886 (arm_lshrdi3_1bit): Remove pattern.
1887 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1888 cost of ashldi3 by 1.
1889 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1890 (<shift>di3_neon): Likewise.
1892 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1894 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1895 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1896 and (*aarch64_simd_mov<VQ:mode>).
1897 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1898 pattern alternative.
1899 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1900 attributes to match pattern alternative.
1902 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1904 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1905 * config/rs6000/emmintrin.h: New file.
1906 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1908 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1910 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1911 * (movdi_vfp_cortexa8): Remove pattern.
1913 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1915 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1917 (Specific, alpha*-dec-osf5.1): Remove.
1918 (Specific, mips-sgi-irix5): Remove.
1919 (Specific, mips-sgi-irix6): Remove.
1921 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1924 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1925 arguments to clear_bit_region_be.
1927 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1929 * gimplify.c: Include memmodel.h.
1931 2017-10-30 Martin Jambor <mjambor@suse.cz>
1933 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1934 condition whether loop should be executed at all.
1936 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1938 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1939 gimple folding of vec_madd() intrinsics.
1940 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1941 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1942 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1944 2017-10-30 Richard Biener <rguenther@suse.de>
1946 PR tree-optimization/82762
1948 2017-10-23 Richard Biener <rguenther@suse.de>
1950 PR tree-optimization/82129
1952 2017-08-01 Richard Biener <rguenther@suse.de>
1954 PR tree-optimization/81181
1955 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1956 (compute_antic): ... end of iteration here.
1958 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1960 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1961 -std=iso9899:2017 and -std=gnu17.
1962 * doc/standards.texi (C Language): Document C17 support.
1963 * doc/cpp.texi (Overview): Mention -std=c17.
1964 (Standard Predefined Macros): Document C11 and C17 values of
1965 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1966 * doc/extend.texi (Inline): Do not list individual options for
1967 standards newer than C99.
1968 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1970 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1973 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1975 * asan.c (asan_finish_file): Align asan globals array by shadow
1978 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1981 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1982 (struct store_immediate_info): Add bitregion_start and bitregion_end
1984 (store_immediate_info::store_immediate_info): Add brs and bre
1985 arguments and initialize bitregion_{start,end} from those.
1986 (struct merged_store_group): Add bitregion_start, bitregion_end,
1987 align_base and mask fields. Drop unnecessary struct keyword from
1988 struct store_immediate_info. Add do_merge method.
1989 (clear_bit_region_be): Use memset instead of loop storing zeros.
1990 (merged_store_group::do_merge): New method.
1991 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
1992 stores as long as the surrounding bitregions have no gaps.
1993 (merged_store_group::merge_overlapping): Use do_merge.
1994 (merged_store_group::apply_stores): Test that bitregion_{start,end}
1995 is byte aligned, rather than requiring that start and width are
1996 byte aligned. Drop unnecessary struct keyword from
1997 struct store_immediate_info. Allocate and populate also mask array.
1998 Make start of the arrays relative to bitregion_start rather than
1999 start and size them according to bitregion_{end,start} difference.
2000 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2001 struct store_immediate_info.
2002 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2003 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2004 struct keyword from struct store_immediate_info.
2005 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2006 between stores as long as the surrounding bitregions have no gaps.
2008 (struct split_store): Add orig non-static data member.
2009 (split_store::split_store): Initialize orig to false.
2010 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2011 if there is exactly a single original stmt. Change stmts argument
2012 to pointer from reference, if NULL, don't push anything to it. Add
2013 first argument, use it to optimize skipping over orig stmts that
2014 are known to be before bitpos already. Simplify.
2015 (split_group): Return unsigned int count how many stores are or
2016 would be needed rather than a bool. Add allow_unaligned argument.
2017 Change split_stores argument from reference to pointer, if NULL,
2018 only do a dry run computing how many stores would be produced.
2019 Rewritten algorithm to use both alignment and misalign if
2020 !allow_unaligned and handle bitfield stores with gaps.
2021 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2022 from bitregion_start instead of start. Compute allow_unaligned
2023 here, if true, do 2 split_group dry runs to compute which one
2024 produces fewer stores and prefer aligned if equal. Punt if
2025 new count is bigger or equal than original before emitting any
2026 statements, rather than during that. Remove no longer needed
2027 new_ssa_names tracking. Replace num_stmts with
2028 split_stores.length (). Use 32-bit stack allocated entries
2029 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2030 if possible. Handle bitfields with gaps.
2031 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2032 Compute bitregion_{start,end} for the stores and construct
2033 store_immediate_info with that. Formatting fixes.
2035 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2038 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2039 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2041 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2043 * gimplify.c: Include tm_p.h.
2045 * common.opt (gcoff): Re-add as ignored option.
2046 (gcoff1, gcoff2, gcoff3): Likewise.
2048 * Makefile.in (OBJS): Delete sdbout.o.
2049 (GTFILES): Delete $(srcdir)/sdbout.c.
2050 * debug.h: Delete sdb_debug_hooks.
2051 * final.c: Delete sdbout.h include.
2052 (final_scan_insn): Delete SDB_DEBUG check.
2053 (rest_of_clean_state): Likewise.
2054 * output.h: Delete sdb_begin_function_line.
2057 * toplev.c: Delete sdbout.h include.
2058 (process_options): Delete SDB_DEBUG check.
2059 * tree-core.h (tree_type_common): Delete pointer field of
2061 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2062 TYPE_SYMTAB_POINTER.
2063 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2064 (TYPE_SYMTAB_IS_POINTER): Delete.
2065 (TYPE_SYMTAB_IS_DIE): Renumber.
2066 * xcoffout.c: Refer to former sdbout.c file.
2067 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2069 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2070 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2071 (Debugging Options): Delete -gcoff.
2072 (-gstabs): Delete SDB reference.
2074 (-gcoff@var{level}): Delete.
2075 * doc/passes.texi (Debugging information output): Delete SDB and
2076 sdbout.c references.
2077 * doc/tm.texi: Regenerate.
2078 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2079 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2081 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2082 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2083 (SDB_DEBUGGING_INFO): Delete.
2084 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2085 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2086 * target.def (output_source_filename): Delete COFF reference.
2088 * common.opt (gcoff): Delete.
2089 (gxcoff+): Update Negative chain.
2090 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2092 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2093 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2094 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2096 (expand_function_start): Change sdb reference to past tense.
2097 (expand_function_end): Change sdb reference to past tense.
2098 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2099 * opts.c (debug_type_names): Delete coff entry.
2100 (common_handle_option): Delete OPT_gcoff case.
2101 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2103 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2104 * config/cris/cris.h: Delete SDB reference in comment.
2105 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2106 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2107 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2108 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2110 (ix86_expand_prologue): Likewise.
2111 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2112 * config/ia64/ia64.h: Likewise.
2113 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2114 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2116 * config/mmix/mmix.h: Likewise.
2117 * config/nds32/nds32.c: Likewise.
2118 * config/stormy/storym16.h: Likewise.
2119 * config/visium/visium.h: Likewise.
2120 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2122 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2124 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2125 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2126 FRAME_POINTER_REGNUM point at high end of local var area.
2128 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2130 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2131 Move comment around. Do not reset best_edge for a copiable
2132 destination if the copy would cause a partition change.
2133 (better_edge_p): Remove redundant check.
2135 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2137 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2139 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2141 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2142 for math functions that have _Float<N> and _Float<N>X variants.
2143 (mathfn_built_in_2): Add support for math functions that have
2144 _Float<N> and _Float<N>X variants.
2145 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2146 (expand_builtin_mathfn_ternary): Add support for fma with
2147 _Float<N> and _Float<N>X variants.
2148 (expand_builtin): Likewise.
2149 (fold_builtin_3): Likewise.
2150 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2151 create math function _Float<N> and _Float<N>X variants as external
2153 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2154 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2155 the __builtin_ prefix and if not strict ansi, without the prefix.
2156 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2157 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2158 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2159 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2160 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2161 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2162 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2163 function signatures for fma _Float<N> and _Float<N>X variants.
2164 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2165 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2166 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2167 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2168 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2169 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2170 * gencfn-macros.c (print_case_cfn): Add support for math functions
2171 that have _Float<N> and _Float<N>X variants.
2172 (print_define_operator_list): Likewise.
2173 (fltfn_suffixes): Likewise.
2175 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
2176 for math functions that have _Float<N> and _Float<N>X variants.
2177 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
2178 and _Float<N>X variants.
2179 (COPYSIGN): Likewise.
2182 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
2183 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
2185 (integer_valued_read_call_p): Likewise.
2186 * fold-const-call.c (fold_const_call_ss): Likewise.
2187 (fold_const_call_sss): Add support for copysign, fmin, and fmax
2188 _Float<N> and _Float<N>X variants.
2189 (fold_const_call_ssss): Add support for fma _Float<N> and
2190 _Float<N>X variants.
2191 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
2192 support for copysign and fma _Float<N> and _Float<N>X variants.
2193 (backprop::process_builtin_call_use): Likewise.
2194 * tree-call-cdce.c (can_test_argument_range); Add support for
2195 sqrt _Float<N> and _Float<N>X variants.
2196 (edom_only_function): Likewise.
2197 (get_no_error_domain): Likewise.
2198 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
2199 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
2200 copysign _Float<N> and _Float<N>X variants.
2201 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
2202 handled by machine independent code.
2203 (FMAF128): Likewise.
2204 * doc/cpp.texi (Common Predefined Macros): Document defining
2205 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
2206 fma _Float<N> and _Float<N>X variants.
2208 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2211 * config/i386/i386-modes.def (CCFPU): Remove definition.
2212 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
2213 (ix86_cc_modes_compatible): Ditto.
2214 (ix86_expand_carry_flag_compare): Ditto.
2215 (ix86_expand_int_movcc): Ditto.
2216 (ix86_expand_int_addcc): Ditto.
2217 (ix86_reverse_condition): Ditto.
2218 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
2219 Return true/false for unordered/ordered fp comparisons.
2220 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
2221 (ix86_prepare_fp_compare_args): Update for rename.
2222 (ix86_expand_fp_compare): Update for rename. Generate unordered
2223 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
2224 (ix86_expand_sse_compare_and_jump): Ditto.
2225 * config/i386/predicates.md (fcmov_comparison_operator):
2226 Remove CCFPU mode handling.
2227 (ix86_comparison_operator): Ditto.
2228 (ix86_carry_flag_operator): Ditto.
2229 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
2230 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
2231 (*cmpu<mode>_cc_i387): Ditto.
2232 (FPCMP): Remove mode iterator.
2233 (unord): Remove mode attribute.
2234 (unord_subst): New define_subst transformation
2235 (unord): New define_subst attribute.
2237 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2238 (*cmpi<unord>xf_i387): Ditto.
2239 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2240 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2241 using unord_subst transformation.
2242 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2243 (round_saeonly): Also handle CCFP mode.
2244 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2245 Remove UNSPEC_SAHF unspec handling.
2247 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2249 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2251 2017-10-27 Jeff Law <law@redhat.com>
2253 * gimple-ssa-sprintf.c: Include domwalk.h.
2254 (class sprintf_dom_walker): New class, derived from dom_walker.
2255 (sprintf_dom_walker::before_dom_children): New function.
2256 (struct call_info): Moved into sprintf_dom_walker class
2257 (compute_formath_length, handle_gimple_call): Likewise.
2258 (sprintf_length::execute): Call the dominator walker rather
2259 than walking the statements.
2261 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2262 gimple statement locations.
2263 (check_array_bounds): Corresponding changes. Get the statement's
2264 location directly from wi->stmt.
2266 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2269 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2271 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2273 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2274 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2276 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2279 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2280 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2281 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2282 maybe_get_pool_constant.
2283 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2286 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2288 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2289 2.26 caveat. Update gas and gld versions.
2290 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2293 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2295 * cgraph.h (set_malloc_flag): Declare.
2296 * cgraph.c (set_malloc_flag_1): New function.
2297 (set_malloc_flag): Likewise.
2298 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2299 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2301 (read_ipa_call_summary): Add support for reading is_return_callee.
2302 (write_ipa_call_summary): Stream is_return_callee.
2303 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2304 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2305 ipa-prop.h, ipa-fnsummary.h.
2306 (pure_const_names): Change to static.
2307 (malloc_state_e): Define.
2308 (malloc_state_names): Define.
2309 (funct_state_d): Add field malloc_state.
2310 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2311 (check_retval_uses): New function.
2312 (malloc_candidate_p): Likewise.
2313 (analyze_function): Add support for malloc attribute.
2314 (pure_const_write_summary): Stream malloc_state.
2315 (pure_const_read_summary): Add support for reading malloc_state.
2316 (dump_malloc_lattice): New function.
2317 (propagate_malloc): New function.
2318 (warn_function_malloc): New function.
2319 (ipa_pure_const::execute): Call propagate_malloc and
2320 ipa_free_fn_summary.
2321 (pass_local_pure_const::execute): Add support for malloc attribute.
2322 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2323 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2325 2017-10-27 Martin Liska <mliska@suse.cz>
2327 PR gcov-profile/82457
2328 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2329 for fork-like functions to be properly instrumented.
2331 2017-10-27 Richard Biener <rguenther@suse.de>
2334 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2335 info when we redirected EH.
2337 2017-10-26 Michael Collison <michael.collison@arm.com>
2339 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2341 (<optab>_trunchf<GPI:mode>2: New pattern.
2342 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2343 * config/aarch64/iterators.md (wv): New mode attribute.
2344 (vf, VF): New mode attributes.
2345 (vgp, VGP): New mode attributes.
2346 (s): Update attribute with SImode and DImode prefixes.
2348 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2350 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2351 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2352 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2353 (nios2_option_overide): Initialize it. Don't allow R0-relative
2354 addressing with PIC.
2355 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2356 (nios2_symbolic_constant_p): Likewise.
2357 (nios2_legitimate_address_p): Likewise.
2358 (nios2_r0rel_section_name_p): New.
2359 (nios2_symbol_ref_in_r0rel_data_p): New.
2360 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2361 (r0rel_constant_p): New.
2362 (nios2_print_operand_address): Handle r0rel_constant_p.
2363 (nios2_cdx_narrow_form_p): Likewise.
2364 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2365 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2366 (Nios II Options): Document -mr0rel-sec.
2368 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2370 * config/nios2/nios2.c: Include xregex.h.
2371 (nios2_gprel_sec_regex): New.
2372 (nios2_option_overide): Initialize it. Don't allow GP-relative
2373 addressing with PIC.
2374 (nios2_small_section_name_p): Check for regex match.
2375 * config/nios2/nios2.opt (mgprel-sec=): New option.
2376 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2377 (Nios II Options): Document -mgprel-sec.
2379 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2381 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2383 2017-10-26 Tom de Vries <tom@codesourcery.com>
2385 PR tree-optimization/82707
2386 * gimple.c (gimple_copy): Fix unsharing of
2387 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2389 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2391 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2392 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2393 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2394 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2395 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2396 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2397 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2398 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2399 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2400 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2401 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2402 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2403 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2404 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2405 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2406 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2407 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2409 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2411 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2413 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2414 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2415 warning to rs6000.c. Remove the Undocumented flag, since it has
2417 (-mabi=ibmlongdouble): Likewise.
2418 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2419 already set, set the default format for long double.
2420 (rs6000_debug_reg_global): Print whether long double is IBM or
2422 (rs6000_option_override_internal): Rework setting long double
2423 format. Only warn if the user is changing the long double default
2424 and they did not use -Wno-psabi.
2425 * doc/invoke.texi (PowerPC options): Update the documentation for
2426 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2428 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2429 Alan Hayward <alan.hayward@arm.com>
2430 David Sherwood <david.sherwood@arm.com>
2432 * rtl.h (wider_subreg_mode): New function.
2433 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2434 rather than an unsigned int *.
2435 * ira-color.c (regno_max_ref_width): Replace with...
2436 (regno_max_ref_mode): ...this new variable.
2437 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2438 Use wider_subreg_mode.
2439 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2440 rather than an unsigned int *.
2441 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2442 (process_alt_operands): Likewise.
2443 (invariant_p): Likewise.
2444 * lra-spills.c (assign_mem_slot): Likewise.
2445 (add_pseudo_to_slot): Likewise.
2446 * lra.c (collect_non_operand_hard_regs): Likewise.
2447 (add_regs_to_insn_regno_info): Likewise.
2448 * reload1.c (regno_max_ref_width): Replace with...
2449 (regno_max_ref_mode): ...this new variable.
2450 (reload): Update accordingly. Update call to
2451 ira_sort_regnos_for_alter_reg.
2452 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2453 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2454 (scan_paradoxical_subregs): Likewise.
2456 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2458 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2459 (aarch64_frame): Add emit_frame_chain boolean.
2460 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2461 Move eh_return case to aarch64_layout_frame.
2462 (aarch64_layout_frame): Initialize emit_frame_chain.
2463 (aarch64_expand_prologue): Use emit_frame_chain.
2465 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2467 * config/aarch64/aarch64.c (aarch64_layout_frame):
2468 Ensure LR is always stored at the bottom of the callee-saves.
2469 Remove rarely used frame layout which saves callee-saves at top of
2470 frame, so the store of LR can be used as a valid probe in all cases.
2472 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2474 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2475 Improve unaligned TImode/TFmode base/offset split.
2477 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2478 Alan Hayward <alan.hayward@arm.com>
2479 David Sherwood <david.sherwood@arm.com>
2481 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2482 * combine.c (find_single_use_1): Likewise.
2483 (expand_field_assignment): Likewise.
2484 (move_deaths): Likewise.
2485 * lra-constraints.c (simplify_operand_subreg): Likewise.
2486 (curr_insn_transform): Likewise.
2487 * lra.c (collect_non_operand_hard_regs): Likewise.
2488 (add_regs_to_insn_regno_info): Likewise.
2489 * rtlanal.c (reg_referenced_p): Likewise.
2490 (covers_regno_no_parallel_p): Likewise.
2492 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2494 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2495 Don't print any bits outside the precision of the value.
2496 * wide-int.cc (test_printing): Add some new tests.
2498 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2500 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2501 supports -xbrace_comment option.
2502 * configure: Regenerate.
2503 * config.in: Regenerate.
2504 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2505 (ASM_CPU_SPEC): Use it.
2507 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2509 * target.def (static_rtx_alignment): New hook.
2510 * targhooks.h (default_static_rtx_alignment): Declare.
2511 * targhooks.c (default_static_rtx_alignment): New function.
2512 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2513 * doc/tm.texi: Regenerate.
2514 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2515 instead of targetm.constant_alignment. Remove call to
2517 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2518 (cris_preferred_mininum_alignment): New function, split out from...
2519 (cris_constant_alignment): ...here.
2520 (cris_static_rtx_alignment): New function.
2521 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2523 (ix86_constant_alignment): ...here.
2524 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2525 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2526 (mmix_static_rtx_alignment): New function.
2527 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2528 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2530 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2533 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2534 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2536 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2538 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2539 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2540 mark as a sign-extending load.
2541 (local_pic_load_u): Define.
2543 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2546 * fold-const.c (operand_equal_for_comparison_p): Also return true
2547 if ARG0 is a simple variant of ARG1 with narrower precision.
2548 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2550 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2552 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2554 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2555 scatter_static, scatter_per_elt.
2556 * x86-tune-costs.h: Add new cost entries.
2558 2017-10-25 Richard Biener <rguenther@suse.de>
2560 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2561 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2562 class pass_fre): Move to ...
2563 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2564 class pass_fre): ... here and adjust for statistics.
2566 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2569 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2570 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2571 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2572 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2575 2017-10-25 Richard Biener <rguenther@suse.de>
2577 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2578 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2579 eliminate_push_avail, eliminate_insert): Move inside...
2580 (class eliminate_dom_walker): ... this class in preparation
2582 (fini_eliminate): Remove by merging with ...
2583 (eliminate): ... this function. Adjust for class changes.
2584 (pass_pre::execute): Remove fini_eliminate call.
2585 (pass_fre::execute): Likewise.
2587 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2590 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2591 (VPERMI2, VPERMI2I): New mode iterators.
2592 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2593 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2595 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2596 mode iterator. Remove 3 old define_insn patterns.
2597 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2598 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2599 VPERMI2 mode iterator, remove the other two expanders.
2600 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2601 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2602 instructions, remove the other two patterns.
2603 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2604 mode iterator, remove the other two patterns.
2605 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2606 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2607 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2608 and adjust argument order accordingly.
2609 (ix86_expand_vec_perm): Adjust caller.
2610 (expand_vec_perm_1): Likewise.
2611 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2612 (expand_vec_perm_vpermt2_vpshub2): ... this.
2613 (ix86_expand_vec_perm_const_1): Adjust caller.
2614 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2617 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2618 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2619 (vec_shl_<mode>): Remove unused expander.
2620 (avx512bw_<shift_insn><mode>3): New define_insn.
2621 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2622 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2624 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2627 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2630 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2632 PR rtl-optimization/82396
2633 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2634 (autopref_multipass_init): Simplify initialization.
2635 (autopref_rank_data): Simplify sort order.
2636 * gcc/sched-int.h (autopref_multipass_data_): Remove
2637 multi_mem_insn_p, min_offset and max_offset.
2639 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2642 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2643 Check special value of flag_omit_frame_pointer.
2644 (aarch64_can_eliminate): Likewise.
2645 (aarch64_override_options_after_change_1): Simplify handling of
2646 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2648 2017-10-24 Richard Biener <rguenther@suse.de>
2650 PR tree-optimization/82697
2651 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2652 zero for conditional load and unconditional store.
2654 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2656 * doc/install.texi: Document bootstrap-cet.
2658 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2661 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2662 ENDBR instruction at function entrance if function is only
2665 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2668 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2669 patterns to better describe from which operation the CF is computed.
2670 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2671 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2672 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
2673 is 0, use _0 suffixed expanders instead of emitting a comparison
2676 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
2678 * config/i386/i386.md(*movsf_internal, *movdf_internal):
2679 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
2681 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
2684 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
2685 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
2686 * loop-iv.c (iv_get_reaching_def): Likewise.
2687 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
2688 variable is promoted and the partition contains undefined values.
2690 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2692 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2694 (nios2_address_cost): Define.
2695 (nios2_legitimize_address): Recognize (exp + constant) directly.
2696 (TARGET_ADDRESS_COST): Define.
2698 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2700 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2701 (nios2_symbolic_memory_operand_p): Declare.
2702 (nios2_split_large_constant): Declare.
2703 (nios2_split_symbolic_memory_operand): Declare.
2704 * config/nios2/nios2.c: Adjust includes.
2705 (nios2_symbolic_constant_allowed): New.
2706 (nios2_symbolic_constant_p): New.
2707 (nios2_plus_symbolic_constant_p): New.
2708 (nios2_valid_addr_expr_p): Recognize addresses involving
2710 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2711 (nios2_symbolic_memory_operand_p): New.
2712 (nios2_large_constant_p): New.
2713 (nios2_split_large_constant): New.
2714 (nios2_split_plus_large_constant): New.
2715 (nios2_split_symbolic_memory_operand): New.
2716 (nios2_legitimize_address): Code refactoring. Handle addresses
2717 involving symbolic constants.
2718 (nios2_emit_move_sequence): Likewise.
2719 (nios2_print_operand): Improve error output.
2720 (nios2_print_operand_address): Handle LO_SUM.
2721 (nios2_cdx_narrow_form_p): Likewise.
2722 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2723 operands involving symbolic constants.
2724 (movhi_internal, movsi_internal): Likewise.
2725 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2726 (extendhisi2, extendqi<mode>2): Likewise.
2728 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2730 * tree-pass.h (PROP_rtl_split_insns): Define.
2731 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2733 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2735 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2737 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2740 * target.def (const_not_ok_for_debug_p): Default to
2741 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2742 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2743 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2744 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2745 which targetm.const_not_ok_for_debug_p returned true.
2746 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2748 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2750 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2751 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2752 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2754 (ix86_const_not_ok_for_debug_p): New function.
2755 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2756 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2758 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2761 * system.h: Conditionally include "unique-ptr.h" if
2762 INCLUDE_UNIQUE_PTR is defined.
2763 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2764 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2766 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2768 * config/rl78/rl78.md: New define_expand "subdi3".
2770 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2773 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2774 DF_REF_INSN if DF_REF_INSN_INFO is false.
2776 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2778 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2779 xmm_move instead of sse_move.
2780 (sse_store_index): New function.
2781 (ix86_register_move_cost): Be more sensible about mismatch stall;
2782 model AVX moves correctly; make difference between sse->integer and
2784 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2785 moves; make difference between SSE and AVX.
2786 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2787 and zmm_move. Increase size of sse load and store tables;
2788 add unaligned load and store tables; add ssemmx_to_integer.
2789 * x86-tune-costs.h: Update all entries according to real
2790 move latencies from Agner Fog's manual and chip documentation.
2792 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2795 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2796 * config/i386/constraints.md (Wf): New constraint.
2797 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2798 (cmp<dwi>_doubleword): Removed.
2799 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2800 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2801 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2802 expand with cmp<dwi>_doubleword. For LTU and GEU use
2803 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2805 * common.opt (gcolumn-info): Enable by default.
2806 * doc/invoke.texi (gcolumn-info): Document new default.
2808 2017-10-23 Richard Biener <rguenther@suse.de>
2810 PR tree-optimization/82672
2811 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2812 Fold the stmt if we propagated into it.
2814 2017-10-23 Richard Biener <rguenther@suse.de>
2816 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2817 (bitmap_remove_expr_from_set): ... this. All callers call this
2818 for non-constant values.
2819 (bitmap_set_subtract): Rename to...
2820 (bitmap_set_subtract_expressions): ... this. Adjust and
2822 (bitmap_set_contains_value): Remove superfluous check.
2823 (bitmap_set_replace_value): Inline into single caller ...
2824 (bitmap_value_replace_in_set): ... here and simplify.
2825 (dependent_clean): Merge into ...
2826 (clean): ... this using an overload. Adjust.
2827 (prune_clobbered_mems): Adjust.
2828 (compute_antic_aux): Likewise.
2829 (compute_partial_antic_aux): Likewise.
2831 2017-10-23 Richard Biener <rguenther@suse.de>
2833 PR tree-optimization/82129
2835 2017-08-01 Richard Biener <rguenther@suse.de>
2837 PR tree-optimization/81181
2838 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2839 (compute_antic): ... end of iteration here.
2841 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2843 * target.def (starting_frame_offset): New hook.
2844 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2845 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2846 * doc/tm.texi.in: Regenerate.
2847 * hooks.h (hook_hwi_void_0): Declare.
2848 * hooks.c (hook_hwi_void_0): New function.
2849 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2850 STARTING_FRAME_OFFSET.
2851 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2852 * reload1.c (reload): Likewise.
2853 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2854 instead of STARTING_FRAME_OFFSET.
2855 * function.c (try_fit_stack_local): Likewise.
2856 (assign_stack_local_1): Likewise
2857 (instantiate_virtual_regs): Likewise.
2858 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2859 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2860 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2861 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2862 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2863 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2864 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2865 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2866 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2867 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2868 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2869 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2870 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2871 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2872 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2873 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2874 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2875 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2876 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2877 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2878 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2879 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2880 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2881 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2882 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2883 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2884 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2885 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2886 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2887 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2888 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2889 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2890 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2891 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2892 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2893 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2894 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2895 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2896 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2897 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2898 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2899 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2900 return a HOST_WIDE_INT.
2901 (avr_builtin_setjmp_frame_value): Use it instead of
2902 STARTING_FRAME_OFFSET.
2903 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2904 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2905 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2907 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2908 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2909 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2910 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2911 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2912 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2913 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2914 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2915 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2916 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2917 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2918 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2920 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2921 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2922 * config/mips/mips.c (mips_compute_frame_info): Refer to
2923 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2924 (mips_starting_frame_offset): New function.
2925 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2926 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2927 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2928 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2929 and return a HOST_WIDE_INT.
2930 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2931 (mmix_initial_elimination_offset): Refer to
2932 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2933 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2934 * config/pa/pa.c (pa_starting_frame_offset): New function.
2935 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2936 (pa_expand_prologue): Likewise.
2937 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2938 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2939 !FRAME_GROWS_DOWNWARD handling to...
2940 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2941 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2942 !FRAME_GROWS_DOWNWARD handling to...
2943 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2944 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2945 !FRAME_GROWS_DOWNWARD handling to...
2946 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2947 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2949 (rs6000_starting_frame_offset): New function.
2950 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2951 !FRAME_GROWS_DOWNWARD handling to...
2952 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2953 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2954 !FRAME_GROWS_DOWNWARD handling to...
2955 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2956 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2957 !FRAME_GROWS_DOWNWARD handling to...
2958 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2959 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2960 (rs6000_starting_frame_offset): New function.
2961 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2962 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2963 * config/vax/vax.c (vax_starting_frame_offset): New function.
2964 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2965 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2966 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2967 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2968 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2969 * system.h (STARTING_FRAME_OFFSET): Poison.
2971 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2973 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2974 SCALAR_TYPE_MODE instead of TYPE_MODE.
2976 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2977 Alan Hayward <alan.hayward@arm.com>
2978 David Sherwood <david.sherwood@arm.com>
2980 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2982 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2983 Alan Hayward <alan.hayward@arm.com>
2984 David Sherwood <david.sherwood@arm.com>
2986 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
2988 2017-10-23 Richard Biener <rguenther@suse.de>
2990 PR tree-optimization/82129
2991 * tree-ssa-pre.c (bitmap_set_and): Remove.
2992 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
2993 canonicalizing expressions in the set to those with lowest
2994 ID rather than taking that from the first edge.
2996 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2998 * combine.c (rtx_equal_for_field_assignment_p): Use
2999 byte_lowpart_offset.
3001 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3002 Alan Hayward <alan.hayward@arm.com>
3003 David Sherwood <david.sherwood@arm.com>
3005 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3006 to a SUBREG_PROMOTED_VAR.
3008 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3009 Alan Hayward <alan.hayward@arm.com>
3010 David Sherwood <david.sherwood@arm.com>
3012 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3013 (expand_debug_source_expr): Likewise.
3014 * combine.c (combine_simplify_rtx): Likewise.
3015 * cse.c (fold_rtx): Likewise.
3016 * optabs.c (expand_float): Likewise.
3017 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3018 (simplify_binary_operation_1): Likewise.
3020 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3021 Alan Hayward <alan.hayward@arm.com>
3022 David Sherwood <david.sherwood@arm.com>
3024 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3025 (record_promoted_value): Likewise.
3026 * expr.c (expand_expr_real_2): Likewise.
3027 * ree.c (update_reg_equal_equiv_notes): Likewise.
3028 (combine_set_extension): Likewise.
3029 * rtlanal.c (low_bitmask_len): Likewise.
3030 * simplify-rtx.c (neg_const_int): Likewise.
3031 (simplify_binary_operation_1): Likewise.
3033 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3034 Alan Hayward <alan.hayward@arm.com>
3035 David Sherwood <david.sherwood@arm.com>
3037 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3038 * regcprop.c (maybe_mode_change): Likewise.
3039 * reload1.c (alter_reg): Likewise.
3041 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3043 * inchash.h (inchash::hash::add_wide_int): New function.
3044 * lto-streamer-out.c (hash_tree): Use it.
3046 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3048 * inchash.h (inchash::hash::add_wide_int): Rename to...
3049 (inchash::hash::add_hwi): ...this.
3050 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3051 (polymorphic_call_target_hasher::hash): Likewise.
3052 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3053 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3054 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3055 * lto-streamer-out.c (hash_tree): Likewise.
3056 * optc-save-gen.awk: Likewise.
3057 * tree.c (add_expr): Likewise.
3059 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3062 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3063 for ordered inequality comparisons even with TARGET_IEEE_FP.
3065 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3068 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3069 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3070 Expand with cmp<dwi>_doubleword.
3072 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3074 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3075 List CET intrinsics.
3076 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3077 specific to -fcf-protection option.
3079 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3081 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3082 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3083 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3084 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3085 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3086 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3087 (extra_objs): Add cet.o for Linux/x86 targets.
3088 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3089 * config/i386/cet.c: New file.
3090 * config/i386/cetintrin.h: Likewise.
3091 * config/i386/t-cet: Likewise.
3092 * config/i386/cpuid.h (bit_SHSTK): New.
3093 (bit_IBT): Likewise.
3094 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3095 pass IBT and SHSTK bits.
3096 * config/i386/i386-builtin-types.def
3097 (VOID_FTYPE_UNSIGNED_PVOID): New.
3098 (VOID_FTYPE_UINT64_PVOID): Likewise.
3099 * config/i386/i386-builtin.def: Add CET intrinsics.
3100 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3101 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3102 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3103 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3105 * config/i386/i386.c (rest_of_insert_endbranch): New.
3106 (pass_data_insert_endbranch): Likewise.
3107 (pass_insert_endbranch): Likewise.
3108 (make_pass_insert_endbranch): Likewise.
3109 (ix86_notrack_prefixed_insn_p): Likewise.
3110 (ix86_target_string): Add -mibt, -mshstk flags.
3111 (ix86_option_override_internal): Add flag_cf_protection
3113 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3114 (ix86_print_operand): Add 'notrack' prefix output.
3115 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3116 (ix86_expand_builtin): Expand CET intrinsics.
3117 (x86_output_mi_thunk): Add 'endbranch' instruction.
3118 * config/i386/i386.h (TARGET_IBT): New.
3119 (TARGET_IBT_P): Likewise.
3120 (TARGET_SHSTK): Likewise.
3121 (TARGET_SHSTK_P): Likewise.
3122 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3123 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3124 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3125 (builtin_setjmp_setup): New pattern.
3126 (builtin_longjmp): Likewise.
3127 (rdssp<mode>): Likewise.
3128 (incssp<mode>): Likewise.
3129 (saveprevssp): Likewise.
3130 (rstorssp): Likewise.
3131 (wrss<mode>): Likewise.
3132 (wruss<mode>): Likewise.
3133 (setssbsy): Likewise.
3134 (clrssbsy): Likewise.
3135 (nop_endbr): Likewise.
3136 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3138 * config/i386/immintrin.h: Include <cetintrin.h>.
3139 * config/i386/linux-common.h
3140 (file_end_indicate_exec_stack_and_cet): New prototype.
3141 (TARGET_ASM_FILE_END): New.
3143 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3145 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3146 latencies instead of having separate table; make difference between
3147 integer and float costs.
3148 * i386.h (processor_costs): Remove scalar_stmt_cost,
3149 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3150 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3152 * x86-tune-costs.h: Remove entries which has been removed in
3153 procesor_costs from all tables; make cond_taken_branch_cost
3154 and cond_not_taken_branch_cost COST_N_INSNS based.
3156 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3158 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3160 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3162 * config/i386/i386.md (isa): Remove fma_avx512f.
3163 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3164 <avx512>_fmadd_<mode>_mask3<round_name>,
3165 <avx512>_fmsub_<mode>_mask<round_name>,
3166 <avx512>_fmsub_<mode>_mask3<round_name>,
3167 <avx512>_fnmadd_<mode>_mask<round_name>,
3168 <avx512>_fnmadd_<mode>_mask3<round_name>,
3169 <avx512>_fnmsub_<mode>_mask<round_name>,
3170 <avx512>_fnmsub_<mode>_mask3<round_name>,
3171 <avx512>_fmaddsub_<mode>_mask<round_name>,
3172 <avx512>_fmaddsub_<mode>_mask3<round_name>,
3173 <avx512>_fmsubadd_<mode>_mask<round_name>,
3174 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
3175 (*vec_widen_umult_even_v16si<mask_name>,
3176 *vec_widen_smult_even_v16si<mask_name>): Likewise.
3177 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
3179 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3181 * extend.texi: Add 'nocf_check' documentation.
3182 * gimple.texi: Add second parameter to
3183 gimple_build_call_from_tree.
3184 * invoke.texi: Add -fcf-protection documentation.
3185 * rtl.texi: Add REG_CALL_NOTRACK documenation.
3187 2017-10-20 Richard Biener <rguenther@suse.de>
3189 PR tree-optimization/82473
3190 * tree-vect-loop.c (vectorizable_reduction): Properly get at
3191 the largest input type.
3193 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3195 * c-attribs.c (handle_nocf_check_attribute): New function.
3196 (c_common_attribute_table): Add 'nocf_check' handling.
3197 * gimple-parser.c: Add second argument NULL to
3198 gimple_build_call_from_tree.
3199 * attrib.c (comp_type_attributes): Check nocf_check attribute.
3200 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
3202 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
3203 * common.opt: Add fcf-protection flag.
3204 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
3205 * flag-types.h: Add enum cf_protection_level.
3206 * gimple.c (gimple_build_call_from_tree): Add second parameter.
3207 Add 'nocf_check' attribute propagation to gimple call.
3208 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
3209 (gimple_build_call_from_tree): Update prototype.
3210 (gimple_call_nocf_check_p): New function.
3211 (gimple_call_set_nocf_check): Likewise.
3212 * gimplify.c: Add second argument to gimple_build_call_from_tree.
3213 * ipa-icf.c: Add nocf_check attribute in statement hash.
3214 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
3215 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
3216 * toplev.c (process_options): Add flag_cf_protection handling.
3218 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3220 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
3222 2017-10-20 Richard Biener <rguenther@suse.de>
3224 PR tree-optimization/82603
3225 * tree-if-conv.c (predicate_mem_writes): Make sure to only
3226 remove false predicated stores.
3228 2017-10-20 Richard Biener <rguenther@suse.de>
3230 * graphite-isl-ast-to-gimple.c
3231 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
3232 Remove return value and simplify, dump copied stmt after lhs
3234 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
3235 Reduce dump verbosity.
3236 (gsi_insert_earliest): Likewise.
3237 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3238 * graphite.c (print_global_statistics): Adjust dumping.
3239 (print_graphite_scop_statistics): Likewise.
3240 (print_graphite_statistics): Do not dump loops here.
3241 (graphite_transform_loops): But here.
3243 2017-10-20 Nicolas Roche <roche@adacore.com>
3245 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3246 * configure: Regenerate.
3248 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3251 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3252 functions when optimizing replace GIMPLE_RETURN stmts with
3253 calls to __builtin_unreachable ().
3256 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3257 for -fsanitize=thread link of executables.
3258 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3259 link of executables.
3262 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3264 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3265 define_insns for logical vector shifts to use VI248_AVX512BW
3266 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3267 condition, useless isa and prefix attributes. Change the first
3268 2 of these define_insns to ...
3269 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3270 define_insn for avx512vl.
3271 (<shift_insn><mode>3): ... and this, new define_insn without
3272 masking for non-avx512vl.
3275 * config/i386/sse.md (*andnot<mode>3,
3276 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3277 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3278 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3279 not applied use empty suffix even for TARGET_AVX512VL.
3280 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3281 is applied, supply evex,evex or evex,evex,evex instead of just
3284 2017-10-20 Julia Koval <julia.koval@intel.com>
3286 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3287 (OPTION_MASK_ISA_GFNI_UNSET): New.
3288 (ix86_handle_option): Handle OPT_mgfni.
3289 * config/i386/cpuid.h (bit_GFNI): New.
3290 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3291 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3292 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3293 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3294 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3295 * config/i386/i386.opt: Add mgfni.
3297 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3299 * config/msp430/msp430.c (msp430_option_override): Disable
3300 -fdelete-null-pointer-checks.
3301 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3303 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3305 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3306 of x87 and SSE instructions.
3308 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3310 * asan.c (create_cond_insert_point): Do not update edge count.
3311 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3312 (afdo_propagate_circuit): Likewise.
3313 (afdo_calculate_branch_prob): Likewise.
3314 (afdo_annotate_cfg): Likewise.
3315 * basic-block.h (struct edge_def): Remove count.
3316 (edge_def::count): New accessor.
3317 * bb-reorder.c (rotate_loop): Update.
3318 (find_traces_1_round): Update.
3319 (connect_traces): Update.
3320 (sanitize_hot_paths): Update.
3321 * cfg.c (unchecked_make_edge): Update.
3322 (make_single_succ_edge): Update.
3323 (check_bb_profile): Update.
3324 (dump_edge_info): Update.
3325 (update_bb_profile_for_threading): Update.
3326 (scale_bbs_frequencies_int): Update.
3327 (scale_bbs_frequencies_gcov_type): Update.
3328 (scale_bbs_frequencies_profile_count): Update.
3329 (scale_bbs_frequencies): Update.
3330 * cfganal.c (connect_infinite_loops_to_exit): Update.
3331 * cfgbuild.c (compute_outgoing_frequencies): Update.
3332 (find_many_sub_basic_blocks): Update.
3333 * cfgcleanup.c (try_forward_edges): Update.
3334 (try_crossjump_to_edge): Update
3335 * cfgexpand.c (expand_gimple_cond): Update
3336 (expand_gimple_tailcall): Update
3337 (construct_exit_block): Update
3338 * cfghooks.c (verify_flow_info): Update
3339 (redirect_edge_succ_nodup): Update
3340 (split_edge): Update
3341 (make_forwarder_block): Update
3342 (duplicate_block): Update
3343 (account_profile_record): Update
3344 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3345 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3346 * cfgloopmanip.c (scale_loop_profile): Update.
3348 (lv_adjust_loop_entry_edge): Update.
3349 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3350 (force_nonfallthru_and_redirect): Update.
3351 (purge_dead_edges): Update.
3352 (rtl_flow_call_edges_add): Update.
3353 * cgraphunit.c (init_lowered_empty_function): Update.
3354 (cgraph_node::expand_thunk): Update.
3355 * gimple-pretty-print.c (dump_probability): Update.
3356 (dump_edge_probability): Update.
3357 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3358 * haifa-sched.c (sched_create_recovery_edges): Update.
3359 * hsa-gen.c (convert_switch_statements): Update.
3360 * ifcvt.c (dead_or_predicable): Update.
3361 * ipa-inline-transform.c (inline_transform): Update.
3362 * ipa-split.c (split_function): Update.
3363 * ipa-utils.c (ipa_merge_profiles): Update.
3364 * loop-doloop.c (add_test): Update.
3365 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3366 * lto-streamer-in.c (input_cfg): Update.
3367 (input_function): Update.
3368 * lto-streamer-out.c (output_cfg): Update.
3369 * modulo-sched.c (sms_schedule): Update.
3370 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3371 * predict.c (maybe_hot_edge_p): Update.
3372 (unlikely_executed_edge_p): Update.
3373 (probably_never_executed_edge_p): Update.
3374 (dump_prediction): Update.
3375 (drop_profile): Update.
3376 (propagate_unlikely_bbs_forward): Update.
3377 (determine_unlikely_bbs): Update.
3378 (force_edge_cold): Update.
3379 * profile.c (compute_branch_probabilities): Update.
3380 * reg-stack.c (better_edge): Update.
3381 * shrink-wrap.c (handle_simple_exit): Update.
3382 * tracer.c (better_p): Update.
3383 * trans-mem.c (expand_transaction): Update.
3384 (split_bb_make_tm_edge): Update.
3385 * tree-call-cdce.c: Update.
3386 * tree-cfg.c (gimple_find_sub_bbs): Update.
3387 (gimple_split_edge): Update.
3388 (gimple_duplicate_sese_region): Update.
3389 (gimple_duplicate_sese_tail): Update.
3390 (gimple_flow_call_edges_add): Update.
3391 (insert_cond_bb): Update.
3392 (execute_fixup_cfg): Update.
3393 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3394 * tree-complex.c (expand_complex_div_wide): Update.
3395 * tree-eh.c (lower_resx): Update.
3396 (unsplit_eh): Update.
3397 (cleanup_empty_eh_move_lp): Update.
3398 * tree-inline.c (copy_edges_for_bb): Update.
3399 (freqs_to_counts): Update.
3400 (copy_cfg_body): Update.
3401 * tree-ssa-dce.c (remove_dead_stmt): Update.
3402 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3403 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3404 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3405 (unloop_loops): Update.
3406 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3407 * tree-ssa-loop-split.c (connect_loops): Update.
3408 (split_loop): Update.
3409 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3410 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3411 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3412 * tree-ssa-reassoc.c (branch_fixup): Update.
3413 * tree-ssa-tail-merge.c (replace_block_by): Update.
3414 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3415 (compute_path_counts): Update.
3416 (update_profile): Update.
3417 (recompute_probabilities): Update.
3418 (update_joiner_offpath_counts): Update.
3419 (estimated_freqs_path): Update.
3420 (freqs_to_counts_path): Update.
3421 (clear_counts_path): Update.
3422 (ssa_fix_duplicate_block_edges): Update.
3423 (duplicate_thread_path): Update.
3424 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3425 (case_bit_test_cmp): Update.
3426 (collect_switch_conv_info): Update.
3427 (gen_inbound_check): Update.
3428 (do_jump_if_equal): Update.
3429 (emit_cmp_and_jump_insns): Update.
3430 * tree-tailcall.c (decrease_profile): Update.
3431 (eliminate_tail_call): Update.
3432 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3433 (vect_do_peeling): Update.
3434 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3435 * ubsan.c (ubsan_expand_null_ifn): Update.
3436 (ubsan_expand_ptr_ifn): Update.
3437 * value-prof.c (gimple_divmod_fixed_value): Update.
3438 (gimple_mod_pow2): Update.
3439 (gimple_mod_subtract): Update.
3440 (gimple_ic): Update.
3441 (gimple_stringop_fixed_value): Update.
3443 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3446 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3448 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3450 PR rtl-optimization/82395
3451 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3452 based on non_spilled_static_chain_regno_p.
3454 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3456 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3457 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3458 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3460 2017-10-19 Martin Sebor <msebor@redhat.com>
3462 PR tree-optimization/82596
3463 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3465 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3467 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3468 (get_mem_refs_of_builtin_call): Likewise.
3469 * builtins.c (expand_builtin_apply): Adjust call to
3470 allocate_dynamic_stack_space.
3471 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3472 the third argument to allocate_dynamic_stack_space, otherwise -1.
3473 (expand_builtin): Deal with all alloca variants.
3474 (is_inexpensive_builtin): Likewise.
3475 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3476 * calls.c (special_function_p): Deal with all alloca variants.
3477 (initialize_argument_information): Adjust call to
3478 allocate_dynamic_stack_space.
3479 (expand_call): Likewise.
3480 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3481 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3482 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3483 use it for the stack usage computation.
3484 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3485 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3486 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3487 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3488 (in_loop_p): Remove first argument and useless check.
3489 (pass_walloca::execute): Remove useless test and adjust call to above.
3490 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3491 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3492 (gimplify_call_expr): Deal with all alloca variants.
3493 * hsa-gen.c (gen_hsa_alloca): Likewise.
3494 (gen_hsa_insns_for_call): Likewise.
3495 * ipa-pure-const.c (special_builtin_state): Likewise.
3496 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3497 * tree-object-size.c (alloc_object_size): Likewise.
3498 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3499 (call_may_clobber_ref_p_1): Likewise.
3500 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3501 (ccp_fold_stmt): Likewise.
3502 (optimize_stack_restore): Likewise.
3503 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3504 (mark_all_reaching_defs_necessary_1): Likewise.
3505 (propagate_necessity): Likewise.
3506 (eliminate_unnecessary_stmts): Likewise.
3507 * tree.c (build_common_builtin_nodes): Build
3508 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3509 (build_alloca_call_expr): New function.
3510 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3511 (CASE_BUILT_IN_ALLOCA): Likewise.
3512 (build_alloca_call_expr): Declare.
3513 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3515 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3518 * dwarf2out.c (new_die_raw): New static inline function.
3519 (new_die): Use it to create the DIE.
3520 (add_AT_external_die_ref): Likewise.
3521 (clone_die): Likewise.
3522 (clone_as_declaration): Likewise.
3523 (dwarf2out_vms_debug_main_pointer): Likewise.
3524 (base_type_die): Likewise. Remove early return for corner cases.
3525 Do not call add_pubtype on the DIE here.
3526 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3527 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3528 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3529 native order exists for base types, attach the DIE manually and call
3530 add_pubtype on it. Do not equate a reverse order DIE to the type.
3532 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3534 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3535 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3537 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3539 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3541 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3542 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3543 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3544 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3545 * opts.c (sanitizer_opts): Add builtin.
3546 * ubsan.c (instrument_builtin): New function.
3547 (pass_ubsan::execute): Call it.
3548 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3549 * doc/invoke.texi: Document -fsanitize=builtin.
3551 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3552 builtins, store max (log2 (align), 0) into uchar field instead of
3553 align into uptr field.
3554 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3555 store uchar 0 field instead of uptr 0 field.
3556 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3557 instead of passing one address of struct with 2 locations pass
3558 two addresses of structs with 1 location each.
3559 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3560 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3561 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3562 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3563 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3564 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3565 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3566 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3568 2017-10-19 Martin Liska <mliska@suse.cz>
3571 * file-find.c (remove_prefix): Remove.
3572 * file-find.h (remove_prefix): Likewise.
3573 * gcc-ar.c: Remove smartness of lookup.
3575 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3577 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3578 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3579 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3581 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3584 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3585 (setcc + and to xor + setcc): New peephole2.
3587 2017-10-19 Tom de Vries <tom@codesourcery.com>
3589 * doc/sourcebuild.texi (Test Directives, Variants of
3590 dg-require-support): Add dg-require-stack-size.
3592 2017-10-19 Martin Liska <mliska@suse.cz>
3595 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3596 that have a large alignment.
3597 (gimplify_target_expr): Likewise.
3599 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3601 PR rtl-optimization/82602
3602 * ira.c (rtx_moveable_p): Return false for volatile asm.
3604 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3607 * config/i386/i386-modes.def (CCGZ): New CC mode.
3608 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3609 * config/i386/predicates.md (ix86_comparison_operator):
3611 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3612 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3613 with double-word subtraction.
3614 (put_condition_code): Handle CCGZmode.
3616 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3618 * wide-int.cc (debug (const wide_int &)): New.
3619 (debug (const wide_int *)): New.
3620 (debug (const widest_int &)): New.
3621 (debug (const widest_int *)): New.
3623 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3626 * lra-constraints.c (curr_insn_transform): Use non-input operand
3627 instead of output one for matched reload.
3629 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3631 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3632 (tree-ssa-loop-ivopts.h): New header file.
3633 (struct builtin_info): New fields.
3634 (classify_builtin_1): Compute and record base and offset parts for
3635 memset builtin partition by calling strip_offset.
3636 (offset_cmp, fuse_memset_builtins): New functions.
3637 (finalize_partitions): Fuse adjacent memset partitions by calling
3639 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3640 Expose the interface.
3641 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3643 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3645 PR tree-optimization/82574
3646 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3647 that data reference must be executed exactly once per iteration
3648 against the outermost loop in nest.
3649 (classify_partition): Update call to above function.
3651 2017-10-18 Richard Biener <rguenther@suse.de>
3653 PR tree-optimization/82591
3654 * graphite.c (graphite_transform_loops): Move code gen message
3656 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3657 Here. Handle scop_to_isl_ast failing.
3658 (scop_to_isl_ast): Limit the number of ISL operations.
3660 2017-10-18 Richard Biener <rguenther@suse.de>
3662 * graphite-isl-ast-to-gimple.c
3663 (translate_isl_ast_to_gimple::set_rename): Simplify.
3664 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3665 (graphite_copy_stmts_from_block): ... here.
3666 (copy_bb_and_scalar_dependences): Simplify.
3667 (add_parameters_to_ivs_params): Canonicalize.
3668 (generate_entry_out_of_ssa_copies): Simplify.
3669 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3670 by passing in ISL dimension.
3671 (parameter_index_in_region_1): Rename to ...
3672 (parameter_index_in_region): ... this.
3673 (extract_affine): Adjust assert, pass down parameter index.
3674 (add_param_constraints): Use range-info when available.
3675 (build_scop_context): Adjust.
3676 * sese.c (new_sese_info): Adjust.
3677 (free_sese_info): Likewise.
3678 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
3679 Remove unused typedefs.
3680 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
3682 2017-10-18 Martin Liska <mliska@suse.cz>
3684 * combine.c (simplify_compare_const): Add gcc_fallthrough.
3686 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3688 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3689 (s390_sched_init): Do not reset s390_sched_state if we entered the
3690 current basic block via a fallthru edge and all others are unlikely.
3692 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3694 * config/s390/s390.c (NUM_SIDES): New variable.
3695 (LONGRUNNING_THRESHOLD): New variable.
3696 (LATENCY_FACTOR): New variable.
3697 (s390_sched_score): Decrease score for long-running instructions on
3699 (s390_sched_variable_issue): Perform bookkeeping for long-running
3702 2017-10-18 Richard Biener <rguenther@suse.de>
3704 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3705 Simplify with removal of the parameter rename map.
3706 (set_rename): Likewise.
3707 (should_copy_to_new_region): Likewise.
3708 (graphite_copy_stmts_from_block): Likewise.
3709 (copy_bb_and_scalar_dependences): Remove initialization of
3710 unused copied_bb_map.
3712 (copy_internal_parameters): Likewise.
3713 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3714 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3715 Use INTEGRAL_TYPE_P.
3716 (parameter_index_in_region_1): Rename to ...
3717 (assign_parameter_index_in_region): ... this. Assert we have
3718 a parameter we handle.
3719 (scan_tree_for_params): Adjust.
3720 * sese.h (parameter_rename_map_t): Remove.
3721 (struct sese_info_t): Remove unused parameter_rename_map and
3722 copied_bb_map members.
3723 * sese.c (new_sese_info): Adjust.
3724 (free_sese_info): Likewise.
3726 2017-10-18 Martin Liska <mliska@suse.cz>
3729 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3730 on an abnormal edge.
3732 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3734 * doc/invoke.texi (ffunction-sections and fdata-sections):
3737 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3739 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3740 the use statement can throw internally.
3742 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3744 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3745 any RTX present on the RHS of a SET.
3746 * compare-elim.c (try_eliminate_compare): Restore comment.
3748 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3750 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3751 may be also called on tcc_type nodes.
3752 * langhooks.c (lhd_tree_size): Likewise.
3754 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3756 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3757 format_warning_at_substring.
3758 (maybe_warn): Convert source_range * param to a location_t. Pass
3759 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3760 (format_directive): Remove code to extract source_ranges and
3761 source_range * in favor of just a location_t.
3762 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3764 * substring-locations.c (format_warning_va): Convert
3765 source_range * param to a location_t.
3766 (format_warning_at_substring): Likewise.
3767 * substring-locations.h (format_warning_va): Likewise.
3768 (format_warning_at_substring): Likewise.
3770 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3772 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3774 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3775 and scatter/gather ops.
3777 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3778 vec_gather_load and vec_scatter_store.
3779 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3780 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3781 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3782 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3783 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3784 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3786 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3788 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3789 Detect FTST instruction and handle its register pops. Only pop
3790 second operand if can_pop_second_op is true.
3791 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3792 set can_pop_second_op to false in the compare_for_stack_reg call.
3794 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3795 output_fp_compare for stack register operands.
3796 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3797 instructions here. Do not emit stack register pops here. Assert
3798 that FCOMPP pops next to top stack register. Rewrite function.
3800 2017-10-17 Nathan Sidwell <nathan@acm.org>
3803 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3804 use DECL_ASSEMBLER_NAME_RAW.
3807 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3810 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3811 Wilco Dijkstra <wilco.dijkstra@arm.com>
3813 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3814 converstion to Pmode from the buf_addr.
3816 2017-10-17 Richard Biener <rguenther@suse.de>
3818 * graphite-dependences.c (scop_get_reads_and_writes): Change
3819 output parameters to references.
3821 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3823 PR 71026/tree-optimization
3824 * fold-const.c (distribute_real_division): Removed.
3825 (fold_binary_loc): Remove calls to distribute_real_divison.
3827 2017-10-17 Richard Biener <rguenther@suse.de>
3829 * graphite-scop-detection.c
3830 (scop_detection::stmt_has_simple_data_refs_p): Always use
3831 the full nest as region.
3832 (try_generate_gimple_bb): Likewise.
3833 * sese.c (scalar_evolution_in_region): Simplify now that
3834 SCEV can handle instantiation in regions.
3835 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3836 in the non-loop part of a function if requested.
3838 2017-10-17 Richard Biener <rguenther@suse.de>
3840 PR tree-optimization/82563
3841 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3843 (graphite_regenerate_ast_isl): Call it.
3844 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3846 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3848 PR tree-optimization/82549
3849 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3850 Formatting fixes. Instead of calling make_bit_field_ref with negative
3853 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3855 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3856 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3857 _mm_maskz_reduce_ss): New.
3858 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3859 __builtin_ia32_reducess_mask): Ditto..
3860 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3861 * config/i386/sse.md (reduces<mode>): Renamed to ...
3862 (reduces<mode><mask_scalar_name>): ... this.
3863 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3865 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3866 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3868 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3870 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3871 * selftest-run-tests.c (selftest::run_tests): Call
3872 selftest::unique_ptr_tests_cc_tests.
3873 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3874 * unique-ptr-tests.cc: New file.
3876 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3879 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3881 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3882 (make_hard_regno_born, make_hard_regno_dead): Update
3883 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3885 2017-10-16 Jeff Law <law@redhat.com>
3887 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3889 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3891 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3893 2017-10-16 Olivier Hainque <hainque@adacore.com>
3895 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3896 with_cpu if we were configured for an e500v2 target cpu name.
3898 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3900 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3901 * doc/invoke.texi: Document +nodsp as a valid extension for
3904 2017-10-16 Martin Liska <mliska@suse.cz>
3906 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3907 (test_set_range): Likewise.
3908 (test_range_functions): Rename to ...
3909 (test_bit_in_range): ... this.
3910 (sbitmap_c_tests): Add new test.
3912 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3914 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3916 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3917 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3919 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3921 * config/aarch64/aarch64-builtins.c
3922 (aarch64_types_quadopu_lane_qualifiers): New.
3923 (TYPES_QUADOPU_LANE): New.
3924 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3925 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3926 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3927 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3928 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3929 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3930 (Vdottype, DOTPROD): New.
3931 (sur): Add SDOT and UDOT.
3933 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3935 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3936 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3937 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3939 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3940 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3941 Enable TARGET_DOTPROD.
3942 (cortex-a75.cortex-a55): Likewise.
3943 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3945 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3947 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3948 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3950 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3952 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3953 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3954 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3955 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3956 * config/arm/types.md (neon_dot, neon_dot_q): New.
3957 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3959 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3961 * config/arm/arm.h (TARGET_DOTPROD): New.
3962 * config/arm/arm.c (arm_arch_dotprod): New.
3963 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3964 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3965 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3966 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3967 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3968 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3969 * doc/invoke.texi (armv8.2-a): Document dotprod
3971 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3973 * i386.c (ix86_vec_cost): New function.
3974 (ix86_rtx_costs): Handle vector operations better.
3975 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3976 * x86-tune-costs.h: Add new costs to all tables.
3978 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3980 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3982 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3983 divsd, sqrtss and sqrtsd
3984 * x86-tune-costs.h: Add new entries to all costs.
3985 (znver1_cost): Fix to match real instruction latencies.
3987 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3988 Michael Collison <michael.collison@arm.com>
3990 * compare-elim.c: Include emit-rtl.h.
3991 (can_merge_compare_into_arith): New function.
3992 (try_validate_parallel): Likewise.
3993 (try_merge_compare): Likewise.
3994 (try_eliminate_compare): Call the above when no previous clobber
3996 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
3999 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4003 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4004 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4008 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4009 to be any operand_equal_p operands. For & (B - 1) require
4010 B to be power of 2. Recognize
4011 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4013 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4016 * optabs.c (expand_memory_blockage): Fix call of
4017 targetm.have_memory_blockage.
4019 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4022 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4023 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4024 objects to extra_objs instead of overwriting it.
4026 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4028 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4029 Use any_fp_register_operand as operand[3] predicate. Simplify
4030 equality test for operands[2] and operands[4] memory location.
4031 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4032 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4033 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4034 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4035 any_fp_register_operand as operand[1] predicate. Simplify
4036 equality test for operands[0] and operands[3] memory location.
4037 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4038 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4039 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4041 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4043 * target-insns.def: Add memory_blockage.
4044 * optabs.c (expand_memory_blockage): New function.
4045 (expand_asm_memory_barrier): Rename ...
4046 (expand_asm_memory_blockage): ... to this.
4047 (expand_mem_thread_fence): Call expand_memory_blockage
4048 instead of expand_asm_memory_barrier.
4049 (expand_mem_singnal_fence): Ditto.
4050 (expand_atomic_load): Ditto.
4051 (expand_atomic_store): Ditto.
4052 * doc/md.texi (Standard Pattern Names For Generation):
4053 Document memory_blockage instruction pattern.
4055 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4057 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4058 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4059 * config/rl78/rl78.md: New define_expand "adddi3".
4061 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4063 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4066 2017-10-13 Jeff Law <law@redhat.com>
4068 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4070 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4073 * internal-fn.c (expand_mul_overflow): If both operands have
4074 the same highpart of -1 or 0 and the topmost bit of lowpart
4075 is different, overflow is if res <= 0 rather than res < 0.
4077 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4079 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4080 TARGET_P9_VECTOR code for unaligned_load case.
4082 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4084 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4086 2017-10-13 Nathan Sidwell <nathan@acm.org>
4088 * tree-core.h (tree_contains_struct): Make bool.
4089 * tree.c (tree_contains_struct): Likewise.
4090 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4091 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4092 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4093 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4094 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4096 2017-10-13 Richard Biener <rguenther@suse.de>
4098 * graphite-isl-ast-to-gimple.c
4099 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4100 parameters and dominance check.
4101 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4102 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4103 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4104 Do not update SSA form here or do intermediate IL verification.
4105 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4106 (graphite_initialize): Remove check on the number of loops in
4107 the function and inline into graphite_transform_loops.
4108 (graphite_finalize): Inline into graphite_transform_loops.
4109 (graphite_transform_loops): Perform SSA update and IL verification
4111 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4113 2017-10-13 Richard Biener <rguenther@suse.de>
4115 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4116 graphite_expression_type_precision): Avoid global constructor
4118 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4119 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4120 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4121 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4122 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4123 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4125 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4128 * config/i386/i386.h (ix86_red_zone_size): New.
4129 * config/i386/i386.md (push peephole2s): Replace
4130 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4132 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4133 Alan Hayward <alan.hayward@arm.com>
4134 David Sherwood <david.sherwood@arm.com>
4136 * combine.c (can_change_dest_mode): Reject changes in
4137 REGMODE_NATURAL_SIZE.
4139 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4140 Alan Hayward <alan.hayward@arm.com>
4141 David Sherwood <david.sherwood@arm.com>
4143 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4144 (expand_debug_source_expr): Likewise.
4145 * combine.c (combine_simplify_rtx): Likewise.
4146 * cse.c (fold_rtx): Likewise.
4147 * fwprop.c (canonicalize_address): Likewise.
4148 * targhooks.c (default_shift_truncation_mask): Likewise.
4150 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4151 Alan Hayward <alan.hayward@arm.com>
4152 David Sherwood <david.sherwood@arm.com>
4154 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4155 (widened_mode): Likewise.
4156 (expand_unop): Likewise.
4157 * ree.c (transform_ifelse): Likewise.
4158 (merge_def_and_ext): Likewise.
4159 (combine_reaching_defs): Likewise.
4160 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4162 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4163 Alan Hayward <alan.hayward@arm.com>
4164 David Sherwood <david.sherwood@arm.com>
4166 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4167 * combine.c (gen_lowpart_for_combine): Likewise.
4168 * dwarf2out.c (rtl_for_decl_location): Likewise.
4169 * final.c (alter_subreg): Likewise.
4170 * rtlhooks.c (gen_lowpart_general): Likewise.
4171 (gen_lowpart_if_possible): Likewise.
4173 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4174 Alan Hayward <alan.hayward@arm.com>
4175 David Sherwood <david.sherwood@arm.com>
4177 * calls.c (expand_call): Use subreg_lowpart_offset.
4178 * cse.c (cse_insn): Likewise.
4179 * regcprop.c (copy_value): Likewise.
4180 (copyprop_hardreg_forward_1): Likewise.
4182 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4185 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
4186 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
4187 =Q constraints to +Q and into insn condition add check
4188 that operands[0] and operands[1] are equal.
4189 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
4190 =Q constraints to +Q and into insn condition add check
4191 that operands[0] is equal to either operands[1] or operands[2].
4194 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
4195 instead of handling MINUS_EXPR twice (once for each argument),
4196 canonicalize operand order and handle just once, use rtype where
4197 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
4200 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
4201 any values of __C while still being pattern recognizable as a simple
4204 2017-10-13 Richard Biener <rguenther@suse.de>
4206 PR tree-optimization/82451
4208 2017-10-02 Richard Biener <rguenther@suse.de>
4210 PR tree-optimization/82355
4211 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4212 a mapping for the enclosing loop but avoid generating one for
4214 (copy_bb_and_scalar_dependences): Remove premature codegen
4215 error on PHIs in blocks duplicated into multiple places.
4216 * graphite-scop-detection.c
4217 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4218 in the region use it as loop and nest to analyze the DR in.
4219 (try_generate_gimple_bb): Likewise.
4220 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4221 (add_loop_constraints): For blocks in a loop not in the region
4222 create a dimension with a single iteration.
4223 * sese.h (gbb_loop_at_index): Remove assert.
4225 * cfgloop.c (loop_preheader_edge): For the loop tree root
4226 return the single successor of the entry block.
4227 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4228 Reset the SCEV hashtable and niters.
4229 * graphite-scop-detection.c
4230 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
4231 assert that we only have POLYNOMIAL_CHREC that vary in loops
4232 contained in the region.
4233 (scop_detection::graphite_can_represent_expr): Adjust.
4234 (scop_detection::stmt_has_simple_data_refs_p): For loops
4235 not in the region set loop to NULL. The nest is now the
4236 entry edge to the region.
4237 (try_generate_gimple_bb): Likewise.
4238 * sese.c (scalar_evolution_in_region): Adjust for
4239 instantiate_scev change.
4240 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4241 Make nest parameter the edge into the region.
4242 (create_data_ref): Likewise.
4243 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4244 entry edge into a region and adjust instantiate_scev calls.
4245 (create_data_ref): Likewise.
4246 (graphite_find_data_references_in_stmt): Likewise.
4247 (find_data_references_in_stmt): Pass the loop preheader edge
4248 from the nest argument.
4249 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4250 parameter the edge into the region.
4251 (instantiate_parameters): Use the loop preheader edge as entry.
4252 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4254 (get_instantiated_value_entry): Make instantiate_below parameter
4255 the edge into the region.
4256 (instantiate_scev_name): Likewise. Adjust dominance checks,
4257 when we cannot use loop-based instantiation instantiate by
4258 walking use-def chains.
4259 (instantiate_scev_poly): Adjust.
4260 (instantiate_scev_binary): Likewise.
4261 (instantiate_scev_convert): Likewise.
4262 (instantiate_scev_not): Likewise.
4263 (instantiate_array_ref): Remove.
4264 (instantiate_scev_3): Likewise.
4265 (instantiate_scev_2): Likewise.
4266 (instantiate_scev_1): Likewise.
4267 (instantiate_scev_r): Do not blindly handle N-operand trees.
4268 Do not instantiate array-refs. Handle all constants and invariants.
4269 (instantiate_scev): Make instantiate_below parameter
4270 the edge into the region.
4271 (resolve_mixers): Use the loop preheader edge for the region
4272 parameter to instantiate_scev_r.
4273 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4275 2017-10-13 Richard Biener <rguenther@suse.de>
4277 PR tree-optimization/82525
4278 * graphite-isl-ast-to-gimple.c
4279 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4281 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4282 Fail code generation when we cannot represent the isl integer.
4283 (binary_op_to_tree): Elide modulo operations that are no-ops
4284 in the type we code generate. Remove now superfluous code
4287 2017-10-13 Richard Biener <rguenther@suse.de>
4289 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4290 (scop_detection::harmful_loop_in_region): Remove premature
4291 IV type restriction.
4292 (scop_detection::graphite_can_represent_scev): We can handle
4293 pointer IVs just fine.
4295 2017-10-13 Alan Modra <amodra@gmail.com>
4297 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4298 "Clobbers and Scratch Registers". Add paragraph on
4299 alternative to clobbers for scratch registers and OpenBLAS
4302 2017-10-13 Alan Modra <amodra@gmail.com>
4304 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4305 example of a memory input for a string of known length. Move
4306 commentary out of table. Add a number of new examples
4307 covering array memory inputs.
4309 2017-10-12 Martin Liska <mliska@suse.cz>
4311 PR tree-optimization/82493
4312 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4313 (test_range_functions): New function.
4314 (sbitmap_c_tests): Likewise.
4315 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4316 * selftest.h (sbitmap_c_tests): New function.
4318 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4320 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4322 * config/rs6000/amo.h: Fix spacing issue.
4324 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4327 * config/i386/i386.md (*ashl<mode>3_mask_1,
4328 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4329 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4332 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4334 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4335 (profile_probability): Set max_probability
4336 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4338 * profile-count.c (profile_probability::differs_from_p): Do not
4339 rely on max_probaiblity == 10000
4341 2017-10-12 Jeff Law <law@redhat.com>
4343 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4346 2017-10-12 Martin Sebor <msebor@redhat.com>
4350 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4351 (handle_alias_pairs): Call it.
4352 * common.opt (-Wattribute-alias): New option.
4353 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4354 * doc/invoke.texi (-Wattribute-alias): Document.
4356 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4359 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4361 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4363 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4364 (make_hard_regno_born, make_hard_regno_dead): Update
4365 bb_killed_pseudos and bb_gen_pseudos.
4367 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4369 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4371 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4373 * config/alpha/alpha.c (alpha_split_conditional_move):
4374 Use std::swap instead of manually swapping.
4375 (alpha_stdarg_optimize_hook): Ditto.
4376 (alpha_canonicalize_comparison): Ditto.
4378 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4380 * tree-loop-distribution.c (struct builtin_info): New struct.
4381 (struct partition): Refactor fields into struct builtin_info.
4382 (partition_free): Free struct builtin_info.
4383 (build_size_arg_loc, build_addr_arg_loc): Delete.
4384 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4385 information from struct builtin_info.
4386 (find_single_drs): New function refactored from classify_partition.
4387 Also moved builtin validity checks to this function.
4388 (compute_access_range, alloc_builtin): New functions.
4389 (classify_builtin_st, classify_builtin_ldst): New functions.
4390 (classify_partition): Refactor code into functions find_single_drs,
4391 classify_builtin_st and classify_builtin_ldst.
4392 (distribute_loop): Don't do runtime alias check when distributing
4394 (find_seed_stmts_for_distribution): New function.
4395 (pass_loop_distribution::execute): Refactor code finding seed
4396 stmts into above function. Support distribution for the innermost
4397 two-level loop nest. Adjust dump information.
4399 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4401 * tree-loop-distribution.c: Adjust the general comment.
4402 (NUM_PARTITION_THRESHOLD): New macro.
4403 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4404 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4405 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4406 in call to build_partition_graph.
4407 (finalize_partitions): New parameter. Make loop distribution more
4408 conservative by fusing more partitions.
4409 (distribute_loop): Don't do runtime alias check in case of loop nest
4411 (find_seed_stmts_for_distribution): New function.
4412 (prepare_perfect_loop_nest): New function.
4413 (pass_loop_distribution::execute): Refactor code finding seed stmts
4414 and loop nest into above functions. Support loop nest distribution.
4415 Adjust dump information accordingly.
4417 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4419 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4420 and set PTYPE_SEQUENTIAL for merged partition.
4422 2017-10-12 Richard Biener <rguenther@suse.de>
4424 PR tree-optimization/69728
4426 2017-09-19 Richard Biener <rguenther@suse.de>
4428 PR tree-optimization/69728
4429 * graphite-sese-to-poly.c (schedule_error): New global.
4430 (add_loop_schedule): Handle empty domain by failing the
4432 (build_original_schedule): Handle schedule_error.
4434 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4435 domain by returning an unchanged schedule.
4437 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4439 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4440 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4442 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4444 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4447 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4450 * expr.c (store_field): Don't optimize away bitsize == 0 store
4451 from CALL_EXPR with addressable return type.
4453 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4455 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4456 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4457 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4458 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4459 TARGET_ISEL instead of TARGET_ISEL<sel>.
4461 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4463 * config/rs6000/rs6000.c
4464 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4466 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4468 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4470 (reg_or_cint_operand): Fix comment.
4471 (reg_or_zero_operand): New predicate.
4472 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4473 * config/rs6000/rs6000.c (output_isel): Delete.
4474 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4475 instead of reg_or_cint_operand. Output instruction directly (not via
4477 (isel_unsigned_<mode>): Ditto.
4478 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4479 gpc_reg_operand. Add an instruction alternative for this. Output
4480 instruction directly.
4481 (*isel_reversed_unsigned_<mode>): Ditto.
4483 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4485 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4486 (TARGET_CANONICALIZE_COMPARISON): Define.
4488 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4491 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4492 Check whether the dest is REG before adding REG_EQUIV note.
4494 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4497 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4499 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4500 (make_hard_regno_born, make_hard_regno_dead): Update
4501 bb_killed_pseudos and bb_gen_pseudos.
4503 2017-10-11 Nathan Sidwell <nathan@acm.org>
4505 * incpath.h (enum incpath_kind): Name enum, prefix values.
4506 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4507 * incpath.c (heads, tails): Use INC_MAX.
4508 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4509 (merge_include_chains, split_quote_chain,
4510 register_include_chains): Update incpath_kind names.
4511 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4512 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4514 (add_framework_path, darwin_register_objc_includes): Likewise.
4515 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4517 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4519 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4520 Do not use float_operator operator predicate.
4521 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4522 * config/i386/predicates.md (float_operator): Remove predicate.
4524 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4526 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4527 (*jccxf_i387): Ditto.
4528 (*jcc<mode>_i387): Ditto.
4529 (*jccu<mode>_i387): Ditto.
4530 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4531 (*jcc_*_i387 splitters): Remove.
4532 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4533 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4534 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4537 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4539 * profile-count.h (slow_safe_scale_64bit): New function.
4540 (safe_scale_64bit): New inline.
4541 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4542 * profile-count.c: Include wide-int.h
4543 (slow_safe_scale_64bit): New.
4545 2017-10-11 Nathan Sidwell <nathan@acm.org>
4547 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4548 HAS_DECL_ASSEMBLER_NAME_P.
4549 * gimple-expr.c (gimple_decl_printable_name: Check
4550 HAS_DECL_ASSEMBLER_NAME_P too.
4551 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4552 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4553 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4554 * passes.c (rest_of_decl_compilation): Check
4555 HAS_DECL_ASSEMBLER_NAME_P too.
4556 * recog.c (verify_changes): Likewise.
4557 * tree-pretty-print.c (dump_decl_name): Likewise.
4558 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4560 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4561 (SET_DECL_ASSEMBLER_NAME): Use it.
4562 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4563 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4564 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4566 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4568 * config.gcc (i386, x86_64): Add extra objects.
4569 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4570 (ix86_min_insn_size): Declare.
4571 (ix86_issue_rate): Declare.
4572 (ix86_adjust_cost): Declare.
4573 (ia32_multipass_dfa_lookahead): Declare.
4574 (ix86_macro_fusion_p): Declare.
4575 (ix86_macro_fusion_pair_p): Declare.
4576 (ix86_bd_has_dispatch): Declare.
4577 (ix86_bd_do_dispatch): Declare.
4578 (ix86_core2i7_init_hooks): Declare.
4579 (ix86_atom_sched_reorder): Declare.
4580 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4581 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4582 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4583 (rip_relative_addr_p): Rename to ...
4584 (ix86_rip_relative_addr_p): ... this one; export.
4585 (memory_address_length): Update.
4586 (ix86_issue_rate): Move to x86-tune-sched.c.
4587 (ix86_flags_dependent): Move to x86-tune-sched.c.
4588 (ix86_agi_dependent): Move to x86-tune-sched.c.
4589 (exact_dependency_1): Move to x86-tune-sched.c.
4590 (exact_store_load_dependency): Move to x86-tune-sched.c.
4591 (ix86_adjust_cost): Move to x86-tune-sched.c.
4592 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4593 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4594 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4595 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4596 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4597 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4598 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4599 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4600 (min_insn_size): Rename to ...
4601 (ix86_min_insn_size): ... this one; export.
4602 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4603 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4604 (core2i7_first_cycle_multipass_backtrack): Move to
4605 x86-tune-sched-core.c.
4606 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4607 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4608 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4609 (ix86_avoid_jump_mispredicts): Update.
4610 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4611 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4612 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4613 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4614 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4615 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4616 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4617 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4618 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4619 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4620 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4621 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4622 (BIG): Move to ix86-tune-sched-bd.c.
4623 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4624 (enum insn_path): Move to ix86-tune-sched-bd.c.
4625 (get_mem_group): Move to ix86-tune-sched-bd.c.
4626 (is_cmp): Move to ix86-tune-sched-bd.c.
4627 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4628 (is_branch): Move to ix86-tune-sched-bd.c.
4629 (is_prefetch): Move to ix86-tune-sched-bd.c.
4630 (init_window): Move to ix86-tune-sched-bd.c.
4631 (allocate_window): Move to ix86-tune-sched-bd.c.
4632 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4633 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4634 (process_end_window): Move to ix86-tune-sched-bd.c.
4635 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4636 (find_constant): Move to ix86-tune-sched-bd.c.
4637 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4638 (has_immediate): Move to ix86-tune-sched-bd.c.
4639 (get_insn_path): Move to ix86-tune-sched-bd.c.
4640 (get_insn_group): Move to ix86-tune-sched-bd.c.
4641 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4642 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4643 (add_insn_window): Move to ix86-tune-sched-bd.c.
4644 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4645 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4646 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4647 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4648 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4649 (do_dispatch): Move to ix86-tune-sched-bd.c.
4650 (has_dispatch): Move to ix86-tune-sched-bd.c.
4651 * config/i386/t-i386: Add new object files.
4652 * config/i386/x86-tune-costs.h: New file.
4653 * config/i386/x86-tune-sched-atom.c: New file.
4654 * config/i386/x86-tune-sched-bd.c: New file.
4655 * config/i386/x86-tune-sched-core.c: New file.
4656 * config/i386/x86-tune-sched.c: New file.
4658 2017-10-11 Liu Hao <lh_mouse@126.com>
4660 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4661 the generic version below instead.
4662 (should_colorize): Recognize Windows consoles as terminals
4664 * pretty-print.c [__MINGW32__] (write_all): New function.
4665 [__MINGW32__] (find_esc_head): Likewise.
4666 [__MINGW32__] (find_esc_terminator): Likewise.
4667 [__MINGW32__] (eat_esc_sequence): Likewise.
4668 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4670 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
4673 2017-10-11 Richard Biener <rguenther@suse.de>
4675 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
4676 Properly call analyze_scalar_evolution with the loop of the stmt.
4678 2017-10-11 Richard Biener <rguenther@suse.de>
4680 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
4681 * tree-core.h (tree_base): Add chrec_var union member.
4682 * tree.h (CHREC_VAR): Remove.
4683 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
4684 * tree-chrec.h (build_polynomial_chrec): Adjust.
4685 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
4686 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
4688 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4690 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4691 * match.pd: ... here.
4692 ((T) X == (T) Y): Relax condition.
4694 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4696 PR tree-optimization/82472
4697 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4699 (break_alias_scc_partitions): Update postorder number.
4701 2017-10-11 Martin Liska <mliska@suse.cz>
4704 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4706 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4708 2017-10-11 Martin Liska <mliska@suse.cz>
4710 * print-rtl.c (print_insn): Move declaration of idbuf
4711 to same scope as name.
4713 2017-10-11 Martin Liska <mliska@suse.cz>
4718 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4721 2017-10-11 Martin Liska <mliska@suse.cz>
4724 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4727 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4729 * config.gcc (default_gnu_indirect_function): Default to yes for
4730 arm*-*-linux* with glibc.
4732 2017-10-11 Richard Biener <rguenther@suse.de>
4734 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4735 default-defs and types we do not want to analyze.
4736 (interpret_loop_phi): Replace unreachable code with an assert.
4737 (compute_scalar_evolution_in_loop): Remove and inline ...
4738 (analyze_scalar_evolution_1): ... here, replacing condition with
4739 what makes the intent clearer. Remove handling of cases
4740 get_scalar_evolution now handles.
4742 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4744 PR rtl-optimization/81434
4745 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4746 comment for main loop. In sched_group_found if, also add checks for
4747 pass and min_cost_group.
4749 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4751 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4752 (rs6000_insn_cost): New function.
4753 * config/rs6000/rs6000.md (cost): New attribute.
4755 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4756 H.J. Lu <hongjiu.lu@intel.com>
4760 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4761 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4762 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4763 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4764 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4765 OPTION_MASK_ISA_MMX and some other bit set require both
4766 mmx and the ISAs without the mmx bit.
4767 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4768 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4769 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4770 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4771 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4772 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4773 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4774 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4775 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4776 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4777 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4778 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4779 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4780 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4781 Add OPTION_MASK_ISA_MMX.
4783 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4785 * config.gcc (armv7*-*-freebsd*): New target.
4786 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4788 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4790 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4791 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4794 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4796 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4797 a template parameter.
4798 (WIDE_INT_REF_FOR): Update accordingly.
4799 * tree.h (wi::int_traits <const_tree>): Delete.
4800 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4801 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4802 (wi::tree_to_wide_ref): New typedef.
4803 (wi::to_wide): New function.
4804 * calls.c (get_size_range): Use wi::to_wide when operating on
4806 * cgraph.c (cgraph_node::create_thunk): Likewise.
4807 * config/i386/i386.c (ix86_data_alignment): Likewise.
4808 (ix86_local_alignment): Likewise.
4809 * dbxout.c (stabstr_O): Likewise.
4810 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4811 * expr.c (const_vector_from_tree): Likewise.
4812 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4813 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4814 (fold_negate_expr_1, int_const_binop_1, const_binop)
4815 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4816 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4817 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4818 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4819 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4820 (fold_not_const, round_up_loc): Likewise.
4821 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4822 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4823 (alloca_call_type): Likewise.
4824 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4825 * godump.c (go_output_typedef): Likewise.
4826 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4827 * internal-fn.c (get_min_precision): Likewise.
4828 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4829 * ipa-polymorphic-call.c
4830 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4831 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4832 (ipa_modify_call_arguments): Likewise.
4833 * match.pd: Likewise.
4834 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4835 * print-tree.c (print_node_brief, print_node): Likewise.
4836 * stmt.c (expand_case): Likewise.
4837 * stor-layout.c (layout_type): Likewise.
4838 * tree-affine.c (tree_to_aff_combination): Likewise.
4839 * tree-cfg.c (group_case_labels_stmt): Likewise.
4840 * tree-data-ref.c (dr_analyze_indices): Likewise.
4841 (prune_runtime_alias_test_list): Likewise.
4842 * tree-dump.c (dequeue_and_dump): Likewise.
4843 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4844 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4845 * tree-pretty-print.c (dump_generic_node): Likewise.
4846 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4847 (simple_iv_with_niters): Likewise.
4848 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4849 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4850 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4851 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4852 (refine_value_range_using_guard, number_of_iterations_ne_max)
4853 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4854 (get_cst_init_from_scev, record_nonwrapping_iv)
4855 (scev_var_range_cant_overflow): Likewise.
4856 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4857 * tree-ssa-pre.c (compute_avail): Likewise.
4858 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4859 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4860 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4861 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4862 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4863 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4864 (dump_case_nodes, try_switch_expansion): Likewise.
4865 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4866 (vect_do_peeling): Likewise.
4867 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4868 * tree-vect-stmts.c (vectorizable_load): Likewise.
4869 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4870 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4871 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4872 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4873 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4874 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4875 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4876 (evrp_dom_walker::before_dom_children): Likewise.
4877 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4878 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4879 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4880 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4881 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4882 (get_range_pos_neg): Likewise.
4883 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4884 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4885 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4886 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4887 * config/avr/avr.c (avr_fold_builtin): Likewise.
4888 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4889 * config/msp430/msp430.c (msp430_attr): Likewise.
4890 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4891 * config/powerpcspe/powerpcspe-c.c
4892 (altivec_resolve_overloaded_builtin): Likewise.
4893 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4894 (rs6000_expand_ternop_builtin): Likewise.
4895 * config/rs6000/rs6000-c.c
4896 (altivec_resolve_overloaded_builtin): Likewise.
4897 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4898 (rs6000_expand_ternop_builtin): Likewise.
4899 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4901 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4903 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4904 when copying loop nest with only one inner loop.
4906 2017-10-10 Richard Biener <rguenther@suse.de>
4908 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4909 blocks if SCEV is active.
4910 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4912 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4913 (scev_initialize): Assert we are not yet initialized.
4915 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4917 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4918 inner loop's exit stmt by making it always exit the loop, otherwise
4919 we would generate an infinite empty loop.
4921 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4923 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4924 renaming variables in new preheader if it's deleted.
4926 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4928 * tree-loop-distribution.c (struct partition): Remove unused field
4929 loops of the structure.
4930 (partition_alloc, partition_free): Ditto.
4931 (build_rdg_partition_for_vertex): Ditto.
4933 2017-10-09 Jeff Law <law@redhat.com>
4935 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4936 return type to match prototype and documentation.
4938 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4940 * config/rs6000/rs6000.c (processor_costs): Move to ...
4941 * config/rs6000/rs6000.h: ... here.
4942 (rs6000_cost): Declare.
4944 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4946 * except.c (setjmp_fn): New global variable.
4947 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4948 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4949 if DONT_USE_BUILTIN_SETJMP is defined.
4951 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4953 * target.def (insn_cost): New hook.
4954 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4955 * doc/tm.texi: Regenerate.
4956 * rtlanal.c (insn_cost): Use the new hook.
4958 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4960 * combine.c (combine_validate_cost): Compute the new insn_cost,
4961 not just pattern_cost.
4962 (try_combine): Adjust comment.
4964 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4966 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4968 * combine.c (uid_insn_cost): Adjust comment.
4969 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4971 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4972 * dse.c (find_shift_sequence): Ditto.
4973 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4974 (bb_valid_for_noce_process_p): Use pattern_cost.
4975 * rtl.h (insn_rtx_cost): Delete.
4976 (pattern_cost): New prototype.
4977 (insn_cost): New prototype.
4978 * rtlanal.c (insn_rtx_cost): Rename to...
4979 (pattern_cost): ... this.
4982 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4984 * config/i386/i386.md (*jcc_2): Remove insn pattern.
4985 (*jcc<mode>_0_r_i387): Ditto.
4986 (*jccxf_r_i387): Ditto.
4987 (*jcc<mode>_r_i387): Ditto.
4988 (*jccu<mode>_r_i387): Ditto.
4989 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
4990 (*jcc): Rename from *jcc_1.
4992 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4994 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
4995 deferred rescans after the lvx/stvx recombination pre-pass.
4997 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
4999 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5000 memory operation instruction support.
5001 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5002 (rs6000-ibm-aix[789]*): Likewise.
5003 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5004 Document new functions.
5006 2017-10-09 Richard Biener <rguenther@suse.de>
5008 PR tree-optimization/82397
5009 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5010 equality only for semantically equal trees.
5012 2017-10-09 Richard Biener <rguenther@suse.de>
5014 PR tree-optimization/82449
5015 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5016 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5017 allow constant addresses.
5018 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5021 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5023 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5026 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5029 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5032 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5035 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5037 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5040 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5041 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5043 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5045 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5046 (WI_BINARY_PREDICATE_RESULT): Likewise.
5047 (wi::binary_traits::operator_result): New type.
5048 (wi::binary_traits::predicate_result): Likewise.
5049 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5050 (generic_wide_int::operator==, generic_wide_int::operator!=)
5051 (generic_wide_int::operator&, generic_wide_int::and_not)
5052 (generic_wide_int::operator|, generic_wide_int::or_not)
5053 (generic_wide_int::operator^, generic_wide_int::operator+
5054 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5056 (operator~, unary operator-, operator==, operator!=, operator&)
5057 (operator|, operator^, operator+, binary operator-, operator*): New
5059 * expr.c (get_inner_reference): Use wi::bit_and_not.
5060 * fold-const.c (fold_binary_loc): Likewise.
5061 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5062 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5063 (bit_value_binop): Likewise.
5064 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5065 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5066 (extract_range_from_binary_expr_1): Likewise.
5067 (masked_increment): Likewise.
5068 (simplify_bit_ops_using_ranges): Likewise.
5070 2017-10-09 Martin Jambor <mjambor@suse.cz>
5073 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5074 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5075 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5077 (hsa_fixup_mov_insn_type): New function.
5078 (hsa_op_with_type::get_in_type): Use it.
5079 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5080 immediates in an assert.
5081 (hsa_op_with_type::extend_int_to_32bit): New method.
5082 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5083 types. Convert to dest type if necessary.
5084 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5085 (reg_for_gimple_ssa): Pass false as min32int to
5086 hsa_type_for_scalar_tree_type.
5087 (gen_hsa_addr): Fixup type when creating addresable temporary.
5088 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5089 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5090 necessary. Call hsa_fixup_mov_insn_type.
5091 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5092 extend operands and convert to dest type if necessary.
5093 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5094 to dest type if necessary.
5095 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5096 if conversion nt necessary and size matches.
5097 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5098 to dest type if necessary.
5099 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5100 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5102 (gen_hsa_clrsb): Likewise.
5103 (gen_hsa_ffs): Likewise.
5104 (gen_hsa_divmod): Extend operands and convert to dest type if
5106 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5108 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5110 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5111 Remove empty default arguments. Use a brace block as output
5113 (conditional return): Ditto.
5115 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5116 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5118 (group_ending_nop): Ditto.
5119 (doloop_end): Ditto.
5120 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5121 (splitters for those): Ditto.
5123 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5125 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5126 a conditional jump (and the compare for it) so that pc_rtx is the
5128 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5129 for the deleted and renamed ctr<mode>_internal[234] patterns.
5130 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5131 Delete second conditional return pattern.
5132 (ctr<mode>_internal2): Delete this second bdnz pattern.
5133 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5134 (ctr<mode>_internal4): Delete this second bdz pattern.
5136 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5138 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5139 (always_initialized_rtx_for_ssa_name_p): New predicate.
5140 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5141 (finish_out_of_ssa): Free new field of SA.
5142 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5143 * tree-ssa-coalesce.c: Include tree-ssa.h.
5144 (get_parm_default_def_partitions): Remove extern keyword.
5145 (get_undefined_value_partitions): New function.
5146 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5147 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5149 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5151 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5153 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5155 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5157 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5158 for targets that preffer 128bit.
5160 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5162 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5164 2017-10-08 Olivier Hainque <hainque@adacore.com>
5166 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5167 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5169 (thumb_set_return_address): Likewise.
5171 2017-10-08 Olivier Hainque <hainque@adacore.com>
5173 * common/config/arm/arm-common.c (arm_except_unwind_info):
5174 Handle DWARF2_UNWIND_INFO.
5176 2017-10-07 Michael Collison <michael.collison@arm.com>
5178 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
5181 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
5183 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
5184 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
5185 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
5186 defined, force the creation of a new block for a dispatch label.
5188 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
5190 * invoke.texi (Wsuggest-attribute=cold): Document.
5191 * common.opt (Wsuggest-attribute=cold): New
5192 * ipa-pure-const.c (warn_function_cold): New function.
5193 * predict.c (compute_function_frequency): Use it.
5194 * predict.h (warn_function_cold): Declare.
5196 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
5198 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
5201 2017-10-06 Martin Liska <mliska@suse.cz>
5203 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
5204 keyword for member functions.
5205 (struct sanopt_tree_couple): New struct.
5206 (struct sanopt_tree_couple_hash): New function.
5207 (struct sanopt_ctx): Add new hash_map.
5208 (has_dominating_ubsan_ptr_check): New function.
5209 (record_ubsan_ptr_check_stmt): Likewise.
5210 (maybe_optimize_ubsan_ptr_ifn): Likewise.
5211 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
5212 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
5214 2017-10-06 Sudakshina Das <sudi.das@arm.com>
5217 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
5218 aarch64_simd_valid_immediate on CONST_VECTORs.
5219 (aarch64_reg_or_bic_imm): Likewise.
5221 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
5223 PR rtl-optimization/82396
5224 * haifa-sched.c (ready_sort_real): Disable qsort checking.
5226 2017-10-06 Sebastian Pop <sebpop@gmail.com>
5228 * graphite-dependences.c (scop_get_reads): Move code to...
5229 (scop_get_must_writes): Move code to...
5230 (scop_get_may_writes): Move code to...
5231 (scop_get_reads_and_writes): ... here.
5232 (scop_get_dependences): Call scop_get_reads_and_writes.
5234 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5236 PR tree-optimization/82434
5237 * fold-const.h (can_native_encode_type_p,
5238 can_native_encode_string_p): Remove.
5239 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5240 don't encode anything, just return what would be otherwise returned.
5241 (native_encode_fixed, native_encode_complex, native_encode_vector):
5243 (native_encode_string): Likewise. Inline by hand
5244 can_native_encode_string_p.
5245 (can_native_encode_type_p): Remove.
5246 (can_native_encode_string_p): Remove.
5247 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5248 STRING_CSTs using can_native_encode_string_p, test all
5249 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5250 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5251 argument from native_encode_expr.
5252 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5253 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5256 2017-10-06 Richard Biener <rguenther@suse.de>
5258 PR tree-optimization/82397
5259 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5260 operand_equal_p but rely on data_ref_compare_tree for detecting
5262 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5263 to match up with dr_group_sort_cmp.
5265 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5268 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5270 * config/s390/s390-builtin-types.def: Regenerate.
5272 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5275 * config/s390/s390-builtin-types.def: Regenerate.
5276 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5277 Change flag from B_VXE to B_VX.
5278 (s390_vec_min_dbl): Remove B_VXE flag.
5280 2017-10-06 Richard Biener <rguenther@suse.de>
5282 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5283 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5284 translate_isl_ast_to_gimple::is_valid_rename,
5285 translate_isl_ast_to_gimple::get_rename,
5286 translate_isl_ast_to_gimple::get_def_bb_for_const,
5287 translate_isl_ast_to_gimple::get_new_name,
5288 translate_isl_ast_to_gimple::collect_all_ssa_names,
5289 translate_isl_ast_to_gimple::copy_loop_phi_args,
5290 translate_isl_ast_to_gimple::collect_all_ssa_names,
5291 translate_isl_ast_to_gimple::copy_loop_phi_args,
5292 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5293 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5294 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5295 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5296 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5297 translate_isl_ast_to_gimple::copy_cond_phi_args,
5298 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5299 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5300 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5301 translate_isl_ast_to_gimple::rename_uses,
5302 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5303 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5304 (set_rename_for_each_def): Likewise.
5305 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5306 here. Handle rewriting SCEV analyzable uses here.
5307 (copy_bb_and_scalar_dependences): Generate code for PHI
5309 (graphite_regenerate_ast_isl): Adjust.
5310 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5311 (add_write, add_read): New functions.
5312 (build_cross_bb_scalars_def): Use it and simplify.
5313 (build_cross_bb_scalars_use): Likewise.
5314 (graphite_find_cross_bb_scalar_vars): Inline into...
5315 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5316 simulating out-of-SSA. Compute liveout and add dependencies.
5317 (build_scops): Force an empty entry block.
5318 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5320 (sese_build_liveouts): Declare.
5321 (sese_trivially_empty_bb_p): Likewise.
5322 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5323 compute liveout and debug_liveout.
5324 (sese_bad_liveouts_use): Remove.
5325 (sese_reset_debug_liveouts_bb): Likewise.
5326 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5327 (sese_build_liveouts): Build liveout and debug_liveout and store
5329 (new_sese_info): Adjust.
5330 (free_sese_info): Likewise.
5331 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5332 do not build liveout here.
5333 (move_sese_in_condition): Adjust region entry.
5334 (scev_analyzable_p): Match up with chrec_apply requirements.
5335 (sese_trivially_empty_bb_p): New.
5336 * tree-into-ssa.c (get_reaching_def): Properly support generating
5337 default-defs for incremental rewrite of anonymous names.
5339 2017-10-06 Richard Biener <rguenther@suse.de>
5341 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5342 precision do not perform modulo reduction.
5344 2017-10-06 Richard Biener <rguenther@suse.de>
5346 PR tree-optimization/82436
5347 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5348 conservatively choose the vectorization factor when checking
5349 whether we can perform the required load permutation.
5350 (vect_transform_slp_perm_load): Assert when we may not fail.
5352 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5354 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5355 message for incompatible -msdata=* and -mcall-* options.
5357 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5359 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5360 rate for post-reload scheduling.
5362 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5364 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5366 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5368 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5369 to improve monte carlo in scimark.
5371 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5373 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5374 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5375 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5376 pentium4_cost, nocona_cost): Set reassociation width to 1.
5377 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5378 width to 2 for fp operations and 1 otherwise.
5379 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5381 (atom_cost): Set reassociation width to 2.
5382 (slm_cost, generic_cost): Set fp reassociation width
5383 to 2 and 1 otherwise.
5384 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5385 (core_cost): Set fp reassociation width to 4 and vector to 2.
5386 (ix86_reassociation_width): Rewrite using cost table; special case
5387 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5388 and TARGET_AVX128_OPTIMAL.
5389 * config/i386/i386.h (processor_costs): Add
5390 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5391 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5392 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5393 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5394 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5395 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5397 2017-10-05 Nathan Sidwell <nathan@acm.org>
5399 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5401 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5403 * config/arm/arm.c (arm_test_fpu_data): New.
5404 (arm_run_selftests): Call arm_test_fpu_data.
5406 2017-10-04 Nathan Sidwell <nathan@acm.org>
5408 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5410 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5412 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5414 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5415 check type for aarch64_simd_valid_immediate.
5416 (aarch64_output_simd_mov_immediate): Update prototype.
5417 (aarch64_simd_valid_immediate): Update prototype.
5418 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5419 support for ORR-immediate.
5420 (and<mode>3): modified pattern to add support for BIC-immediate.
5421 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5422 now checks for valid immediate for BIC and ORR based on new enum
5424 (aarch64_output_simd_mov_immediate): Function now used to output
5425 BIC/ORR imm as well based on new enum argument.
5426 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5428 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5429 (aarch64_reg_or_bic_imm): Likewise.
5431 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5433 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5434 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5436 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5440 PR rtl-optimization/82396
5441 * haifa-sched.c (autopref_multipass_init): Simplify
5443 (autopref_rank_data): Simplify sort order.
5444 * sched-int.h (autopref_multipass_data_): Remove
5445 multi_mem_insn_p, min_offset and max_offset.
5447 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5449 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5451 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5453 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5456 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5458 * doc/sourcebuild.texi: Document vect_long_mult.
5460 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5462 PR tree-optimization/82413
5463 * fold-const.c (build_range_check): Use widest_int when comparing
5464 the maximum ETYPE value with HIGH.
5466 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5468 PR rtl-optimization/82396
5469 * haifa-sched.c (autopref_multipass_init): Simplify
5471 (autopref_rank_data): Simplify sort order.
5472 * sched-int.h (autopref_multipass_data_): Remove
5473 multi_mem_insn_p, min_offset and max_offset.
5475 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5477 PR tree-optimization/82381
5478 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5479 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5482 PR tree-optimization/82374
5483 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5484 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5485 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5486 current_function_decl to the new decl.
5488 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5490 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5491 helper macro for IEEE float128 hardware built-in functions.
5492 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5494 (TRUNCF128_ODD): Likewise.
5495 (ADDF128_ODD): Likewise.
5496 (SUBF128_ODD): Likewise.
5497 (MULF128_ODD): Likewise.
5498 (DIVF128_ODD): Likewise.
5499 (FMAF128_ODD): Likewise.
5500 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5501 UNSPEC_TRUNC_ROUND_TO_ODD.
5502 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5503 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5504 floating point round to odd instructions.
5505 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5506 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5507 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5508 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5509 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5510 (trunc<mode>sf2_hw): Change the truncate with round to odd
5511 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5512 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5513 to odd hardware instructions.
5514 (sub<mode>3_odd): Likewise.
5515 (mul<mode>3_odd): Likewise.
5516 (div<mode>3_odd): Likewise.
5517 (sqrt<mode>2_odd): Likewise.
5518 (fma<mode>4_odd): Likewise.
5519 (fms<mode>4_odd): Likewise.
5520 (nfma<mode>4_odd): Likewise.
5521 (nfms<mode>4_odd): Likewise.
5522 (trunc<mode>df2_odd): Change the truncate with round to odd
5523 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5525 * doc/extend.texi (PowerPC built-in functions): Update documentation
5526 for existing IEEE float128-bit built-in functions. Add built-in
5527 functions that generate the IEEE 128-bit floating point round to
5530 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5532 PR rtl-optimization/77729
5533 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5534 to (X&(C1&~C2))|C2 transformations.
5536 2017-10-03 Martin Jambor <mjambor@suse.cz>
5538 PR tree-optimization/82363
5539 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5540 mismatch, mark lacc written regardless of racc.
5542 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5544 PR tree-optimization/82381
5545 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5546 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5547 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5548 fallthrough into reassoc_stmt_dominates_stmt_p.
5551 * combine.c (combine_instructions): Don't combine in unreachable
5554 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5557 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5558 function to not use the have_cpu variable. Do not set cpu_index,
5559 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5561 (rs6000_valid_attribute_p): Remove duplicate initializations of
5562 old_optimize and func_optimize.
5563 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5564 (rs6000_activate_target_options): Make global.
5565 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5568 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5570 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5571 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5572 Set *poffset to 0 and *psize and *pmax_size to -1 if
5573 *poffset + *psize overflows in HOST_WIDE_INT.
5575 PR tree-optimization/82387
5576 PR tree-optimization/82388
5577 PR tree-optimization/82389
5578 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5579 instead of live_bytes non-NULL.
5581 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5584 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5587 2017-10-02 Richard Biener <rguenther@suse.de>
5589 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5590 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5591 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5593 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5595 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5596 requested precision matches the type's.
5597 * calls.c (alloc_max_size): Calculate the new candidate size as
5598 a widest_int and use wi::to_widest when comparing it with the
5599 current candidate size.
5600 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5601 zero rather than integer_zero_node.
5602 * match.pd: Check for a no-op conversion before using wi::add
5603 rather than after. Use tree_to_uhwi when summing small shift
5604 counts into an unsigned int.
5606 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5607 Alan Hayward <alan.hayward@arm.com>
5608 David Sherwood <david.sherwood@arm.com>
5611 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5612 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5613 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5614 POINTER_AND_FP_REGS.
5616 2017-10-02 Richard Biener <rguenther@suse.de>
5618 PR tree-optimization/82355
5619 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5620 a mapping for the enclosing loop but avoid generating one for
5622 (copy_bb_and_scalar_dependences): Remove premature codegen
5623 error on PHIs in blocks duplicated into multiple places.
5624 * graphite-scop-detection.c
5625 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5626 in the region use it as loop and nest to analyze the DR in.
5627 (try_generate_gimple_bb): Likewise.
5628 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5629 (add_loop_constraints): For blocks in a loop not in the region
5630 create a dimension with a single iteration.
5631 * sese.h (gbb_loop_at_index): Remove assert.
5633 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5635 * omp-expand.c (adjust_context_scope): New function.
5636 (expand_parallel_call): Call adjust_context_scope.
5638 2017-10-01 Jeff Law <law@redhat.com>
5640 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5641 dom_opt_dom_walker class with direct access to private members.
5642 Add comments. Call test_for_singularity.
5643 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5644 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5645 m_dummy_cond anymore.
5646 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5648 (pass_dominator:execute): Build the dummy_cond here and pass it
5649 to the dom_opt_dom_walker ctor.
5650 (test_for_singularity): New function.
5652 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5653 Maya Rashish <coypu@sdf.org>
5655 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5656 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5657 (alpha*-*-netbsd*) Use nbsd_tm_file.
5658 (arm*-*-netbsdelf*) Likewise.
5659 (i[34567]86-*-netbsdelf*) Likewise.
5660 (x86_64-*-netbsd*) Likewise.
5661 (mips*-*-netbsd*) Likewise.
5662 (powerpc-*-netbsd*) Likewise.
5663 (sh*-*-netbsd*) Likewise.
5664 (sparc-*-netbsdelf*) Likewise.
5665 (sparc64-*-netbsd*) Likewise.
5666 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5668 (vax-*-netbsdelf*) Likewise.
5669 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5670 (UINT_FAST8_TYPE) Likewise.
5671 (INT_FAST16_TYPE) Check CHAR_FAST16.
5672 (UINT_FAST16_TYPE) Likewise.
5674 2017-09-30 Jakub Jelinek <jakub@redhat.com>
5677 * config/i386/i386.md
5678 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
5679 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
5680 *divmodsi4_zext_2): New define_insn_and_split.
5681 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
5682 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
5683 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
5684 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
5685 New define_insn_and_split.
5686 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
5687 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
5688 operands[1] having DImode when mode is SImode.
5690 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5691 always SImode for DIV and MOD in REG_EQUAL notes.
5693 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5696 * match.pd: Fix handling of NaNs in pattern.
5698 2017-09-29 Jeff Law <law@redhat.com>
5700 * sbitmap.c (bitmap_bit_in_range_p): New function.
5701 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5702 * tree-ssa-dse.c (live_bytes_read): New function.
5703 (dse_classify_store): Ignore reads of dead bytes.
5705 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5706 typos and whitespace errors.
5707 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5708 * config/s390/s390.c (s390_emit_prologue): Likewise.
5710 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5713 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5714 with a symbol for LRA.
5716 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5718 PR rtl-optimization/82338
5719 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5721 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5723 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5724 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5725 (qsort_chk): Declare.
5726 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5727 (qsort_chk): New function.
5729 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5731 PR tree-optimization/82337
5732 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5733 phi definition if the PHI result appears in an abnormal PHI.
5734 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5735 basis appears in an abnormal PHI.
5737 2017-09-29 Richard Biener <rguenther@suse.de>
5739 * graphite-isl-ast-to-gimple.c
5740 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5741 (binary_op_to_tree): Use it.
5742 (get_rename_from_scev): Likewise.
5743 (copy_loop_phi_nodes): Likewise.
5744 (copy_bb_and_scalar_dependences): Likewise.
5745 (translate_pending_phi_nodes): Likewise.
5747 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5750 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5751 for movabsq $(i32 << shift), r64.
5753 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5755 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5756 index when encoding %esp as %rsp to avoid 0x67 prefix.
5758 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5760 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5761 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5763 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5765 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5766 Extensions with more than 16 double VFP registers.
5767 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5768 to_clear_mask and all code related to it. Replace the remaining
5769 entry by a sbitmap and adapt code accordingly.
5771 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5773 * brig-builtins.def: Change pure attributes to const.
5775 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5777 * config.gcc (default_gnu_indirect_function): Default to yes for
5778 sparc*-*-linux* with glibc.
5780 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5782 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5783 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5784 when creating .init_array and .fini_array sections with priority
5787 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5790 * config/aarch64/aarch64.c
5791 (aarch64_builtin_support_vector_misalignment): Always return false
5792 when misalignment is unknown.
5794 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5796 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5797 this function to return false if the definition used by the swap
5798 instruction is artificial, or if the memory address from which the
5799 constant value is loaded is not represented by a base address held
5800 in a register or if the base address register is a frame or stack
5801 pointer. Additionally, return false if the base address of the
5802 loaded constant is a SYMBOL_REF but is not considered to be a
5804 (replace_swapped_load_constant): New function.
5805 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5806 loaded constant vector with a load of a swapped constant vector.
5808 2017-09-27 Carl Love <cel@us.ibm.com>
5810 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5811 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5812 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5815 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5817 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5818 first, always call autopref_rank_data otherwise.
5820 2017-09-27 Richard Biener <rguenther@suse.de>
5822 * graphite-scop-detection.c (find_scop_parameters): Move
5823 loop bound handling ...
5824 (gather_bbs::before_dom_children): ... here, avoiding the need
5825 to build scop_info->loop_nest.
5826 (record_loop_in_sese): Remove.
5827 * sese.h (sese_info_t::loop_nest): Remove.
5828 * sese.c (new_sese_info): Do not allocate loop_nest.
5829 (free_sese_info): Do not free loop_nest.
5831 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5834 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5835 lhs from calls if the lhs has addressable type.
5837 2017-09-27 Richard Biener <rguenther@suse.de>
5839 * graphite.h (scop::max_alias_set): New member.
5840 * graphite-scop-detection.c: Remove references to non-existing
5841 --param in comments.
5842 (build_alias_sets): Record the maximum alias set used for drs.
5843 (build_scops): Support zero as unlimited for
5844 --param graphite-max-arrays-per-scop.
5845 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5847 (build_poly_sr_1): ... here. Compute alias set based on the
5848 maximum alias set used for drs rather than
5849 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5851 2017-09-27 Richard Biener <rguenther@suse.de>
5853 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5854 --param loop-block-tile-size=0 to disable tiling.
5856 2017-09-27 Richard Biener <rguenther@suse.de>
5858 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5859 (graphite-max-nb-scop-params): Document special value zero.
5860 * domwalk.h (dom_walker::STOP): New symbolical constant.
5861 (dom_walker::dom_walker): Add optional parameter for bb to
5863 (dom_walker::~dom_walker): Declare.
5864 (dom_walker::before_dom_children): Document STOP return value.
5865 (dom_walker::m_user_bb_to_rpo): New member.
5866 (dom_walker::m_bb_to_rpo): Likewise.
5867 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5868 mapping here if not provided by the user.
5869 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5870 provided by the user.
5871 (dom_walker::STOP): Define.
5872 (dom_walker::walk): Do not compute bb to RPO mapping here.
5873 Support STOP return value from before_dom_children to stop
5875 * graphite-optimize-isl.c (optimize_isl): If the schedule
5876 is the same still generate code if -fgraphite-identity
5877 or -floop-parallelize-all are given.
5878 * graphite-scop-detection.c: Include cfganal.h.
5879 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5881 (gather_bbs::before_dom_children): Return STOP for BBs
5883 (build_scops): Compute bb to RPO mapping and pass it to
5884 the domwalk. Treat --param graphite-max-nb-scop-params=0
5885 as not limiting the number of params.
5886 * graphite.c (graphite_initialize): Remove limit on the
5887 number of basic-blocks in a function.
5888 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5889 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5890 default value of 10.
5892 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5894 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5895 Adjust code to eliminate needing to do the shift right 32-bits
5896 operation after XSCVDPSPN.
5898 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5900 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5901 ((X / Y) != 0 -> X >= Y): Likewise.
5903 2017-09-26 Carl Love <cel@us.ibm.com>
5905 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5906 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5907 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5908 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5909 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5910 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5911 definitions and overloading.
5912 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5913 statement for P9V_BUILTIN_XST_LEN_R.
5914 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5915 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5916 define_expand and define_insn for the instructions and builtins.
5917 * doc/extend.texi: Update the built-in documentation file for the new
5919 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5920 define_insn for the instructions
5922 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5925 * gcc/config/netbsd-protos.h: New file.
5926 * gcc/config/netbsd.c: New file.
5927 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5928 * gcc/config/t-netbsd: New file.
5929 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5930 (tmake_file) Add t-netbsd.
5931 (extra_objs) Add netbsd.o.
5933 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5937 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5939 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5941 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5942 sign extension from a vector register to a GPR by doing a 32-bit
5943 direct move and then an EXTSW.
5944 (extendsi<mode>2 splitter): Likewise.
5945 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5946 right or vector extract after doing XSCVDPSPN. Use
5947 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5949 (movdi_from_sf_zero_ext): Likewise.
5950 (reload_gpr_from_vsxsf): Likewise.
5951 (p8_mfvsrd_4_disf): Delete, no longer used.
5952 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5953 then needing to move the SFmode to a GPR to use the XSCVDPSP
5954 instruction instead of FRSP and XSCVDPSPN.
5955 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5956 it is adjacent to the other XSCVSPDP insns.
5957 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5958 SFmode to be in traditional Altivec registers.
5959 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5960 (vsx_xscvspdpn): Likewise.
5961 (vsx_xscvspdpn_scalar): Likewise.
5963 2017-09-26 Martin Jambor <mjambor@suse.cz>
5965 * tree-sra.c (compare_access_positions): Put integral types first,
5966 stabilize sorting of integral types, remove conditions putting
5967 non-full-precision integers last.
5968 (sort_and_splice_var_accesses): Disable scalarization if a
5969 non-integert would be represented by a non-full-precision integer.
5971 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5973 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5974 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5975 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5976 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5977 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5978 conditionals inside the function instead of around it. Call
5979 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5980 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5982 2017-09-26 Richard Biener <rguenther@suse.de>
5984 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5986 (scop_detection::build_scop_breadth): ... this. Removed.
5987 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
5988 (scop_detection::harmful_stmt_in_bb): Likewise.
5989 (scop_detection::graphite_can_represent_stmt): Likewise.
5990 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
5991 (scop_detection::can_represent_loop): Remove recursion, fold in ...
5992 (scop_detection::can_represent_loop_1): ... this. Removed.
5993 (scop_detection::harmful_loop_in_region): Simplify after inlining
5994 the above and remove more quadraticness.
5995 (build_scops): Adjust.
5996 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
5999 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6002 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6003 REGNO (base) == SP_REG if base is a REG.
6006 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6007 if it is different SSA_NAME.
6008 (optimize_range_tests_cmp_bitwise): New function.
6009 (optimize_range_tests): Call it.
6011 2017-09-26 Richard Biener <rguenther@suse.de>
6013 PR tree-optimization/82321
6014 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6015 for the def being inside the loop.
6017 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6019 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6021 * config/s390/s390-builtins.def: Fix constraint on op4.
6023 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6025 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6026 independent expanders.
6027 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6028 ("vec_ordered", "vec_unordered"): New expanders.
6030 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6032 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6035 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6037 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6038 vec_unpacks_lo_v16qi.
6039 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6041 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6043 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6044 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6045 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6047 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6049 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6051 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6053 ("*vec_slb<mode>"): New insn pattern.
6054 ("vec_shr_<mode>"): New expander.
6055 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6056 and force the shift count operand to V16QImode.
6057 ("vec_srb<mode>"): Set shift count mode to V16QI.
6059 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6061 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6062 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6063 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6065 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6068 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6070 2017-09-26 Richard Biener <rguenther@suse.de>
6072 PR tree-optimization/82320
6073 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6076 2017-09-25 Jeff Law <law@redhat.com>
6078 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6079 prototype for new argument.
6080 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6081 mostly extracted from rs6000_emit_allocate_stack.
6082 (rs6000_emit_probe_stack_range_stack_clash): New function.
6083 (rs6000_emit_allocate_stack): Call
6084 rs6000_emit_probe_stack_range_stack_clash as needed.
6085 (rs6000_emit_probe_stack_range): Add additional argument
6086 to call to gen_probe_stack_range{si,di}.
6087 (output_probe_stack_range): New.
6088 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6089 (output_probe_stack_range_stack_clash): New.
6090 (rs6000_emit_prologue): Emit notes into dump file as requested.
6091 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6092 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6093 Add additional operand and pass it to output_probe_stack_range.
6095 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6097 PR tree-optimization/82163
6098 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6099 (checking_verify_loop_closed_ssa): New parameter.
6100 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6101 (check_loop_closed_ssa_stmt): Delete.
6102 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6103 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6104 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6107 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6109 * brig-builtins.def: Treat HSAIL barrier builtins as
6110 setjmp/longjump style functions.
6112 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6114 * target.def (constant_alignment): New hook.
6115 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6116 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6117 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6118 * doc/tm.texi: Regenerate.
6119 * targhooks.h (default_constant_alignment): Declare.
6120 (constant_alignment_word_strings): Likewise.
6121 * targhooks.c (default_constant_alignment): New function.
6122 (constant_alignment_word_strings): Likewise.
6123 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6124 instead of CONSTANT_ALIGNMENT.
6125 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6126 (force_const_mem): Likewise.
6127 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6128 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6129 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6130 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6131 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6133 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6134 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6135 constant_alignment_word_strings.
6136 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6137 (CONSTANT_ALIGNMENT): Likewise.
6138 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6139 (arm_constant_alignment): New function.
6140 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6141 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6142 constant_alignment_word_strings.
6143 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6144 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6145 constant_alignment_word_strings.
6146 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6147 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6148 (cris_constant_alignment): New function.
6149 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6150 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6151 (epiphany_constant_alignment): New function.
6152 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6153 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6154 constant_alignment_word_strings.
6155 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6156 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6157 constant_alignment_word_strings.
6158 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6159 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6160 constant_alignment_word_strings.
6161 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6162 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6163 * config/i386/i386.c (ix86_constant_alignment): Make static.
6164 Use the same interface as the target hook.
6165 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6166 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6167 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6168 constant_alignment_word_strings.
6169 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6170 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6171 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6172 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
6173 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6174 constant_alignment_word_strings.
6175 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
6176 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6177 constant_alignment_word_strings.
6178 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
6179 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6180 constant_alignment_word_strings.
6181 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
6182 * config/microblaze/microblaze.c (microblaze_constant_alignment):
6184 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6185 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
6186 * config/mips/mips.c (mips_constant_alignment): New function.
6187 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6188 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
6189 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
6190 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6191 (mmix_constant_alignment): Make static. Use the same interface
6193 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
6194 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6195 constant_alignment_word_strings.
6196 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
6197 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6198 constant_alignment_word_strings.
6199 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
6200 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6201 constant_alignment_word_strings.
6202 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
6203 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6204 (rs6000_constant_alignment): New function.
6205 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
6206 * config/riscv/riscv.c (riscv_constant_alignment): New function.
6207 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6208 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
6209 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6210 (rs6000_constant_alignment): New function.
6211 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
6212 * config/s390/s390.c (s390_constant_alignment): New function.
6213 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6214 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
6215 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6216 constant_alignment_word_strings.
6217 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
6218 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6219 (sparc_constant_alignment): New function.
6220 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
6221 * config/spu/spu.c (spu_constant_alignment): New function.
6222 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6223 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
6224 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6225 constant_alignment_word_strings.
6226 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
6227 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6228 constant_alignment_word_strings.
6229 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
6230 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6231 constant_alignment_word_strings.
6232 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
6233 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6234 (visium_constant_alignment): New function.
6235 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6236 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6237 (xtensa_constant_alignment): New function.
6238 * system.h (CONSTANT_ALIGNMENT): Poison.
6240 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6242 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6243 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6244 (rs6000_builtin_valid_without_lhs): New helper function.
6245 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6246 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6248 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6250 * target.h (vec_perm_indices): Use unsigned short rather than
6252 (auto_vec_perm_indices): Likewise.
6253 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6254 Use unsigned int rather than unsigned char.
6255 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6257 2017-09-25 Richard Biener <rguenther@suse.de>
6259 * cfgloop.h (sort_sibling_loops): Declare.
6260 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6261 (sort_sibling_loops): New function sorting the sibling loop list
6263 * graphite.c (graphite_transform_loops): Sort sibling loops.
6265 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6267 * target.def (vec_perm_const_ok): Change sel parameter to
6269 * optabs-query.c (can_vec_perm_p): Update accordingly.
6270 * doc/tm.texi: Regenerate.
6271 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6272 auto_vec_perm_indices and remove separate nelt field.
6273 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6274 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6275 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6276 (aarch64_expand_vec_perm_const): Update accordingly.
6277 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6278 to vec_perm_indices.
6279 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6280 auto_vec_perm_indices and remove separate nelt field.
6281 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6282 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6283 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6285 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6286 to vec_perm_indices.
6287 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6288 sel to vec_perm_indices.
6289 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6290 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6291 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6293 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6296 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6299 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6300 on the FUNCTION_DECL function context if it has a DIE that is a
6303 2017-09-25 Richard Biener <rguenther@suse.de>
6305 PR tree-optimization/82285
6306 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6309 2017-09-25 Tom de Vries <tom@codesourcery.com>
6313 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6316 2017-09-25 Richard Biener <rguenther@suse.de>
6318 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6319 ISL errors other than isl_error_quota happen. Dump if the
6320 schedule is the same.
6321 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6322 errors instead of aborting inside ISL.
6324 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6327 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6328 of libgcc_eh for m64.
6329 * config/i386/darwin64.h: Likewise.
6331 2017-09-25 Richard Biener <rguenther@suse.de>
6334 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6335 attribute for incomplete types nor twice for complete ones.
6337 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6340 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6341 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6344 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6347 * config/i386/i386.opt (mprefer-avx256): Use
6348 ix86_target_flags variable.
6349 * config/i386/i386.c (ix86_target_string): Move
6350 -mprefer-avx256 to flag2_opts.
6352 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6355 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6356 and x != -1 | y != -1 into (x & y) != -1.
6358 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6360 * config.gcc: Add new case statement to set
6361 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6362 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6363 s390x-*-linux* case statements. Added aarch64 to the list of
6364 supported architectures.
6366 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6368 PR tree-optimization/82289
6369 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6370 STMT_VINFO_RELEVANT_P.
6372 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6373 Alan Hayward <alan.hayward@arm.com>
6374 David Sherwood <david.sherwood@arm.com>
6376 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6377 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6378 (extract_range_from_binary_expr_1): Don't call
6379 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6381 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6382 Alan Hayward <alan.hayward@arm.com>
6383 David Sherwood <david.sherwood@arm.com>
6385 * target.def (preferred_vector_alignment): New hook.
6386 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6388 * doc/tm.texi: Regenerate.
6389 * targhooks.h (default_preferred_vector_alignment): Declare.
6390 * targhooks.c (default_preferred_vector_alignment): New function.
6391 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6393 (DR_TARGET_ALIGNMENT): New macro.
6394 (aligned_access_p): Update commentary.
6395 (vect_known_alignment_in_bytes): New function.
6396 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6398 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6399 Calculate the misalignment based on the target alignment rather than
6401 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6402 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6403 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6404 the target alignment, rather than masking the element misalignment
6405 with the number of elements in a vector. Also use the target
6406 alignment when calculating the maximum number of peels.
6407 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6408 instead of TYPE_ALIGN_UNIT.
6409 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6410 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6411 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6412 (vect_create_data_ref_ptr): Likewise.
6413 (vect_setup_realignment): Realign by ANDing with
6414 -DR_TARGET_MISALIGNMENT.
6415 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6416 the number of peels based on DR_TARGET_ALIGNMENT.
6417 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6418 with the guaranteed alignment boundary when deciding whether
6420 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6421 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6422 (ensure_base_align): Remove stmt_info parameter. Get the
6423 target base alignment from DR_TARGET_ALIGNMENT.
6424 (vectorizable_store): Update call accordingly. Interpret
6425 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6427 (vectorizable_load): Likewise.
6429 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6430 Alan Hayward <alan.hayward@arm.com>
6431 David Sherwood <david.sherwood@arm.com>
6433 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6434 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6435 (vect_enhance_data_refs_alignment): Likewise.
6437 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6439 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6440 error. Only quit immediately if parsing is complete.
6441 (BEGIN): Initialize fatal_err and parse_done.
6442 (begin fpu, end fpu): Check number of arguments.
6443 (begin arch, end arch): Likewise.
6444 (begin cpu, end cpu): Likewise.
6445 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6446 (optalias): Likewise.
6448 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6450 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6451 * config/arm/arm-isa.h: Delete. Move definitions to ...
6452 * arm-cpus.in: ... here. Use new feature and fgroup values.
6453 * config/arm/arm.c (arm_option_override): Use lower case for feature
6455 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6456 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6457 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6459 (print_isa_bits_for): New function.
6460 (gen_isa): New function.
6461 (gen_comm_data): Use print_isa_bits_for.
6462 (define feature): New keyword.
6463 (define fgroup): New keyword.
6464 * config/arm/t-arm (TM_H): Remove.
6465 (GTM_H): Add arm-isa.h.
6466 (arm-isa.h): Add rule to generate file.
6467 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6468 case for feature bit names.
6470 2017-09-22 Richard Biener <rguenther@suse.de>
6472 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6474 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6475 print of no dependency loops ...
6476 * graphite.c (graphite_transform_loops): ... here.
6477 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6479 (same_close_phi_node, remove_duplicate_close_phi,
6480 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6481 (canonicalize_loop_closed_ssa): ... here and simplify.
6482 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6483 (optimize_isl): Use dump_printf_loc to tell when we stopped
6484 optimizing because of an ISL timeout.
6486 2017-09-22 Richard Biener <rguenther@suse.de>
6488 PR tree-optimization/82291
6489 * tree-if-conv.c (predicate_mem_writes): Make sure to
6490 remove writes in blocks predicated with false.
6492 2017-09-22 Richard Biener <rguenther@suse.de>
6494 * sese.c: Include cfganal.h.
6495 (if_region_set_false_region): Remove.
6496 (create_if_region_on_edge): Likewise.
6497 (move_sese_in_condition): Re-implement without destroying
6500 2017-09-22 Richard Biener <rguenther@suse.de>
6502 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6503 Verify both BBs contain loop PHI nodes before dispatching to
6505 (graphite_regenerate_ast_isl): Do not recompute dominators,
6506 do not verify three times. Restructure for clarity.
6507 * graphite-scop-detection.c (same_close_phi_node,
6508 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6509 defined_in_loop_p, canonicalize_loop_closed_ssa,
6510 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6511 checking and SSA rewrite, move to ...
6512 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6513 (graphite_initialize): Do not pass in ctx, do not reset the
6514 SCEV cache, compute only dominators.
6515 (graphite_transform_loops): Allocate ISL ctx after
6516 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6517 Maintain post-dominators only around build_scops.
6518 * sese.c (if_region_set_false_region): Make static. Free
6519 and recompute dominators.
6520 (move_sese_in_condition): Assert we don't get called with
6521 post-dominators computed.
6522 * sese.h (if_region_set_false_region): Remove.
6524 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6526 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6527 mode attribute for TARGET_AVX512VL.
6529 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6531 * config/i386/i386.opt (mprefer-avx256): New option.
6532 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6534 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6535 for TARGET_PREFER_AVX256.
6536 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6538 2017-09-21 Jeff Law <law@redhat.com>
6540 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6541 Fix dump output if the only stack space is for pushed registers.
6543 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6545 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6548 2017-09-21 Martin Sebor <msebor@redhat.com>
6551 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6552 code (in C++) or code that triggers warnings.
6554 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6556 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6558 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6560 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6561 * sched-rgn.c: Ditto.
6562 * sel-sched-ir.c: Ditto.
6564 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6566 * toplev.h (set_random_seed): Adjust return type.
6567 * toplev.c (init_local_tick): Move eager initialization of random_seed
6568 to get_random_seed. Adjust comment.
6569 (init_random_seed): Inline to get_random_seed, delete.
6570 (get_random_seed): Initialize random_seed lazily.
6571 (set_random_seed): Do not return previous value.
6572 (print_switch_value): Do not call get_random_seed.
6574 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6576 * cgraph.c (delete_function_version): New, broken out from...
6577 (cgraph_node::delete_function_version): ...here. Rename to
6578 cgraph_node::delete_function_version_by_decl. Update all uses.
6579 (cgraph_node::remove): Call delete_function_version.
6581 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6584 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6585 VAR_DECLs to which addressable non-volatile parameters are mapped
6586 and for id->retvar after the return value assignment. Clear
6587 id->retval and id->retbnd after inlining.
6589 2017-09-21 Richard Biener <rguenther@suse.de>
6591 PR tree-optimization/82276
6592 PR tree-optimization/82244
6593 * tree-vrp.c (build_assert_expr_for): Set
6594 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6596 (remove_range_assertions): Revert earlier change.
6598 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6601 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6603 2017-09-21 Richard Biener <rguenther@suse.de>
6605 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6606 Restore valid IL after code generation errors.
6607 * graphite.c (graphite_transform_loops): Diagnose code
6608 generation issues as MSG_MISSED_OPTIMIZATION and continue
6609 with processing SCOPs.
6611 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6612 Alan Hayward <alan.hayward@arm.com>
6613 David Sherwood <david.sherwood@arm.com>
6615 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6616 rather than choosing between plus_constant and gen_rtx_<CODE>.
6617 * expr.c (emit_push_insn): Likewise.
6618 (expand_expr_real_2): Likewise.
6620 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6621 Alan Hayward <alan.hayward@arm.com>
6622 David Sherwood <david.sherwood@arm.com>
6624 * loop-unroll.c (split_iv): Call copy_rtx on the step.
6626 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6627 Alan Hayward <alan.hayward@arm.com>
6628 David Sherwood <david.sherwood@arm.com>
6630 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6631 calling tree_to_uhwi.
6633 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6634 Alan Hayward <alan.hayward@arm.com>
6635 David Sherwood <david.sherwood@arm.com>
6637 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6638 INTEGER_CST rather than a negative test for ADDR_EXPR.
6640 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6641 Alan Hayward <alan.hayward@arm.com>
6642 David Sherwood <david.sherwood@arm.com>
6644 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6645 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6647 2017-09-21 Richard Biener <rguenther@suse.de>
6649 PR tree-optimization/71351
6650 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6651 graphite_create_new_loop_guard): Remove, fold remaining parts
6653 (translate_isl_ast_node_for): ... here and simplify.
6655 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6658 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6659 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6660 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6661 alternative always use QI mode, for -Os imov (=R,R) alternative
6662 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6665 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6666 Jeff Law <law@redhat.com>
6668 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6669 (allocate_stack_space): New function, partially extracted from
6671 (s390_emit_prologue): Track offset to most recent stack probe.
6672 Code to allocate space moved into allocate_stack_space.
6673 Dump actions when no stack is allocated.
6674 (s390_prologue_plus_offset): New function.
6675 (s390_emit_stack_probe): Likewise.
6677 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
6679 * common.opt (Wa, Wl, Wp, g, gz=): Add
6681 (gno-column-info): Remove.
6682 (gcolumn-info): Drop RejectNegative.
6684 (gno-record-gcc-switches): Remove.
6685 (grecord-gcc-switches): Drop RejectNegative.
6686 (gno-split-dwarf): Remove.
6687 (gsplit-dwarf): Drop RejectNegative.
6688 (gno-strict-dwarf): Remove.
6689 (gstrict-dwarf): Drop RejectNegative.
6690 * config/darwin.opt (gfull, gused): Add RejectNegative.
6691 * dwarf2out.c (gen_producer_string): Drop
6692 gno-record-gcc-switches handler.
6693 * optc-gen.awk: Add g to prefixes with negative forms.
6694 * opts-common.c (remapping_prefix_p): New.
6695 (find_opt): Check it.
6696 (generate_canonical_option): Test g prefix.
6697 (option_map): Add -gno- mapping.
6698 (add_misspelling_candidates): Check remapping_prefix_p.
6700 2017-09-20 Jeff Law <law@redhat.com>
6702 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6703 thinko in stack clash protection support.
6705 * explow.c (compute_stack_clash_protection_loop_data): Use
6706 CONST_INT_P instead of explicit test. Verify object is a
6707 CONST_INT_P before looking at INTVAL.
6708 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6709 instead of explicit test.
6711 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6714 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6715 address instead of to r1 and r11.
6717 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6719 * config.gcc: Support "knm".
6720 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6721 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6723 * config/i386/i386.c (m_KNM): Define.
6724 (processor_target_table): Add "knm".
6726 (ix86_option_override_internal): Add "knm".
6727 (ix86_issue_rate): Add PROCESSOR_KNM.
6728 (ix86_adjust_cost): Ditto.
6729 (ia32_multipass_dfa_lookahead): Ditto.
6730 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6731 (fold_builtin_cpu): Add M_INTEL_KNM.
6732 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6733 (processor_type): Add PROCESSOR_KNM.
6734 * config/i386/x86-tune.def: Add m_KNM.
6735 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6737 2017-09-20 Richard Biener <rguenther@suse.de>
6739 PR tree-optimization/80213
6740 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6741 are allowed in empty BBs as well.
6742 (canonicalize_loop_closed_ssa): Also look for other complex
6744 (scop_detection::get_sese): Include the loop-closed PHI block
6746 (scop_detection::merge_sese): Remove code adding extra blocks.
6747 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6748 (build_scops): Assert the final returned scop is invalid.
6750 2017-09-20 Richard Biener <rguenther@suse.de>
6752 PR tree-optimization/82264
6753 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6755 (vn_phi_lookup): Likewise.
6756 (vn_phi_insert): Likewise.
6758 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6760 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6761 that fits into uhwi or shwi, add DW_AT_const_value regardless
6762 of early_dwarf without going through RTL, using add_AT_unsigned
6765 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6766 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6767 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6768 (DEBUG_MACRO_SECTION): Likewise.
6769 (DEBUG_LTO_MACRO_SECTION): Likewise.
6770 (DEBUG_STR_DWO_SECTION): Likewise.
6771 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6772 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6773 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6774 (DEBUG_LTO_LINE_STR_SECTION): Define.
6775 (init_sections_and_labels): Initialize debug_line_str_section
6776 variable. Initialize debug_loc_section for -gdwarf-5 to
6777 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6779 2017-09-20 Richard Biener <rguenther@suse.de>
6781 * graphite-sese-to-poly.c (extract_affine): Properly handle
6782 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6784 2017-09-20 Richard Biener <rguenther@suse.de>
6786 PR tree-optimization/81373
6787 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6788 Force SESE live-out defs to be handled even if they are
6791 2017-09-19 Jeff Law <law@redhat.com>
6793 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6794 nothing for stack adjustments with REG_STACK_CHECK.
6795 * sched-deps.c (parse_add_or_inc): Reject insns with
6796 REG_STACK_CHECK from dependency breaking.
6797 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6798 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6799 * reg-notes.def (STACK_CHECK): New note.
6801 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6802 (ix86_expand_prologue): Dump stack clash info as needed.
6803 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6805 * function.c (dump_stack_clash_frame_info): New function.
6806 * function.h (dump_stack_clash_frame_info): Prototype.
6807 (enum stack_clash_probes): New enum.
6809 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6810 flag_stack_clash_protection.
6811 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6812 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6813 (arm_frame_pointer_required): Likewise.
6814 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6815 (ia64_expand_prologue): Likewise.
6816 * config/mips/mips.c (mips_expand_prologue): Likewise.
6817 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6818 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6819 (sparc_flat_expand_prologue): Likewise.
6820 * config/spu/spu.c (spu_expand_prologue): Likewise.
6822 * explow.c: Include "params.h".
6823 (anti_adjust_stack_and_probe_stack_clash): New function.
6824 (get_stack_check_protect): Likewise.
6825 (compute_stack_clash_protection_loop_data): Likewise.
6826 (emit_stack_clash_protection_loop_start): Likewise.
6827 (emit_stack_clash_protection_loop_end): Likewise.
6828 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6829 Use anti_adjust_stack_and_probe_stack_clash.
6830 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6831 (emit_stack_clash_protection_loop_start): Likewise.
6832 (emit_stack_clash_protection_loop_end): Likewise.
6833 * rtl.h (get_stack_check_protect): Prototype.
6834 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6835 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6836 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6838 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6840 * doc/tm.texi: Rebuilt.
6841 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6842 get_stack_check_protect.
6843 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6844 * config/arm/arm.c (arm_expand_prologue): Likewise.
6845 (arm_frame_pointer_required): Likewise.
6846 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6847 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6848 * config/mips/mips.c (mips_expand_prologue): Likewise.
6849 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6850 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6851 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6852 (sparc_flat_expand_prologue): Likewise.
6854 * common.opt (-fstack-clash-protection): New option.
6855 * flag-types.h (enum stack_check_type): Note difference between
6856 -fstack-check= and -fstack-clash-protection.
6857 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6858 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6859 * toplev.c (process_options): Issue warnings/errors for cases
6860 not handled with -fstack-clash-protection.
6861 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6862 (-fstack-check): Note additional problem with -fstack-check=generic.
6863 Note that -fstack-check is primarily for Ada and refer users
6864 to -fstack-clash-protection for stack-clash-protection.
6865 Document new params for stack clash protection.
6867 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6869 * config/i386/i386.c (ix86_split_long_move): Do not handle
6870 address used for LEA in a special way.
6872 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6874 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6876 2017-09-19 Martin Sebor <msebor@redhat.com>
6879 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6880 of incompatible types.
6882 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6884 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6885 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6886 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6887 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6889 2017-09-19 Richard Biener <rguenther@suse.de>
6891 PR tree-optimization/82244
6892 * tree-vrp.c (remove_range_assertions): Do not propagate
6893 a constant to abnormals but replace the assert with a copy.
6895 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6897 PR rtl-optimization/57878
6898 PR rtl-optimization/68988
6899 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6900 avoidance test involving non_reload_pseudos. Move frequency test
6901 below the general fragmentation avoidance test.
6903 2017-09-19 Richard Biener <rguenther@suse.de>
6905 PR tree-optimization/69728
6906 * graphite-sese-to-poly.c (schedule_error): New global.
6907 (add_loop_schedule): Handle empty domain by failing the
6909 (build_original_schedule): Handle schedule_error.
6911 2017-09-19 Richard Biener <rguenther@suse.de>
6913 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6914 Do not iterate to sibling loops but only to siblings of inner
6917 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6920 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6921 that the registers are different.
6923 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6925 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6926 to processor_model and "amdfam17h" to arch_names_table.
6927 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6929 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6932 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6934 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6935 Alan Hayward <alan.hayward@arm.com>
6936 David Sherwood <david.sherwood@arm.com>
6938 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6940 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6942 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6943 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6944 vect_schedule_slp_instance.
6945 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6946 Update call to vect_slp_analyze_node_operations. Simplify return
6948 (vect_slp_analyze_bb_1): Update call accordingly.
6949 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6950 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6951 (vect_schedule_slp): Update call accordingly.
6953 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6954 Alan Hayward <alan.hayward@arm.com>
6955 David Sherwood <david.sherwood@arm.com>
6957 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6958 with types that aren't in fact scalar.
6960 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6962 * tree-vect-slp.c (vect_record_max_nunits): New function,
6964 (vect_build_slp_tree_1): ...here.
6965 (vect_build_slp_tree_2): Call it for phis too.
6967 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6969 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6970 to vect_get_vec_def_for_operand when getting the mask operand.
6972 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6973 Alan Hayward <alan.hayward@arm.com>
6974 David Sherwood <david.sherwood@arm.com>
6976 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
6979 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6980 Alan Hayward <alan.hayward@arm.com>
6981 David Sherwood <david.sherwood@arm.com>
6983 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
6984 calculation for vector booleans.
6986 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6987 Alan Hayward <alan.hayward@arm.com>
6988 David Sherwood <david.sherwood@arm.com>
6990 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
6992 (vect_transform_stmt): ...here.
6993 (vect_analyze_stmt): Use it instead of calling
6994 vectorizable_live_operation directly.
6996 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
6998 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
6999 non-SIMT targets in acc vector loops.
7001 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7003 * configure.ac: Add arc and check if assembler supports gdwarf2.
7004 * configure: Regenerate.
7006 2017-09-18 Richard Biener <rguenther@suse.de>
7008 PR tree-optimization/82220
7009 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7010 epilogue niters from the min_profitable_iters compute.
7012 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7015 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7016 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7017 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7019 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7022 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7023 switching to a new text section.
7025 2017-09-18 Richard Biener <rguenther@suse.de>
7027 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7029 (build_alias_set): Reject aliases with no access function.
7031 2017-09-18 Richard Biener <rguenther@suse.de>
7033 PR tree-optimization/79622
7034 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7036 (build_cross_bb_scalars_use): Likewise.
7038 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7040 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7042 2017-09-18 Alan Modra <amodra@gmail.com>
7045 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7046 stack_pointer_rtx for count 0. Update comments. Break up
7047 large rtl expression.
7049 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7051 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7052 Increase to 20 bytes.
7053 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7054 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7055 or avx version of the stub.
7057 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7060 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7061 compute the minimum stack alignment. Also update preferred stack
7062 boundary for leaf functions.
7064 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7066 PR tree-optimization/82228
7067 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7070 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7072 * common/config/nds32/nds32-common.c
7073 (nds32_option_optimization_table): Refine formatting.
7074 (nds32_option_optimization_table): Use -fsched-pressure and
7075 -fomit-frame-pointer for specific optimization level.
7077 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7079 * config/nds32/nds32.c: Refine formatting and comments.
7080 * config/nds32/nds32.h: Likewise.
7081 * config/nds32/nds32.md: Likewise.
7082 * config/nds32/nds32-cost.c: Likewise.
7083 * config/nds32/nds32-isr.c: Likewise.
7084 * config/nds32/nds32-md-auxiliary.c: Likewise.
7085 * config/nds32/nds32-multiple.md: Likewise.
7086 * config/nds32/nds32-predicates.c: Likewise.
7088 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7089 Jakub Jelinek <jakub@redhat.com>
7091 Add support for -std=c++2a.
7092 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7094 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7096 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7099 * doc/extend.texi (Common Function Attributes): Add
7100 references to ARM, AArch64, and S/390 specific attributes.
7101 (Function Specific Option Pragmas): Add AArch64 and S/390
7102 to list of back ends that support the target pragma.
7104 2017-09-15 Nathan Sidwell <nathan@acm.org>
7106 * doc/standards.texi: Fix C++17 description. Update URLs for
7109 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7111 * common.opt (Wcast-align=strict): New warning option.
7112 * doc/invoke.texi: Document -Wcast-align=strict.
7114 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7116 * cgraph.h (cgraph_thunk_info): Add comments.
7117 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7118 assert for VIRTUAL_* arguments stricter.
7120 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7122 PR tree-optimization/71026
7123 * match.pd: Move RDIV patterns from fold-const.c
7124 * fold-const.c (distribute_real_division): Removed.
7125 (fold_binary_loc): Remove calls to distribute_real_divison.
7127 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7129 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7130 c++1z and gnu++1z as deprecated. Change other references to
7131 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7132 Change -Wc++1z-compat to -Wc++17-compat.
7133 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7134 * dwarf2out.c (highest_c_language): Handle C++17.
7135 (gen_compile_unit_die): Likewise.
7137 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7139 PR rtl-optimization/82192
7140 * combine.c (make_extraction): Don't look through non-paradoxical
7141 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7144 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7145 Alan Hayward <alan.hayward@arm.com>
7146 David Sherwood <david.sherwood@arm.com>
7148 * target.def (function_arg_offset): New hook.
7149 * targhooks.h (default_function_arg_offset): Declare.
7150 * targhooks.c (default_function_arg_offset): New function.
7151 * function.c (locate_and_pad_parm): Use
7152 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7153 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7154 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7155 * doc/tm.texi: Regenerate.
7156 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7157 * config/spu/spu.c (spu_function_arg_offset): New function.
7158 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7159 * system.h (FUNCTION_ARG_OFFSET): Poison.
7161 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7162 Alan Hayard <alan.hayward@arm.com>
7163 David Sherwood <david.sherwood@arm.com>
7165 * target.def (truly_noop_truncation): New hook.
7166 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7167 than TRULY_NOOP_TRUNCATION.
7168 * hooks.h (hook_bool_uint_uint_true): Declare.
7169 * hooks.c (hook_bool_uint_uint_true): New function.
7170 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7171 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7172 * doc/tm.texi: Regenerate.
7173 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
7174 rather than TRULY_NOOP_TRUNCATION in comments.
7175 (simplify_comparison): Likewise.
7176 (record_truncated_value): Likewise.
7177 * expmed.c (extract_bit_field_1): Likewise.
7178 (extract_split_bit_field): Likewise.
7179 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
7180 instead of TRULY_NOOP_TRUNCATION.
7181 * function.c (assign_parm_setup_block): Likewise.
7182 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
7183 * rtlhooks.c: Include target.h.
7184 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
7185 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
7186 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
7187 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
7188 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
7189 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
7190 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
7191 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
7192 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
7193 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
7194 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
7195 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
7196 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
7197 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
7198 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
7199 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
7200 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
7201 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
7202 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
7203 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
7204 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
7205 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
7206 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
7207 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
7208 * config/mips/mips.c (mips_truly_noop_truncation): New function.
7209 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7210 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7211 rather than TRULY_NOOP_TRUNCATION in comments.
7212 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
7213 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
7214 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
7215 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
7216 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
7217 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
7218 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
7219 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
7220 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
7221 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
7222 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
7223 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7224 rather than TRULY_NOOP_TRUNCATION in comments.
7225 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
7226 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
7227 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
7228 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
7229 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
7230 TRULY_NOOP_TRUNCATION condition.
7231 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
7232 (TRULY_NOOP_TRUNCATION): Delete.
7233 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
7234 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
7235 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7236 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7237 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7238 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7239 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7240 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7241 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7242 rather than TRULY_NOOP_TRUNCATION in comments.
7243 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7244 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7245 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7246 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7247 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7248 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7250 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7253 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7254 (*cmp_ior): Likewise.
7255 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7256 (*ior_scc_scc_cmp): Likewise.
7257 (*and_scc_scc): Likewise.
7258 (*and_scc_scc_cmp): Likewise.
7260 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7261 Alan Hayard <alan.hayward@arm.com>
7262 David Sherwood <david.sherwood@arm.com>
7264 * target.def (can_change_mode_class): New hook.
7265 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7266 (hard_regno_nregs): Likewise.
7267 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7268 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7269 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7270 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7271 (LOAD_EXTEND_OP): Update accordingly.
7272 * doc/tm.texi: Regenerate.
7273 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7274 CANNOT_CHANGE_MODE_CLASS.
7275 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7276 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7277 * combine.c (simplify_set): Update accordingly.
7278 * emit-rtl.c (validate_subreg): Likewise.
7279 * recog.c (general_operand): Likewise.
7280 * regcprop.c (mode_change_ok): Likewise.
7281 * reload1.c (choose_reload_regs): Likewise.
7282 (inherit_piecemeal_p): Likewise.
7283 * rtlanal.c (simplify_subreg_regno): Likewise.
7284 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7285 instead of CANNOT_CHANGE_MODE_CLASS.
7286 (reload_cse_simplify_operands): Likewise.
7287 * reload.c (push_reload): Use targetm.can_change_mode_class
7288 instead of CANNOT_CHANGE_MODE_CLASS.
7289 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7290 REG_CANNOT_CHANGE_MODE_P.
7291 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7292 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7293 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7294 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7295 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7296 (arm_can_change_mode_class): New function.
7297 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7298 than CANNOT_CHANGE_MODE_CLASS in comments.
7299 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7300 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7301 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7302 (ix86_can_change_mode_class): ...this new function, inverting the
7303 sense of the return value.
7304 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7305 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7306 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7307 (ia64_can_change_mode_class): New function.
7308 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7309 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7310 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7311 (m32c_can_change_mode_class): ...this new function, inverting the
7312 sense of the return value.
7313 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7314 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7315 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7316 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7317 (mips_can_change_mode_class): ...this new function, inverting the
7318 sense of the return value.
7319 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7320 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7321 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7322 (msp430_can_change_mode_class): New function.
7323 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7324 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7325 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7326 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7327 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7328 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7329 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7330 (pa_cannot_change_mode_class): Replace with...
7331 (pa_can_change_mode_class): ...this new function, inverting the
7332 sense of the return value.
7333 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7334 than CANNOT_CHANGE_MODE_CLASS in comments.
7335 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7336 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7337 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7338 (pdp11_cannot_change_mode_class): Replace with...
7339 (pdp11_can_change_mode_class): ...this new function, inverting the
7340 sense of the return value.
7341 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7342 * config/powerpcspe/powerpcspe-protos.h
7343 (rs6000_cannot_change_mode_class_ptr): Delete.
7344 * config/powerpcspe/powerpcspe.c
7345 (rs6000_cannot_change_mode_class_ptr): Delete.
7346 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7347 (rs6000_option_override_internal): Assign to
7348 targetm.can_change_mode_class instead of
7349 rs6000_cannot_change_mode_class_ptr.
7350 (rs6000_cannot_change_mode_class): Replace with...
7351 (rs6000_can_change_mode_class): ...this new function, inverting the
7352 sense of the return value.
7353 (rs6000_debug_cannot_change_mode_class): Replace with...
7354 (rs6000_debug_can_change_mode_class): ...this new function.
7355 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7356 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7357 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7358 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7359 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7361 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7362 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7363 (rs6000_option_override_internal): Assign to
7364 targetm.can_change_mode_class instead of
7365 rs6000_cannot_change_mode_class_ptr.
7366 (rs6000_cannot_change_mode_class): Replace with...
7367 (rs6000_can_change_mode_class): ...this new function, inverting the
7368 sense of the return value.
7369 (rs6000_debug_cannot_change_mode_class): Replace with...
7370 (rs6000_debug_can_change_mode_class): ...this new function.
7371 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7372 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7373 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7374 (s390_can_change_mode_class): ...this new function, inverting the
7375 sense of the return value.
7376 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7377 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7378 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7379 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7380 (sh_cannot_change_mode_class): Replace with...
7381 (sh_can_change_mode_class): ...this new function, inverting the
7382 sense of the return value.
7383 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7384 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7385 (sparc_can_change_mode_class): New function.
7386 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7387 * config/spu/spu.c (spu_can_change_mode_class): New function.
7388 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7389 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7390 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7391 (visium_can_change_mode_class): New function.
7392 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7394 2017-09-15 Richard Biener <rguenther@suse.de>
7396 PR tree-optimization/82217
7397 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7398 but not undefined case.
7400 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7403 * postreload.c (reload_cse_simplify_operands): Skip
7404 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7406 2017-09-15 Richard Biener <rguenther@suse.de>
7408 PR tree-optimization/68823
7409 * graphite-scop-detection.c (build_alias_set): If we have a
7410 possible dependence check whether we can handle them by just
7411 looking at the DRs DR_ACCESS_FNs.
7412 (build_scops): If build_alias_set fails, fail the SCOP.
7414 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7416 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7417 to support float128 built-in functions that require the ISA 3.0
7419 (BU_FLOAT128_3_HW): Likewise.
7420 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7422 (FMAF128): Likewise.
7424 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7425 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7426 floating point instructions.
7427 (rs6000_invalid_builtin): Likewise.
7428 (rs6000_builtin_mask_names): Likewise.
7429 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7430 (RS6000_BTM_FLOAT128_HW): Likewise.
7431 (RS6000_BTM_COMMON): Likewise.
7432 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7434 * doc/extend.texi (RS/6000 built-in functions): Document the
7435 IEEE 128-bit floating point square root and fused multiply-add
7438 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7440 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7441 reg (r2) isn't in the set of registers defined in the prologue.
7443 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7444 Alan Hayward <alan.hayward@arm.com>
7445 David Sherwood <david.sherwood@arm.com>
7447 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7448 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7449 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7450 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7451 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7453 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7454 max_vectorization_factor.
7455 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7457 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7458 Alan Hayward <alan.hayward@arm.com>
7459 David Sherwood <david.sherwood@arm.com>
7461 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7462 (vect_worthwhile_without_simd_p): Declare.
7463 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7464 (vectorizable_reduction): Use it.
7465 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7466 (vectorizable_operation): Likewise.
7468 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7469 Alan Hayward <alan.hayward@arm.com>
7470 David Sherwood <david.sherwood@arm.com>
7472 * tree-vectorizer.h (vect_get_num_copies): New function.
7473 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7474 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7475 (vectorizable_induction): Likewise.
7476 (vectorizable_live_operation): Likewise.
7477 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7478 (vectorizable_bswap): Likewise.
7479 (vectorizable_call): Likewise.
7480 (vectorizable_conversion): Likewise.
7481 (vectorizable_assignment): Likewise.
7482 (vectorizable_shift): Likewise.
7483 (vectorizable_operation): Likewise.
7484 (vectorizable_store): Likewise.
7485 (vectorizable_load): Likewise.
7486 (vectorizable_condition): Likewise.
7487 (vectorizable_comparison): Likewise.
7488 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7490 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7491 Alan Hayward <alan.hayward@arm.com>
7492 David Sherwood <david.sherwood@arm.com>
7494 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7495 of vect_init_vector.
7497 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7498 Alan Hayward <alan.hayward@arm.com>
7499 David Sherwood <david.sherwood@arm.com>
7501 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7502 an inline wrapper that provides a location.
7503 (gimple_build_vector): Likewise.
7504 * gimple-fold.c (gimple_build_vector_from_val): New function.
7505 (gimple_build_vector): Likewise.
7506 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7507 functions to build the initial value. Always return a gimple value.
7508 (get_initial_defs_for_reduction): Likewise. Only compute
7510 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7511 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7512 (vectorizable_induction): Use gimple_build_vector rather than
7515 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7516 Alan Hayward <alan.hayward@arm.com>
7517 David Sherwood <david.sherwood@arm.com>
7519 * target.h (vec_perm_indices): New typedef.
7520 (auto_vec_perm_indices): Likewise.
7521 * optabs-query.h: Include target.h
7522 (can_vec_perm_p): Take a vec_perm_indices *.
7523 * optabs-query.c (can_vec_perm_p): Likewise.
7524 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7525 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7526 * tree-vect-generic.c (lower_vec_perm): Likewise.
7527 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7528 (vect_grouped_load_supported): Likewise.
7529 (vect_shift_permute_load_chain): Likewise.
7530 (vect_permute_store_chain): Use auto_vec_perm_indices.
7531 (vect_permute_load_chain): Likewise.
7532 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7533 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7534 Update uses of can_vec_perm_p.
7535 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7536 mode with a number of elements. Take a vec_perm_indices *.
7537 (vect_create_epilog_for_reduction): Update accordingly.
7538 Use auto_vec_perm_indices.
7539 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7540 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7541 (vect_transform_slp_perm_load): Likewise.
7542 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7543 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7544 (vect_gen_perm_mask_checked): Likewise.
7545 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7546 (vect_gen_perm_mask_checked): Likewise.
7547 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7548 (vectorizable_store): Likewise.
7549 (vectorizable_load): Likewise.
7550 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7551 (vectorizable_bswap): Likewise.
7553 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7554 Alan Hayward <alan.hayward@arm.com>
7555 David Sherwood <david.sherwood@arm.com>
7557 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7558 * tree.c (build_vector): Likewise.
7559 (build_vector_from_ctor): Update accordingly.
7560 (build_vector_from_val): Likewise.
7561 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7562 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7563 * tree-vect-generic.c (add_rshift): Likewise.
7564 (expand_vector_divmod): Likewise.
7565 (optimize_vector_constructor): Likewise.
7566 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7567 (vect_transform_slp_perm_load): Likewise.
7568 (vect_schedule_slp_instance): Likewise.
7569 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7570 (vectorizable_call): Likewise.
7571 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7572 * expmed.c (make_tree): Likewise.
7573 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7574 a vector passed to build_vector.
7575 (fold_convert_const): Likewise.
7576 (exact_inverse): Likewise.
7577 (fold_ternary_loc): Likewise.
7578 (fold_relational_const): Likewise.
7579 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7580 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7581 (const_unop): Likewise. Store the reduction accumulator in a
7582 variable rather than an array.
7583 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7584 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7585 the new vector, rather than constructing it after the input arrays.
7586 (native_interpret_vector): Use auto_vec<tree> when building
7587 a vector passed to build_vector. Add elements in order.
7588 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7589 auto_vec<tree> when building a vector passed to build_vector.
7590 (vect_create_epilog_for_reduction): Likewise.
7591 (vectorizable_induction): Likewise.
7592 (get_initial_def_for_reduction): Likewise. Fix indentation of
7594 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7596 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7597 passed to build_vector.
7599 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7600 Alan Hayward <alan.hayward@arm.com>
7601 David Sherwood <david.sherwood@arm.com>
7603 * tree-core.h (tree_base::u): Add an "nelts" field.
7604 (tree_vector): Use VECTOR_CST_NELTS as the length.
7605 * tree.c (tree_size): Likewise.
7606 (make_vector): Initialize VECTOR_CST_NELTS.
7607 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7608 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7609 TYPE_VECTOR_SUBPARTS.
7610 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7611 as the number of units, setting it from VECTOR_CST_NELTS.
7612 (const_vector_from_tree): Likewise.
7613 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7614 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7615 (fold_negate_expr_1): Likewise.
7616 (fold_convert_const): Likewise.
7617 (const_binop): Likewise. Differentiate the number of output and
7619 (const_unop): Likewise.
7620 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7621 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7622 in cases that did the opposite.
7624 2017-09-14 Richard Biener <rguenther@suse.de>
7626 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7629 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7631 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7633 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7636 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7637 if and where to split a bb, except for splitting before debug insn
7638 sequences followed by non-label real insn. Delete debug insns
7639 in between basic blocks.
7641 * combine.c (make_compound_operation_int): Formatting fixes.
7643 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7644 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7645 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7646 * config/sol2.h (LINK_EH_SPEC): Likewise.
7647 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7648 * config/s390/linux.h (LINK_SPEC): Likewise.
7649 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7650 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7651 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7652 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7653 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7654 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7655 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7656 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7658 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7660 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7662 (ENDFILE_LINUX_SPEC): Likewise.
7663 (LINK_EH_SPEC): Likewise.
7664 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7665 (LINK_OS_LINUX_SPEC32): Likewise.
7666 (LINK_OS_LINUX_SPEC64): Likewise.
7667 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7668 (LINK_OS_LINUX_SPEC): Likewise.
7670 2017-09-13 Martin Liska <mliska@suse.cz>
7673 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
7674 CASE_HIGH is NULL_TREE.
7676 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7677 Alan Hayward <alan.hayward@arm.com>
7678 David Sherwood <david.sherwood@arm.com>
7680 * target.def (secondary_memory_needed): New hook.
7681 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
7682 instead of SECONDARY_MEMORY_NEEDED.
7683 (secondary_memory_needed_mode): Likewise.
7684 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
7685 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
7686 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
7687 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
7688 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7689 * doc/tm.texi: Regenerate.
7690 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7691 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7692 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7693 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7694 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7695 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7696 mode argument first and change the reg_class arguments to reg_class_t.
7697 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7698 Make static. Update the call to inline_secondary_memory_needed.
7699 (ix86_register_move_cost): Update the call to
7700 inline_secondary_memory_needed.
7701 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7702 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7704 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7705 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7707 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7708 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7709 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7710 and match hook interface. Add comment from mips.h.
7711 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7712 * config/mmix/mmix.md (truncdfsf2): Refer to
7713 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7715 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7716 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7717 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7718 (pa_secondary_memory_needed): New function.
7719 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7720 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7721 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7722 (pdp11_secondary_memory_needed): Make static and match hook interface.
7723 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7724 * config/powerpcspe/powerpcspe-protos.h
7725 (rs6000_secondary_memory_needed_ptr): Delete.
7726 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7728 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7729 (rs6000_option_override_internal): Assign to
7730 targetm.secondary_memory_needed rather than
7731 rs6000_secondary_memory_needed_ptr.
7732 (rs6000_secondary_memory_needed): Match hook interface.
7733 (rs6000_debug_secondary_memory_needed): Likewise.
7734 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7735 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7736 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7737 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7738 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7739 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7741 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7742 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7743 (rs6000_option_override_internal): Assign to
7744 targetm.secondary_memory_needed rather than
7745 rs6000_secondary_memory_needed_ptr.
7746 (rs6000_secondary_memory_needed): Match hook interface.
7747 (rs6000_debug_secondary_memory_needed): Likewise.
7748 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7749 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7750 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7751 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7752 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7753 (sparc_secondary_memory_needed): New function.
7754 * lra-constraints.c (check_and_process_move): Refer to
7755 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7757 (curr_insn_transform): Likewise.
7758 (process_alt_operands): Use targetm.secondary_memory_needed
7759 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7760 (check_secondary_memory_needed_p): Likewise.
7761 (choose_split_class): Likewise.
7762 * reload.c: Unconditionally include code that was previously
7763 conditional on SECONDARY_MEMORY_NEEDED.
7764 (push_secondary_reload): Use targetm.secondary_memory_needed
7765 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7766 (push_reload): Likewise.
7767 * reload1.c: Unconditionally include code that was previously
7768 conditional on SECONDARY_MEMORY_NEEDED.
7769 (choose_reload_regs): Use targetm.secondary_memory_needed
7770 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7771 (gen_reload): Likewise.
7772 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7774 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7775 Alan Hayward <alan.hayward@arm.com>
7776 David Sherwood <david.sherwood@arm.com>
7778 * target.def (secondary_memory_needed_mode): New hook:
7779 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7780 * targhooks.h (default_secondary_memory_needed_mode): New function.
7781 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7782 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7783 * doc/tm.texi: Regenerate.
7784 * lra-constraints.c (check_and_process_move): Use
7785 targetm.secondary_memory_needed_mode instead of
7786 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7787 (curr_insn_transform): Likewise.
7788 * reload.c (get_secondary_mem): Likewise.
7789 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7790 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7792 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7793 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7794 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7795 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7796 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7798 * config/powerpcspe/powerpcspe-protos.h
7799 (rs6000_secondary_memory_needed_mode): Delete.
7800 * config/powerpcspe/powerpcspe.c
7801 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7802 (rs6000_secondary_memory_needed_mode): Make static.
7803 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7804 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7806 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7808 (rs6000_secondary_memory_needed_mode): Make static.
7809 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7810 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7811 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7812 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7813 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7815 (sparc_secondary_memory_needed_mode): New function.
7816 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7818 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7820 * config/aarch64/constraints.md (Umq): New constraint.
7821 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7823 (mov<mode>): Update condition.
7825 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7827 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7828 when bitposition is the same.
7830 2017-09-13 Richard Biener <rguenther@suse.de>
7832 * dwarf2out.c (output_die_symbol): Remove.
7833 (output_die): Do not output a DIEs symbol.
7835 2017-09-13 Richard Biener <rguenther@suse.de>
7838 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7839 default-def to avoid breaking iterator update with the weird
7840 interaction with cgraph_update_edges_for_call_stmt_node.
7842 2017-09-13 Richard Biener <rguenther@suse.de>
7844 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7845 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7846 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7847 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7849 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7851 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7852 Disable pc relative literal load irrespective of
7853 TARGET_FIX_ERR_A53_84341 for default.
7855 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7857 * config/sparc/sparc.c (output_return): Output the source location of
7858 the insn in the delay slot, if any.
7859 (output_sibcall): Likewise.
7861 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7864 * common.opt (-static-pie): New alias.
7865 (shared): Negate static-pie.
7866 (-no-pie): Update help text.
7868 (static-pie): New option.
7869 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7870 -static-pie support.
7871 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7872 (LINK_EH_SPEC): Likewise.
7873 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7874 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7875 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7876 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7877 (init_gcc_specs): Likewise.
7878 (init_spec): Likewise.
7879 (display_help): Update help message for -pie.
7880 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7883 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7885 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7886 (movdi_aarch64): Likewise.
7887 (movti_aarch64): Likewise.
7889 2017-09-12 Simon Wright <simon@pushface.org>
7892 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7893 calculation of the minor version, always output as 0.
7895 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7898 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7899 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7900 on it early, rather than manual conversion late. For
7901 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7902 instead of performing manual conversion.
7904 2017-09-12 Carl Love <cel@us.ibm.com>
7906 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7907 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7909 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7910 VMULOSW): Add definitions.
7911 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7912 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7913 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7914 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7915 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7917 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7919 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7921 (movti_aarch64): Likewise.
7922 (movdf_aarch64): Likewise.
7923 (movtf_aarch64): Likewise.
7924 (load_pairdi): Likewise.
7925 (store_pairdi): Likewise.
7926 (load_pairdf): Likewise.
7927 (store_pairdf): Likewise.
7928 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7929 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7930 (ldr_got_small_<mode>): Likewise.
7931 (ldr_got_small_28k_<mode>): Likewise.
7932 (ldr_got_tiny): Likewise.
7933 * config/aarch64/iterators.md (ldst_sz): New.
7934 (ldpstp_sz): Likewise.
7935 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7937 (thunderx_load): Split load_8 to load_16.
7938 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7940 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7941 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7942 (xgene1_store_pair): Split store_8 to store_16.
7943 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7944 (falkor_st_0_st_sd): Split store_8 to store_16.
7946 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7948 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7949 and store1/2/3/4 to store_4/8/12/16.
7950 * config/aarch64/aarch64.md: Update for rename.
7951 * config/arm/arm.md: Likewise.: Likewise.
7952 * config/arm/arm.c: Likewise.
7953 * config/arm/thumb1.md: Likewise.
7954 * config/arm/thumb2.md: Likewise.
7955 * config/arm/vfp.md: Likewise.
7956 * config/arm/arm-generic.md: Likewise.
7957 * config/arm/arm1020e.md: Likewise.
7958 * config/arm/arm1026ejs.md: Likewise.
7959 * config/arm/arm1136jfs.md: Likewise.
7960 * config/arm/arm926ejs.md: Likewise.
7961 * config/arm/cortex-a15.md: Likewise.
7962 * config/arm/cortex-a17.md: Likewise.
7963 * config/arm/cortex-a5.md: Likewise.
7964 * config/arm/cortex-a53.md: Likewise.
7965 * config/arm/cortex-a57.md: Likewise.
7966 * config/arm/cortex-a7.md: Likewise.
7967 * config/arm/cortex-a8.md: Likewise.
7968 * config/arm/cortex-a9.md: Likewise.
7969 * config/arm/cortex-m4.md: Likewise.
7970 * config/arm/cortex-m7.md: Likewise.
7971 * config/arm/cortex-r4.md: Likewise.
7972 * config/arm/exynos-m1.md: Likewise.
7973 * config/arm/fa526.md: Likewise.
7974 * config/arm/fa606te.md: Likewise.
7975 * config/arm/fa626te.md: Likewise.
7976 * config/arm/fa726te.md: Likewise.
7977 * config/arm/fmp626.md: Likewise.
7978 * config/arm/iwmmxt.md: Likewise.
7979 * config/arm/ldmstm.md: Likewise.
7980 * config/arm/marvell-pj4.md: Likewise.
7981 * config/arm/xgene1.md: Likewise.
7982 * config/aarch64/thunderx.md: Likewise.
7983 * config/aarch64/thunderx2t99.md: Likewise.
7984 * config/aarch64/falkor.md: Likewise.
7986 2017-09-12 Martin Liska <mliska@suse.cz>
7988 * attribs.c (private_lookup_attribute): New function.
7989 * attribs.h (private_lookup_attribute): Declared here.
7990 (lookup_attribute): Called from this place.
7992 2017-09-12 Richard Biener <rguenther@suse.de>
7994 PR tree-optimization/82157
7995 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
7996 stmts with side-effects.
7998 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7999 Alan Hayward <alan.hayward@arm.com>
8000 David Sherwood <david.sherwood@arm.com>
8002 * target.def (hard_regno_nregs): New hook.
8003 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8004 * targhooks.h (default_hard_regno_nregs): Declare.
8005 * targhooks.c (default_hard_regno_nregs): New function.
8006 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8007 (TARGET_HARD_REGNO_NREGS): ...this hook.
8008 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8009 (CLASS_MAX_NREGS): Likewise.
8010 * doc/tm.texi: Regenerate.
8011 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8012 instead of HARD_REGNO_NREGS.
8013 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8014 HARD_REGNO_NREGS in the comment.
8015 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8016 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8017 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8018 Return an unsigned int.
8019 (TARGET_HARD_REGNO_NREGS): Redefine.
8020 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8021 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8022 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8023 (arc_hard_regno_nregs): New function.
8024 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8025 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8026 (arm_hard_regno_nregs): New function.
8027 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8028 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8029 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8030 (TARGET_HARD_REGNO_NREGS): Redefine.
8031 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8032 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8033 (HARD_REGNO_NREGS): Delete.
8034 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8035 (cr16_hard_regno_nregs): New function.
8036 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8037 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8038 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8039 (cris_hard_regno_nregs): New function.
8040 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8041 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8042 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8043 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8044 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8045 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8046 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8047 (frv_hard_regno_nregs): Make static. Take and return an
8049 (frv_class_max_nregs): Remove outdated copy of documentation.
8050 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8051 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8052 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8053 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8054 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8055 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8056 (TARGET_HARD_REGNO_NREGS): Redefine.
8057 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8058 (CLASS_MAX_NREGS): Update comment.
8059 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8060 (ia64_hard_regno_nregs): New function.
8061 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8062 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8063 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8064 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8065 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8067 (m32c_hard_regno_nregs): Likewise. Make static.
8068 (TARGET_HARD_REGNO_NREGS): Redefine.
8069 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8070 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8071 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8072 (m68k_hard_regno_nregs): New function.
8073 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8074 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8075 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8076 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8077 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8078 Take and return an unsigned int.
8079 (TARGET_HARD_REGNO_NREGS): Redefine.
8080 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8081 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8082 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8083 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8084 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8085 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8086 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8087 (msp430_hard_regno_nregs): Make static. Take and return an
8089 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8090 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8091 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8092 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8093 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8094 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8095 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8096 (TARGET_HARD_REGNO_NREGS): Redefine.
8097 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8098 (PA_HARD_REGNO_NREGS): ...this.
8099 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8100 (PA_HARD_REGNO_NREGS): ...this.
8101 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8102 (pa_hard_regno_nregs): New function.
8103 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8104 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8105 (pdp11_hard_regno_nregs): New function.
8106 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8107 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8108 (rs6000_hard_regno_nregs_hook): New function.
8109 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8110 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8111 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8112 Take and return an unsigned int. Move earlier in file.
8113 (TARGET_HARD_REGNO_NREGS): Redefine.
8114 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8115 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8116 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8117 (rl78_hard_regno_nregs): Make static. Take and return an
8119 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8120 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8121 (rs6000_hard_regno_nregs_hook): New function.
8122 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8123 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8124 (TARGET_HARD_REGNO_NREGS): Redefine.
8125 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8126 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8127 instead of HARD_REGNO_NREGS.
8128 (s390_hard_regno_nregs): New function.
8129 (s390_hard_regno_mode_ok): Add comment from s390.h.
8130 (TARGET_HARD_REGNO_NREGS): Redefine.
8131 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8132 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8133 (sh_hard_regno_nregs): New function.
8134 (sh_pass_in_reg_p): Use it.
8135 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8136 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8137 (sparc_hard_regno_nregs): New function.
8138 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8139 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8140 (spu_function_arg_advance): Use it, supplying a valid register number.
8141 (TARGET_HARD_REGNO_NREGS): Redefine.
8142 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8143 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8144 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8145 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8146 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8147 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8148 (CLASS_MAX_NREGS): Remove copy of old documentation.
8149 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8150 (visium_hard_regno_nregs): New function.
8151 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8152 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8153 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8154 xtensa_hard_regno_nregs): New function.
8155 * system.h (HARD_REGNO_NREGS): Poison.
8157 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8159 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8160 hard_regno_nregs instead of HARD_REGNO_NREGS.
8161 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8162 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8163 (c6x_expand_epilogue): Likewise.
8164 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8165 (frv_read_iacc_argument): Likewise.
8166 * config/sh/sh.c: Include regs.h.
8167 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8168 (regs_used): Likewise.
8169 (output_stack_adjust): Likewise.
8170 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8171 * expmed.c: Include regs.h.
8172 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8173 * ree.c: Include regs.h.
8174 (combine_reaching_defs): Use hard_regno_nregs instead of
8176 (add_removable_extension): Likewise.
8178 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8180 * regs.h (hard_regno_nregs): Turn into a function.
8181 (end_hard_regno): Update accordingly.
8182 * caller-save.c (setup_save_areas): Likewise.
8183 (save_call_clobbered_regs): Likewise.
8184 (replace_reg_with_saved_mem): Likewise.
8185 (insert_restore): Likewise.
8186 (insert_save): Likewise.
8187 * combine.c (can_change_dest_mode): Likewise.
8188 (move_deaths): Likewise.
8189 (distribute_notes): Likewise.
8190 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
8191 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
8192 (rs6000_split_multireg_move): Likewise.
8193 (rs6000_register_move_cost): Likewise.
8194 (rs6000_memory_move_cost): Likewise.
8195 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
8196 (rs6000_split_multireg_move): Likewise.
8197 (rs6000_register_move_cost): Likewise.
8198 (rs6000_memory_move_cost): Likewise.
8199 * cselib.c (cselib_reset_table): Likewise.
8200 (cselib_lookup_1): Likewise.
8201 * emit-rtl.c (set_mode_and_regno): Likewise.
8202 * function.c (aggregate_value_p): Likewise.
8203 * ira-color.c (setup_profitable_hard_regs): Likewise.
8204 (check_hard_reg_p): Likewise.
8205 (calculate_saved_nregs): Likewise.
8206 (assign_hard_reg): Likewise.
8207 (improve_allocation): Likewise.
8208 (calculate_spill_cost): Likewise.
8209 * ira-emit.c (modify_move_list): Likewise.
8210 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
8211 (ira_hard_reg_in_set_p): Likewise.
8212 * ira.c (setup_reg_mode_hard_regset): Likewise.
8213 (clarify_prohibited_class_mode_regs): Likewise.
8214 (check_allocation): Likewise.
8215 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8216 (lra_setup_reg_renumber): Likewise.
8217 (setup_try_hard_regno_pseudos): Likewise.
8218 (spill_for): Likewise.
8219 (assign_hard_regno): Likewise.
8220 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
8221 * lra-constraints.c (in_class_p): Likewise.
8222 (lra_constraint_offset): Likewise.
8223 (simplify_operand_subreg): Likewise.
8224 (lra_constraints): Likewise.
8225 (split_reg): Likewise.
8226 (split_if_necessary): Likewise.
8227 (invariant_p): Likewise.
8228 (inherit_in_ebb): Likewise.
8229 * lra-lives.c (process_bb_lives): Likewise.
8230 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
8231 (get_hard_regs): Likewise.
8232 (do_remat): Likewise.
8233 * lra-spills.c (assign_spill_hard_regs): Likewise.
8234 * mode-switching.c (create_pre_exit): Likewise.
8235 * postreload.c (reload_combine_recognize_pattern): Likewise.
8236 * recog.c (peep2_find_free_register): Likewise.
8237 * regcprop.c (kill_value_regno): Likewise.
8238 (set_value_regno): Likewise.
8239 (copy_value): Likewise.
8240 (maybe_mode_change): Likewise.
8241 (find_oldest_value_reg): Likewise.
8242 (copyprop_hardreg_forward_1): Likewise.
8243 * regrename.c (check_new_reg_p): Likewise.
8244 (regrename_do_replace): Likewise.
8245 * reload.c (push_reload): Likewise.
8246 (combine_reloads): Likewise.
8247 (find_dummy_reload): Likewise.
8248 (operands_match_p): Likewise.
8249 (find_reloads): Likewise.
8250 (find_equiv_reg): Likewise.
8251 (reload_adjust_reg_for_mode): Likewise.
8252 * reload1.c (count_pseudo): Likewise.
8253 (count_spilled_pseudo): Likewise.
8254 (find_reg): Likewise.
8255 (clear_reload_reg_in_use): Likewise.
8256 (free_for_value_p): Likewise.
8257 (allocate_reload_reg): Likewise.
8258 (choose_reload_regs): Likewise.
8259 (reload_adjust_reg_for_temp): Likewise.
8260 (emit_reload_insns): Likewise.
8261 (delete_output_reload): Likewise.
8262 * rtlanal.c (subreg_get_info): Likewise.
8263 * sched-deps.c (sched_analyze_reg): Likewise.
8264 * sel-sched.c (init_regs_for_mode): Likewise.
8265 (mark_unavailable_hard_regs): Likewise.
8266 (choose_best_reg_1): Likewise.
8267 (verify_target_availability): Likewise.
8268 * valtrack.c (dead_debug_insert_temp): Likewise.
8269 * var-tracking.c (track_loc_p): Likewise.
8270 (emit_note_insn_var_location): Likewise.
8271 * varasm.c (make_decl_rtl): Likewise.
8272 * reginfo.c (choose_hard_reg_mode): Likewise.
8273 (init_reg_modes_target): Refer directly to
8274 this_target_regs->x_hard_regno_nregs.
8276 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8278 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8279 instead of hard_regno_nregs.
8281 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8283 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8284 end_hard_regno instead of hard_regno_nregs.
8285 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8286 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8287 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8288 * ira-color.c (improve_allocation): Likewise.
8289 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8290 * lra-lives.c (mark_regno_live): Likewise.
8291 (mark_regno_dead): Likewise.
8292 * lra-remat.c (operand_to_remat): Likewise.
8293 * lra.c (collect_non_operand_hard_regs): Likewise.
8294 * postreload.c (reload_combine_note_store): Likewise.
8295 (move2add_valid_value_p): Likewise.
8296 * reload.c (regno_clobbered_p): Likewise.
8298 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8300 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8302 * config/v850/v850.c (v850_reorg): Likewise.
8303 * reload.c (refers_to_regno_for_reload_p): Likewise.
8304 (find_equiv_reg): Likewise.
8305 * reload1.c (reload_reg_reaches_end_p): Likewise.
8307 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8309 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8311 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8312 * config/arm/arm.c (output_move_neon): Likewise.
8313 (arm_attr_length_move_neon): Likewise.
8314 (neon_split_vcombine): Likewise.
8315 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8316 (c6x_mark_reg_written): Likewise.
8317 (c6x_dwarf_register_span): Likewise.
8318 * config/i386/i386.c (ix86_save_reg): Likewise.
8319 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8320 (rws_access_reg): Likewise.
8321 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8322 * mode-switching.c (create_pre_exit): Likewise.
8323 * ree.c (combine_reaching_defs): Likewise.
8324 (add_removable_extension): Likewise.
8325 * regcprop.c (find_oldest_value_reg): Likewise.
8326 (copyprop_hardreg_forward_1): Likewise.
8327 * reload.c (reload_inner_reg_of_subreg): Likewise.
8328 (push_reload): Likewise.
8329 (combine_reloads): Likewise.
8330 (find_dummy_reload): Likewise.
8331 (reload_adjust_reg_for_mode): Likewise.
8332 * reload1.c (find_reload_regs): Likewise.
8333 (forget_old_reloads_1): Likewise.
8334 (reload_reg_free_for_value_p): Likewise.
8335 (reload_adjust_reg_for_temp): Likewise.
8336 (emit_reload_insns): Likewise.
8337 (delete_output_reload): Likewise.
8338 * sel-sched.c (choose_best_reg_1): Likewise.
8339 (choose_best_pseudo_reg): Likewise.
8341 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8342 Alan Hayward <alan.hayward@arm.com>
8343 David Sherwood <david.sherwood@arm.com>
8345 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8346 * target.def (slow_unaligned_access): New hook.
8347 * targhooks.h (default_slow_unaligned_access): Declare.
8348 * targhooks.c (default_slow_unaligned_access): New function.
8349 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8350 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8351 * doc/tm.texi: Regenerate.
8352 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8353 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8354 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8356 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8357 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8359 (rs6000_slow_unaligned_access): New function.
8360 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8361 (expand_block_compare): Likewise.
8362 (expand_strn_compare): Likewise.
8363 (rs6000_rtx_costs): Likewise.
8364 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8365 (riscv_slow_unaligned_access): Likewise.
8366 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8367 (riscv_slow_unaligned_access_p): ...this and make static.
8368 (riscv_option_override): Update accordingly.
8369 (riscv_slow_unaligned_access): New function.
8370 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8371 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8372 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8373 (rs6000_slow_unaligned_access): New function.
8374 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8375 (rs6000_rtx_costs): Likewise.
8376 * config/rs6000/rs6000-string.c (expand_block_compare)
8377 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8378 of SLOW_UNALIGNED_ACCESS.
8379 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8380 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8381 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8382 of SLOW_UNALIGNED_ACCESS.
8383 * expmed.c (simple_mem_bitfield_p): Likewise.
8384 * expr.c (alignment_for_piecewise_move): Likewise.
8385 (emit_group_load_1): Likewise.
8386 (emit_group_store): Likewise.
8387 (copy_blkmode_from_reg): Likewise.
8388 (emit_push_insn): Likewise.
8389 (expand_assignment): Likewise.
8390 (store_field): Likewise.
8391 (expand_expr_real_1): Likewise.
8392 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8393 * lra-constraints.c (simplify_operand_subreg): Likewise.
8394 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8395 * gimple-ssa-store-merging.c: Likewise in block comment at start
8397 * tree-ssa-strlen.c: Include target.h.
8398 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8399 of SLOW_UNALIGNED_ACCESS.
8400 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8402 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8404 PR rtl-optimization/82185
8405 * expmed.c (emit_store_flag_int): Only test tem if it has been
8408 2017-09-12 Richard Biener <rguenther@suse.de>
8411 * match.pd ((FTYPE) N CMP CST): Fix typo.
8413 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8415 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8417 (mips_near_type_p): Add 'short_call' attribute as a synonym
8419 * doc/extend.texi (short_call): Document new function attribute.
8421 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8424 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8425 assertion check that in the condition.
8426 (get_atomic_generic_size): Likewise. Before testing if parameter
8427 has pointer type, if it has array type, call for C++
8428 default_conversion to perform array-to-pointer conversion.
8430 2017-09-12 Richard Biener <rguenther@suse.de>
8432 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8433 for operations we cannot scalarize.
8435 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8437 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8438 vectors heap vectors. Clean up comments.
8439 Make visited_bbs a reference.
8440 (profitable_jump_thread_path): Make GC
8441 vectors heap vectors. Clean up comments.
8443 (convert_and_register_jump_thread_path): Make GC vectors heap
8445 (check_subpath_and_update_thread_path): Same. Clean up comments.
8446 Make visited_bbs a reference.
8447 (handle_phi): Abstract common code to to
8448 register_jump_thread_path_if_profitable.
8449 Rename VAR_BB to DEF_BB.
8451 Make GC vectors heap vectors.
8452 Make visited_bbs a reference.
8453 (handle_assignment): Same.
8454 (register_jump_thread_path_if_profitable): New.
8455 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8457 Make GC vectors heap vectors. Clean up comments.
8458 Make visited_bbs a reference.
8459 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8460 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8463 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8466 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8467 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8469 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8471 Revert r251800 and r251799.
8473 2017-09-11 Martin Jambor <mjambor@suse.cz>
8476 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8477 arguments in advance.
8478 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8479 use it to find predecessor edges.
8480 (naive_outof_ssa): Collect vector of predecessors.
8482 2017-09-08 Jason Merrill <jason@redhat.com>
8484 PR c++/70029 - ICE with ref-qualifier and -flto
8485 * langhooks.h (struct lang_hooks_for_types): Add
8486 copy_lang_qualifiers.
8487 * attribs.c (build_type_attribute_qual_variant): Use it.
8488 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8490 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8491 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8493 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8496 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8497 (*mulsi3_sp64): New instruction.
8498 (mulsi3): New expander.
8500 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8502 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8504 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8506 * sancov.c: Include memmodel.h.
8508 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8511 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8514 2017-09-07 Carl Love <cel@us.ibm.com>
8516 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8517 the sldi instruction.
8519 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8521 * sancov.c: Include tm_p.h.
8523 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8526 * output.h (switch_to_other_text_partition): New declaration.
8527 * varasm.c (switch_to_other_text_partition): New function.
8528 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8529 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8530 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8531 to the other text partition before emitting LCL label and switch back
8532 after emitting the word after it.
8534 2017-09-07 Richard Biener <rguenther@suse.de>
8536 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8537 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8538 critical edges here, after loop init.
8539 (pass_data_pre): Remove PROP_no_crit_edges flags.
8540 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8541 for valueization of call args to avoid leaking VN_TOP.
8542 (visit_use): Assert we do not visit default defs.
8543 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8544 Use error_mark_node to more easily detect leaking VN_TOP.
8545 All default-defs are varying, not VN_TOP. Mark them visited.
8546 (run_scc_vn): Make code match comment.
8548 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8550 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8551 OPTION_MASK_FLOAT128_KEYWORD.
8552 (POWERPC_MASKS): Likewise.
8553 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8554 support for the -mfloat128-type option, and make -mfloat128
8555 default on PowerPC Linux systems. Define or undefine
8556 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8557 Define __float128 to be __ieee128 if IEEE 128-bit support is
8558 enabled, or undefine it.
8559 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8560 Delete defining __FLOAT128_TYPE__.
8561 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8562 -mfloat128-type option and make -mfloat128 default on PowerPC
8564 (TARGET_FLOAT128_TYPE): Likewise.
8565 (-mfloat128-type): Likewise.
8566 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8567 Delete the -mfloat128-type option and make -mfloat128 default on
8568 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8569 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8570 128-bit floating point is enabled. Change tests from using
8571 -mfloat128-type to -mfloat128.
8572 (rs6000_mangle_type): Use the correct mangling for the __float128
8573 type even if normal long double is restricted to 64-bits.
8574 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8576 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8577 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8578 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8580 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8581 documentation for -mfloat128.
8583 2017-09-06 Olivier Hainque <hainque@adacore.com>
8585 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8587 2017-09-06 Wish Wu <wishwu007@gmail.com>
8588 Jakub Jelinek <jakub@redhat.com>
8590 * asan.c (initialize_sanitizer_builtins): Add
8591 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8592 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8593 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8594 BT_FN_VOID_UINT64_PTR variables.
8595 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8596 (BT_FN_VOID_UINT16_UINT16): Likewise.
8597 (BT_FN_VOID_UINT32_UINT32): Likewise.
8598 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8599 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8600 (BT_FN_VOID_UINT64_PTR): Likewise.
8601 * common.opt (flag_sanitize_coverage): New variable.
8602 (fsanitize-coverage=trace-pc): Remove.
8603 (fsanitize-coverage=): Add.
8604 * flag-types.h (enum sanitize_coverage_code): New enum.
8605 * fold-const.c (fold_range_test): Disable non-short-circuit
8606 optimization if flag_sanitize_coverage.
8607 (fold_truth_andor): Likewise.
8608 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8609 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8610 (coverage_sanitizer_opts): New array.
8611 (get_closest_sanitizer_option): Add OPTS argument, handle also
8612 OPT_fsanitize_coverage_.
8613 (parse_sanitizer_options): Adjusted to also handle
8614 OPT_fsanitize_coverage_.
8615 (common_handle_option): Add OPT_fsanitize_coverage_.
8616 * sancov.c (instrument_comparison, instrument_switch): New function.
8617 (sancov_pass): Add trace-cmp support.
8618 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8619 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8620 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8621 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8622 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8623 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8624 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8625 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8626 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8627 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8629 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8631 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8632 error. Only quit immediately if parsing is complete.
8633 (BEGIN): Initialize fatal_err and parse_done.
8634 (begin fpu, end fpu): Check number of arguments.
8635 (begin arch, end arch): Likewise.
8636 (begin cpu, end cpu): Likewise.
8637 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8638 (optalias): Likewise.
8640 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8642 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8643 * config/arm/arm-isa.h: Delete. Move definitions to ...
8644 * arm-cpus.in: ... here. Use new feature and fgroup values.
8645 * config/arm/arm.c (arm_option_override): Use lower case for feature
8647 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8648 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8649 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8651 (print_isa_bits_for): New function.
8652 (gen_isa): New function.
8653 (gen_comm_data): Use print_isa_bits_for.
8654 (define feature): New keyword.
8655 (define fgroup): New keyword.
8656 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8657 (arm-isa.h): Add rule to generate file.
8658 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8659 case for feature bit names.
8661 2017-09-06 Richard Biener <rguenther@suse.de>
8663 * tree-ssa-pre.c (NECESSARY): Remove.
8664 (create_expression_by_pieces): Do not touch pass-local flags.
8665 (insert_into_preds_of_block): Likewise.
8666 (do_pre_regular_insertion): Likewise.
8667 (eliminate_insert): Likewise.
8668 (eliminate_dom_walker::before_dom_children): Likewise.
8669 (fini_eliminate): Do not look at inserted_exprs.
8670 (mark_operand_necessary): Remove.
8671 (remove_dead_inserted_code): Replace with simple work-list
8672 algorithm based on inserted_exprs and SSA uses.
8673 (pass_pre::execute): Re-order fini_eliminate and
8674 remove_dead_inserted_code.
8676 2017-09-06 Olivier Hainque <hainque@adacore.com>
8678 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
8679 for VxWorks 7. Adjust surrounding comments.
8681 2017-09-06 Richard Biener <rguenther@suse.de>
8683 * gimple-ssa-strength-reduction.c
8684 (find_candidates_dom_walker::before_dom_children): Also allow
8687 2017-09-06 Richard Biener <rguenther@suse.de>
8689 PR tree-optimization/82108
8690 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8691 for gap in the non-permutation SLP case.
8693 2017-09-06 Martin Jambor <mjambor@suse.cz>
8695 PR tree-optimization/82078
8696 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8697 add_access_to_work_queue...
8698 (build_accesses_from_assign): ...here.
8699 (propagate_all_subaccesses): Make sure racc is the group
8700 representative, if there is one.
8702 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8705 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8708 2017-09-06 Richard Biener <rguenther@suse.de>
8710 * gimple-ssa-strength-reduction.c
8711 (find_candidates_dom_walker::before_doom_children): Use a
8712 type and not a mode check.
8714 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8717 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8719 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8721 2017-09-05 Jeff Law <law@redhat.com>
8723 PR tree-optimization/64910
8724 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8725 cases where we have 3 or more operands.
8727 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8730 * omp-low.c (lower_omp_for): Recompute tree invariant if
8731 gimple_omp_for_initial/final is ADDR_EXPR.
8734 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8735 into gimple val before gimplification fo the COND_EXPR.
8737 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8739 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8740 REGION_COPY argument.
8741 (thread_through_all_blocks): Remove unused argument to
8742 duplicate_thread_path.
8744 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8745 Alan Hayward <alan.hayward@arm.com>
8746 David Sherwood <david.sherwood@arm.com>
8748 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8749 Take a scalar_mode rather than a machine_mode.
8750 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8751 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8752 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8753 (aarch64_gen_adjusted_ldpstp): Likewise.
8754 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8756 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8757 Alan Hayward <alan.hayward@arm.com>
8758 David Sherwood <david.sherwood@arm.com>
8760 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8761 Take a scalar_int_mode instead of a machine_mode.
8762 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8763 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8764 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8765 (aarch64_simd_attr_length_rglist): Delete.
8766 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8767 a scalar_int_mode instead of a machine_mode.
8768 (aarch64_add_offset): Likewise.
8769 (aarch64_internal_mov_immediate): Likewise
8770 (aarch64_add_constant_internal): Likewise.
8771 (aarch64_add_constant): Likewise.
8772 (aarch64_movw_imm): Likewise.
8773 (aarch64_rtx_arith_op_extract_p): Likewise.
8774 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8775 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8776 Remove assert that the mode isn't a vector.
8777 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8778 (aarch64_expand_mov_immediate): Update calls after above changes.
8779 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8780 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8781 (aarch64_move_imm): Likewise.
8782 (aarch64_can_const_movi_rtx_p): Likewise.
8783 (aarch64_strip_extend): Likewise.
8784 (aarch64_extr_rtx_p): Likewise.
8785 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8786 a CONST_INT when the mode parameter is VOIDmode.
8787 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8789 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8791 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8792 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8793 (bitwise_type_for_mode): Update accordingly.
8795 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8797 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8798 * stor-layout.c (mode_for_size_tree): Likewise.
8799 (mode_for_array): Update accordingly.
8800 (layout_decl): Likewise.
8801 (compute_record_mode): Likewise. Only set the mode once.
8803 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8805 * target.def (get_mask_mode): Change return type to opt_mode.
8807 * doc/tm.texi: Regenerate.
8808 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8809 * targhooks.c (default_get_mask_mode): Likewise.
8810 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8811 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8812 targetm.get_mask_mode.
8813 * tree.c (build_truth_vector_type): Likewise.
8815 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8817 * machmode.h (mode_for_vector): Return an opt_mode.
8818 * stor-layout.c (mode_for_vector): Likewise.
8819 (mode_for_int_vector): Update accordingly.
8820 (layout_type): Likewise.
8821 * config/i386/i386.c (emit_memmov): Likewise.
8822 (ix86_expand_set_or_movmem): Likewise.
8823 (ix86_expand_vector_init): Likewise.
8824 (ix86_get_mask_mode): Likewise.
8825 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8827 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8828 * expmed.c (extract_bit_field_1): Likewise.
8829 * expr.c (expand_expr_real_2): Likewise.
8830 * optabs-query.c (can_vec_perm_p): Likewise.
8831 (can_vec_mask_load_store_p): Likewise.
8832 * optabs.c (expand_vec_perm): Likewise.
8833 * targhooks.c (default_get_mask_mode): Likewise.
8834 * tree-vect-stmts.c (vectorizable_store): Likewise.
8835 (vectorizable_load): Likewise.
8836 (get_vectype_for_scalar_type_and_size): Likewise.
8838 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8840 * machmode.h (mode_for_int_vector): New function.
8841 * stor-layout.c (mode_for_int_vector): Likewise.
8842 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8843 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8844 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8845 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8846 (s390_expand_vcond): Likewise.
8848 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8850 * machmode.h (opt_machine_mode): New type.
8851 (opt_mode<T>): Allow construction from anything that can be
8853 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8854 (mode_for_size): Return an opt_machine_mode.
8855 * stor-layout.c (mode_for_size): Likewise.
8856 (mode_for_size_tree): Update call accordingly.
8857 (bitwise_mode_for_mode): Likewise.
8858 (make_fract_type): Likewise.
8859 (make_accum_type): Likewise.
8860 * caller-save.c (replace_reg_with_saved_mem): Update call
8862 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8863 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8864 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8865 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8866 * expmed.c (extract_bit_field_1): Likewise.
8867 * reload.c (get_secondary_mem): Likewise.
8868 * varasm.c (assemble_integer): Likewise.
8869 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8872 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8874 * machmode.h (decimal_float_mode_for_size): New function.
8875 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8876 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8877 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8878 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8879 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8880 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8882 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8884 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8885 (get_builtin_sync_mode): Likewise.
8886 (expand_ifn_atomic_compare_exchange): Likewise.
8887 (expand_builtin_atomic_clear): Likewise.
8888 (expand_builtin_atomic_test_and_set): Likewise.
8889 (fold_builtin_atomic_always_lock_free): Likewise.
8890 * calls.c (compute_argument_addresses): Likewise.
8891 (emit_library_call_value_1): Likewise.
8892 (store_one_arg): Likewise.
8893 * combine.c (combine_instructions): Likewise.
8894 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8895 * config/arm/arm.c (arm_function_value): Likewise.
8896 (aapcs_allocate_return_reg): Likewise.
8897 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8898 * config/i386/i386.c (construct_container): Likewise.
8899 (ix86_gimplify_va_arg): Likewise.
8900 (ix86_expand_sse_cmp): Likewise.
8901 (emit_memmov): Likewise.
8902 (emit_memset): Likewise.
8903 (expand_small_movmem_or_setmem): Likewise.
8904 (ix86_expand_pextr): Likewise.
8905 (ix86_expand_pinsr): Likewise.
8906 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8907 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8909 * config/mips/mips.c (mips_function_value_1) Likewise.
8910 (mips_block_move_straight): Likewise.
8911 (mips_expand_ins_as_unaligned_store): Likewise.
8912 * config/powerpcspe/powerpcspe.c
8913 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8914 (rs6000_darwin64_record_arg_flush): Likewise.
8915 * config/rs6000/rs6000.c
8916 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8917 (rs6000_darwin64_record_arg_flush): Likewise.
8918 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8919 (sparc_function_value_1): Likewise.
8920 * config/spu/spu.c (adjust_operand): Likewise.
8921 (spu_emit_branch_or_set): Likewise.
8922 (arith_immediate_p): Likewise.
8923 * emit-rtl.c (gen_lowpart_common): Likewise.
8924 * expr.c (expand_expr_real_1): Likewise.
8925 * function.c (assign_parm_setup_block): Likewise.
8926 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8927 * reload1.c (alter_reg): Likewise.
8928 * stor-layout.c (mode_for_vector): Likewise.
8929 (layout_type): Likewise.
8931 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8933 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8934 (spu_convert_move): Likewise.
8935 * lower-subreg.c (resolve_simple_move): Likewise.
8937 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8940 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8941 define_insn to a define_expand.
8942 (altivec_vsum2sws_direct): New define_insn.
8943 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8945 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8947 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8950 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8955 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8956 realign stack if stack alignment needed is less than incoming
8959 2017-09-05 Marek Polacek <polacek@redhat.com>
8962 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8965 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8967 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8969 2017-09-05 Richard Biener <rguenther@suse.de>
8971 PR tree-optimization/82084
8972 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8974 2017-09-05 Richard Biener <rguenther@suse.de>
8976 PR tree-optimization/82102
8977 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8979 2017-09-05 Martin Liska <mliska@suse.cz>
8981 PR tree-optimization/82032
8982 * tree-cfg.c (generate_range_test): New function.
8983 * tree-cfg.h (generate_range_test): Declared here.
8984 * tree-cfgcleanup.c (convert_single_case_switch): New function.
8985 (cleanup_control_expr_graph): Use it.
8986 * tree-switch-conversion.c (try_switch_expansion): Remove
8988 (emit_case_nodes): Use generate_range_test.
8990 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8993 * config/i386/i386.md (*<btsc><mode>_mask): Add
8994 TARGET_USE_BT to insn constraint.
8995 (*btr<mode>_mask): Ditto.
8997 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
8999 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9000 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9002 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9005 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9006 TARGET_NEON and TARGET_IWMMXT.
9007 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9008 TARGET_NEON and TARGET_IWMMXT.
9009 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9011 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9013 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9014 (ix86_rewrite_tls_address): Ditto.
9015 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9016 (ix86_rewrite_tls_address_1): Ditto.
9017 (ix86_rewrite_tls_address): Ditto.
9018 * config/i386/predicates.md (tls_address_pattern): New predicate.
9019 * config/i386/i386.md (TLS address splitter): New splitter.
9021 2017-09-04 Richard Biener <rguenther@suse.de>
9023 PR tree-optimization/82084
9024 * fold-const.h (can_native_encode_string_p): Declare.
9025 * fold-const.c (can_native_encode_string_p): Factor out from ...
9026 (native_encode_string): ... here.
9027 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9028 vectorizing stores from constants we later cannot handle.
9030 2017-09-04 Marek Polacek <polacek@redhat.com>
9033 * doc/invoke.texi: Update -Wtautological-compare documentation.
9035 2017-09-04 Jeff Law <law@redhat.com>
9037 PR tree-optimization/64910
9038 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9039 swap the first and last operand if the last is a constant.
9041 2017-09-04 Marek Polacek <polacek@redhat.com>
9044 * convert.c (do_narrow): When sanitizing signed integer overflows,
9045 bail out for signed types.
9046 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9048 2017-09-04 Richard Biener <rguenther@suse.de>
9050 PR tree-optimization/82060
9051 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9052 Move devirtualization after stmt folding and before EH/AB/noreturn
9053 cleanup to get the stmt refs canonicalized. Use a bool instead
9054 of gimple_modified_p since that doesn't work for NOPs. Schedule
9055 NOPs generated by folding for removal.
9057 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9058 Alan Hayward <alan.hayward@arm.com>
9059 David Sherwood <david.sherwood@arm.com>
9061 * coretypes.h (pad_direction): New enum.
9062 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9063 (FUNCTION_ARG_PADDING): Likewise.
9064 * target.def (function_arg_padding): New hook.
9065 * targhooks.h (default_function_arg_padding): Declare.
9066 * targhooks.c (default_function_arg_padding): New function.
9067 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9068 (TARGET_FUNCTION_ARG_PADDING): ...this.
9069 * doc/tm.texi: Regenerate.
9070 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9071 instead of direction.
9072 (compute_argument_addresses): Likewise.
9073 (load_register_parameters): Likewise.
9074 (emit_library_call_value_1): Likewise.
9075 (store_one_arg): Use targetm.calls.function_arg_padding instead
9076 of FUNCTION_ARG_PADDING.
9077 (must_pass_in_stack_var_size_or_pad): Likewise.
9078 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9079 (emit_group_store): Likewise.
9080 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9081 instead of FUNCTION_ARG_PADDING.
9082 (emit_push_insn): Likewise, and propagate enum change throughout
9084 * function.h (direction): Delete.
9085 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9087 * function.c (assign_parm_find_stack_rtl): Likewise.
9088 (assign_parm_setup_block_p): Likewise.
9089 (assign_parm_setup_block): Likewise.
9090 (gimplify_parameters): Likewise.
9091 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9092 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9094 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9095 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9096 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9097 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9098 (aarch64_function_arg_padding): ...this new function.
9099 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9100 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9101 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9102 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9103 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9104 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9105 (arm_pad_arg_upward): Replace with...
9106 (arm_function_arg_padding): ...this new function.
9107 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9109 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9110 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9111 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9112 (ia64_hpux_function_arg_padding): Replace with...
9113 (ia64_function_arg_padding): ...this new function. Use pad_direction
9114 instead of direction. Check for TARGET_HPUX.
9115 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9116 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9117 (iq2000_function_arg_padding): New function.
9118 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9119 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9120 (mips_function_arg_padding): ...this new function.
9121 (mips_pad_reg_upward): Update accordingly.
9122 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9123 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9124 targetm.calls.function_arg_padding.
9125 (FUNCTION_ARG_PADDING): Delete.
9126 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9127 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9128 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9129 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9130 (nios2_block_reg_padding): Return pad_direction instead of direction.
9131 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9132 instead of direction.
9133 (nios2_function_arg_padding): Likewise. Make static.
9134 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9135 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9136 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9137 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9138 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9139 (pa_function_arg_padding): Make static. Return pad_direction instead
9141 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9142 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9143 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9144 instead of direction. Use targetm.calls.function_arg_padding.
9145 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9146 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9147 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9148 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9149 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9151 (function_arg_padding): Rename to...
9152 (rs6000_function_arg_padding): ...this. Make static. Return
9153 pad_direction instead of direction.
9154 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9155 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9156 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9157 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9158 instead of direction. Use targetm.calls.function_arg_padding.
9159 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9160 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9161 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9162 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9163 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9164 (function_arg_padding): Rename to...
9165 (rs6000_function_arg_padding): ...this. Make static. Return
9166 pad_direction instead of direction.
9167 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9168 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9169 * config/s390/s390.c (s390_function_arg_padding): New function.
9170 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9171 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9172 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
9173 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9174 (function_arg_padding): Rename to...
9175 (sparc_function_arg_padding): ...this. Make static. Return
9176 pad_direction instead of direction.
9177 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
9178 * config/spu/spu.c (spu_function_arg_padding): New function.
9179 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9180 * system.h (FUNCTION_ARG_PADDING): Poison.
9182 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9183 Alan Hayward <alan.hayward@arm.com>
9184 David Sherwood <david.sherwood@arm.com>
9186 * target.def (modes_tieable_p): New hook.
9187 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
9188 (TARGET_MODES_TIEABLE_P): ...this.
9189 * doc/tm.texi.in: Regenerate.
9190 * hooks.h (hook_bool_mode_mode_true): Declare.
9191 * hooks.c (hook_bool_mode_mode_true): New function.
9192 * combine.c (subst): Use targetm.modes_tieable_p instead of
9194 * dse.c (find_shift_sequence): Likewise.
9195 * expmed.c (extract_low_bits): Likewise.
9196 * lower-subreg.c: Include target.h.
9197 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
9199 * rtlanal.c (rtx_cost): Likewise.
9200 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
9201 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
9202 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
9203 (TARGET_MODES_TIEABLE_P): Redefine.
9204 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
9205 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
9206 (TARGET_MODES_TIEABLE_P): Redefine.
9207 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
9208 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
9209 (arc_modes_tieable_p): New function.
9210 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
9211 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
9212 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
9213 (arm_modes_tieable_p): Make static.
9214 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
9215 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
9216 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
9217 (TARGET_MODES_TIEABLE_P): Redefine.
9218 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
9219 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
9220 (TARGET_MODES_TIEABLE_P): Redefine.
9221 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
9222 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
9223 (cr16_modes_tieable_p): New function.
9224 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
9225 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
9226 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
9227 (TRULY_NOOP_TRUNCATION): Update comment.
9228 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
9229 (TRULY_NOOP_TRUNCATION): Update comment.
9230 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
9231 (frv_modes_tieable_p): New function.
9232 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
9233 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
9234 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
9235 (TARGET_MODES_TIEABLE_P): Redefine.
9236 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9237 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9238 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9239 (TARGET_MODES_TIEABLE_P): Redefine.
9240 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9241 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9242 (ia64_modes_tieable_p): New function.
9243 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9244 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9245 (iq2000_modes_tieable_p): New function.
9246 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9247 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9248 (lm32_modes_tieable_p): New function.
9249 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9250 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9251 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9252 (TARGET_MODES_TIEABLE_P): Redefine.
9253 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9254 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9255 (m32r_modes_tieable_p): New function.
9256 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9257 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9258 (m68k_modes_tieable_p): New function.
9259 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9260 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9261 (mcore_modes_tieable_p): New function.
9262 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9263 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9265 (TARGET_MODES_TIEABLE_P): Redefine.
9266 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9267 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9268 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9269 (TARGET_MODES_TIEABLE_P): Redefine.
9270 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9271 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9272 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9273 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9274 (mn10300_modes_tieable_p): ...this and make static.
9275 (TARGET_MODES_TIEABLE_P): Redefine.
9276 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9277 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9278 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9279 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9280 (msp430_modes_tieable_p): Make static.
9281 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9282 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9283 (TARGET_MODES_TIEABLE_P): Redefine.
9284 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9285 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9286 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9287 (TARGET_MODES_TIEABLE_P): Redefine.
9288 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9289 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9290 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9291 (TARGET_MODES_TIEABLE_P): Redefine.
9292 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9293 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9294 (pdp11_modes_tieable_p): New function.
9295 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9296 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9297 (rs6000_modes_tieable_p): New function.
9298 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9299 * config/powerpcspe/powerpcspe.md: Update comment.
9300 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9301 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9302 (TARGET_MODES_TIEABLE_P): Redefine.
9303 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9304 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9305 (rl78_modes_tieable_p): New function.
9306 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9307 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9308 (rs6000_modes_tieable_p): New function.
9309 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9310 * config/rs6000/rs6000.md: Update comment.
9311 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9312 * config/rx/rx.c (rx_modes_tieable_p): New function.
9313 (TARGET_MODES_TIEABLE_P): Redefine.
9314 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9315 * config/s390/s390.c (s390_modes_tieable_p): New function.
9316 (TARGET_MODES_TIEABLE_P): Redefine.
9317 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9318 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9319 (sh_modes_tieable_p): New function.
9320 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9321 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9322 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9323 (sparc_modes_tieable_p): Make static.
9324 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9325 * config/spu/spu.c (spu_modes_tieable_p): New function.
9326 (TARGET_MODES_TIEABLE_P): Redefine.
9327 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9328 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9329 (TARGET_MODES_TIEABLE_P): Redefine.
9330 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9331 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9332 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9333 * config/v850/v850.c (v850_modes_tieable_p): New function.
9334 (TARGET_MODES_TIEABLE_P): Redefine.
9335 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9336 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9337 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9338 (visium_modes_tieable_p): New function.
9339 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9340 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9341 (xtensa_modes_tieable_p): New function.
9342 * system.h (MODES_TIEABLE_P): Poison.
9344 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9345 Alan Hayward <alan.hayward@arm.com>
9346 David Sherwood <david.sherwood@arm.com>
9348 * target.def (hard_regno_mode_ok): New hook.
9349 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9350 (TARGET_HARD_REGNO_MODE_OK): ...this.
9351 * doc/tm.texi.in: Regenerate.
9352 * hooks.h (hook_bool_uint_mode_true): Declare.
9353 * hooks.c (hook_bool_uint_mode_true): New function.
9354 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9356 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9357 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9358 instead of HARD_REGNO_MODE_OK.
9359 * caller-save.c: Include target.h.
9360 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9362 * combine.c (can_combine_p): Likewise.
9363 (combinable_i3pat): Likewise.
9364 (can_change_dest_mode): Likewise.
9365 * expr.c (init_expr_target): Likewise.
9366 (convert_move): Likewise.
9367 (convert_modes): Likewise.
9368 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9369 (setup_prohibited_mode_move_regs): Likewise.
9370 * ira.h (target_ira): Likewise.
9371 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9372 * lra-constraints.c (process_alt_operands): Likewise.
9373 (split_reg): Likewise.
9374 * recog.c (peep2_find_free_register): Likewise.
9375 * ree.c (combine_reaching_defs): Likewise.
9376 * regcprop.c (maybe_mode_change): Likewise.
9377 * reginfo.c (init_reg_sets_1): Likewise.
9378 (choose_hard_reg_mode): Likewise.
9379 (simplifiable_subregs): Likewise.
9380 * regrename.c (check_new_reg_p): Likewise.
9381 * reload.c (find_valid_class): Likewise.
9382 (find_valid_class_1): Likewise.
9383 (reload_inner_reg_of_subreg): Likewise.
9384 (push_reload): Likewise.
9385 (combine_reloads): Likewise.
9386 (find_dummy_reload): Likewise.
9387 (find_reloads): Likewise.
9388 * reload1.c (find_reg): Likewise.
9389 (set_reload_reg): Likewise.
9390 (allocate_reload_reg): Likewise.
9391 (choose_reload_regs): Likewise.
9392 (reload_adjust_reg_for_temp): Likewise.
9393 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9394 (simplify_subreg_regno): Likewise.
9395 * sel-sched.c (init_regs_for_mode): Likewise.
9396 * varasm.c (make_decl_rtl): Likewise.
9397 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9398 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9400 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9401 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9402 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9403 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9404 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9405 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9406 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9407 (arc_mode_class): Delete.
9408 (HARD_REGNO_MODE_OK): Delete.
9409 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9410 (arc_hard_regno_mode_ok): Rename old array to...
9411 (arc_hard_regno_mode_ok_modes): ...this.
9412 (arc_conditional_register_usage): Update accordingly.
9413 (arc_mode_class): Make static.
9414 (arc_hard_regno_mode_ok): New function.
9415 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9416 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9417 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9418 (arm_hard_regno_mode_ok): Make static.
9419 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9421 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9422 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9423 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9425 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9426 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9427 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9428 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9429 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9430 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9431 * config/bfin/predicates.md (valid_reg_operand): Use
9432 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9433 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9434 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9435 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9436 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9437 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9438 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9439 (cr16_hard_regno_mode_ok): Make static and return a bool.
9440 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9441 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9442 (cris_hard_regno_mode_ok): New function.
9443 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9444 (epiphany_mode_class): Delete.
9445 (HARD_REGNO_MODE_OK): Delete.
9446 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9447 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9448 (hard_regno_mode_ok): Rename to...
9449 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9450 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9451 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9453 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9454 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9455 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9456 (frv_hard_regno_mode_ok): Make static and return a bool.
9457 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9459 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9460 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9461 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9462 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9464 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9465 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9466 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9467 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9469 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9470 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9471 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9472 (ia64_hard_regno_mode_ok): New function.
9473 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9474 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9475 (iq2000_hard_regno_mode_ok): New function.
9476 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9477 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9478 (lm32_hard_regno_mode_ok): New function.
9479 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9480 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9481 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9482 instead of HARD_REGNO_MODE_OK.
9483 (m32c_hard_regno_ok): Rename to...
9484 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9485 (m32c_cannot_change_mode_class): Update accordingly.
9486 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9487 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9488 (m32r_mode_class): Delete.
9489 (HARD_REGNO_MODE_OK): Delete.
9490 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9491 (m32r_hard_regno_mode_ok): Rename to...
9492 (m32r_hard_regno_modes): ...this.
9493 (m32r_mode_class): Make static.
9494 (m32r_hard_regno_mode_ok): New function.
9495 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9496 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9497 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9498 (m68k_hard_regno_mode_ok): Make static.
9499 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9500 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9501 (mcore_hard_regno_mode_ok): New function.
9502 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9503 (HARD_REGNO_MODE_OK): Delete.
9504 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9506 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9507 (microblaze_hard_regno_mode_ok): New function.
9508 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9509 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9510 (mips_hard_regno_mode_ok): Delete.
9511 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9512 (mips_hard_regno_mode_ok_p): ...this and make static.
9513 (mips_hard_regno_mode_ok_p): Rename to...
9514 (mips_hard_regno_mode_ok_uncached): ...this.
9515 (mips_hard_regno_mode_ok): New function.
9516 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9517 of HARD_REGNO_MODE_OK.
9518 (mips_option_override): Update after above name changes.
9519 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9520 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9521 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9522 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9523 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9524 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9525 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9526 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9527 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9528 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9529 (msp430_hard_regno_mode_ok): Make static and return a bool.
9530 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9531 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9532 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9534 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9535 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9536 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9537 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9538 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9539 (PA_HARD_REGNO_MODE_OK): ...this
9540 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9541 (PA_HARD_REGNO_MODE_OK): ...this.
9542 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9543 (pa_hard_regno_mode_ok): New function.
9544 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9545 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9546 (pdp11_hard_regno_mode_ok): New function.
9547 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9548 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9550 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9552 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9553 (rs6000_hard_regno_mode_ok): Rename to...
9554 (rs6000_hard_regno_mode_ok_uncached): ...this.
9555 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9556 (rs6000_hard_regno_mode_ok): New function.
9557 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9558 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9559 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9560 (riscv_hard_regno_mode_ok): ...this and make static.
9561 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9562 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9563 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9564 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9565 (rl78_hard_regno_mode_ok): Make static and return bool.
9566 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9567 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9569 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9570 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9571 (rs6000_hard_regno_mode_ok): Rename to...
9572 (rs6000_hard_regno_mode_ok_uncached): ...this.
9573 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9574 (rs6000_hard_regno_mode_ok): New function.
9575 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9576 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9577 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9578 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9579 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9580 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9581 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9582 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9583 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9584 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9585 (sh_hard_regno_mode_ok): Make static.
9586 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9587 instead of HARD_REGNO_MODE_OK.
9588 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9589 (sparc_mode_class): Delete.
9590 (HARD_REGNO_MODE_OK): Delete.
9591 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9592 (hard_regno_mode_classes): Make static.
9593 (sparc_mode_class): Likewise.
9594 (sparc_hard_regno_mode_ok): New function.
9595 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9596 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9597 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9599 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9600 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9601 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9602 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9603 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9604 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9605 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9606 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9607 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9608 (visium_hard_regno_mode_ok): New function.
9609 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9610 instead of HARD_REGNO_MODE_OK.
9611 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9612 (HARD_REGNO_MODE_OK): Delete.
9613 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9614 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9615 (xtensa_option_override): Update accordingly.
9616 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9617 (xtensa_hard_regno_mode_ok): New function.
9618 * system.h (HARD_REGNO_MODE_OK): Poison.
9620 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9621 Alan Hayward <alan.hayward@arm.com>
9622 David Sherwood <david.sherwood@arm.com>
9624 * target.def (hard_regno_call_part_clobbered): New hook.
9625 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9626 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9627 * doc/tm.texi: Regenerate.
9628 * hooks.h (hook_bool_uint_mode_false): Declare.
9629 * hooks.c (hook_bool_uint_mode_false): New function.
9630 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9631 * cselib.c (cselib_process_insn): Use
9632 targetm.hard_regno_call_part_clobbered instead of
9633 HARD_REGNO_CALL_PART_CLOBBERED.
9634 * ira-conflicts.c (ira_build_conflicts): Likewise.
9635 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9636 * lra-constraints.c (need_for_call_save_p): Likewise.
9637 * lra-lives.c: Include target.h.
9638 (check_pseudos_live_through_calls): Use
9639 targetm.hard_regno_call_part_clobbered instead of
9640 HARD_REGNO_CALL_PART_CLOBBERED.
9641 * regcprop.c: Include target.h.
9642 (copyprop_hardreg_forward_1): Use
9643 targetm.hard_regno_call_part_clobbered instead of
9644 HARD_REGNO_CALL_PART_CLOBBERED.
9645 * reginfo.c (choose_hard_reg_mode): Likewise.
9646 * regrename.c (check_new_reg_p): Likewise.
9647 * reload.c (find_equiv_reg): Likewise.
9648 * reload1.c (emit_reload_insns): Likewise.
9649 * sched-deps.c (deps_analyze_insn): Likewise.
9650 * sel-sched.c (init_regs_for_mode): Likewise.
9651 (mark_unavailable_hard_regs): Likewise.
9652 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9653 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9654 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9656 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9657 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9658 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9660 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9662 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9663 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9664 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9666 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9667 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9668 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9670 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9671 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
9673 * config/powerpcspe/powerpcspe.c
9674 (rs6000_hard_regno_call_part_clobbered): New function.
9675 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9676 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9677 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
9679 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9680 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9681 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
9683 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9684 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9685 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
9687 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9688 Alan Hayward <alan.hayward@arm.com>
9689 David Sherwood <david.sherwood@arm.com>
9691 * rtl.h (subreg_memory_offset): Declare.
9692 * emit-rtl.c (subreg_memory_offset): New function.
9693 * expmed.c (store_bit_field_1): Use it.
9694 * expr.c (undefined_operand_subword_p): Likewise.
9695 * simplify-rtx.c (simplify_subreg): Likewise.
9697 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9699 PR rtl-optimization/57448
9702 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9703 using atomic_load pattern.
9704 (expand_atomic_store): Likewise.
9706 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9709 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9710 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9711 handling. Use replace_call_with_value with NULL instead of
9712 gsi_replace, unlink_stmt_vdef and release_defs.
9714 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9717 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9719 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9722 * rtl.h (emit_library_call_value_1): Declare.
9723 (emit_library_call): Replace declaration with a series of overloads.
9724 Remove the parameter count argument.
9725 (emit_library_call_value): Likewise.
9726 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9727 with an "rtx_mode_t *".
9728 (emit_library_call_value): Delete.
9729 (emit_library_call): Likewise.
9730 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9731 (asan_emit_allocas_unpoison): Likewise.
9732 * builtins.c (expand_builtin_powi): Likewise.
9733 (expand_asan_emit_allocas_unpoison): Likewise.
9734 * cfgexpand.c (expand_main_function): Likewise.
9735 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9736 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9737 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9738 * config/arm/arm.c (arm_trampoline_init): Likewise.
9739 (arm_call_tls_get_addr): Likewise.
9740 (arm_expand_divmod_libfunc): Likewise.
9741 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9742 (smulsi3_highpart): Likewise.
9743 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9744 (c6x_expand_compare): Likewise.
9745 (c6x_expand_movmem): Likewise.
9746 * config/frv/frv.c (frv_trampoline_init): Likewise.
9747 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9748 (ix86_expand_divmod_libfunc): Likewise.
9749 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9750 (ia64_expand_compare): Likewise.
9751 (ia64_profile_hook): Likewise.
9752 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9753 (nonlocal_goto): Likewise.
9754 (restore_stack_nonlocal): Likewise.
9755 * config/m32r/m32r.c (block_move_call): Likewise.
9756 (m32r_trampoline_init): Likewise.
9757 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9758 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9759 (m68k_call_m68k_read_tp): Likewise.
9760 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9761 (microblaze_expand_divide): Likewise.
9762 * config/mips/mips.h (mips_args): Likewise.
9763 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9764 (MIPS_ICACHE_SYNC): Likewise.
9765 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9766 (nios2_trampoline_init): Likewise.
9767 * config/pa/pa.c (hppa_tls_call): Likewise.
9768 (pa_trampoline_init): Likewise.
9769 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9770 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9771 (expand_strn_compare): Likewise.
9772 (rs6000_generate_compare): Likewise.
9773 (rs6000_expand_float128_convert): Likewise.
9774 (output_profile_hook): Likewise.
9775 (rs6000_trampoline_init): Likewise.
9776 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9777 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9778 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9779 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9780 (rs6000_generate_compare): Likewise.
9781 (rs6000_expand_float128_convert): Likewise.
9782 (output_profile_hook): Likewise.
9783 (rs6000_trampoline_init): Likewise.
9784 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9785 * config/sh/sh.c (sh_trampoline_init): Likewise.
9786 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9787 (sparc_emit_float_lib_cmp): Likewise.
9788 (sparc32_initialize_trampoline): Likewise.
9789 (sparc64_initialize_trampoline): Likewise.
9790 (sparc_profile_hook): Likewise.
9791 * config/spu/spu.c (ea_load_store): Likewise.
9792 * config/spu/spu.md (floatunssidf2): Likewise.
9793 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9794 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9795 * config/visium/visium.c (expand_block_move_4): Likewise.
9796 (expand_block_move_2): Likewise.
9797 (expand_block_move_1): Likewise.
9798 (expand_block_set_4): Likewise.
9799 (expand_block_set_2): Likewise.
9800 (expand_block_set_1): Likewise.
9801 (visium_trampoline_init): Likewise.
9802 (visium_profile_hook): Likewise.
9803 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9804 (xtensa_setup_frame_addresses): Likewise.
9805 (xtensa_trampoline_init): Likewise.
9806 * except.c (sjlj_emit_function_enter): Likewise.
9807 (sjlj_emit_function_exit): Likewise.
9808 * explow.c (allocate_dynamic_stack_space): Likewise.
9809 (probe_stack_range): Likewise.
9810 * expr.c (convert_mode_scalar): Likewise.
9811 * optabs.c (expand_binop): Likewise.
9812 (expand_twoval_binop_libfunc): Likewise.
9813 (expand_unop): Likewise.
9814 (prepare_cmp_insn): Likewise.
9815 (prepare_float_lib_cmp): Likewise.
9816 (expand_float): Likewise.
9817 (expand_fix): Likewise.
9818 (expand_fixed_convert): Likewise.
9819 (maybe_emit_sync_lock_test_and_set): Likewise.
9820 (expand_atomic_compare_and_swap): Likewise.
9821 (expand_mem_thread_fence): Likewise.
9822 (expand_atomic_fetch_op): Likewise.
9824 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9826 * doc/generic.texi (OpenACC): Adjust URL.
9827 * doc/invoke.texi (C Dialect Options): Ditto.
9829 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9831 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9832 predicate for operand 1. Add (m,<S>) constraint.
9833 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9834 Prevent memory operand 1 with register operand 2.
9836 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9838 PR rtl-optimization/82024
9839 * combine.c (try_combine): If the combination result is a PARALLEL,
9840 and we only need to retain the SET in there that would be placed
9841 at I2, check that we can place that at I3 instead, before doing so.
9843 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9846 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9848 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9849 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9852 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9853 Jeff Law <law@redhat.com>
9855 * varasm.c (bss_initializer_p): Do not put constants into .bss
9856 (categorize_decl_for_section): Handle bss_initializer_p returning
9857 false when DECL_INITIAL is NULL.
9859 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9862 * config/s390/s390.c (s390_can_inline_p): New function.
9864 2017-09-01 Jeff Law <law@redhat.com>
9866 PR tree-optimization/82052
9867 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9868 Always initialize the returned slot after a hash table miss
9869 when INSERT is true.
9871 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9873 * config/s390/s390.md (mem_signal_fence): Remove.
9874 * doc/md.texi (mem_signal_fence): Remove.
9875 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9877 * target-insns.def (mem_signal_fence): Remove.
9879 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9882 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9885 * asan.c (create_odr_indicator): Strip name encoding from assembler
9886 name before appending it after __odr_asan_.
9888 2017-09-01 Martin Liska <mliska@suse.cz>
9890 PR tree-optimization/82059
9891 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9892 frequency only when an edge is redirected.
9894 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9896 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9897 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9898 (arc_conditional_register_usage): Remove ARC600 lp_count
9900 (arc_file_start): Emit Tag_ARC_CPU_variation.
9901 (arc_can_use_doloop_p): New conditions to use ZOLs.
9902 (hwloop_fail): New function.
9903 (hwloop_optimize): Likewise.
9904 (hwloop_pattern_reg): Likewise.
9905 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9906 (arc_reorg_loops): New function, calls reorg_loops.
9907 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9908 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9910 (arc_loop_hazard): Remove function, functionality moved into
9912 (arc_hazard): Remove arc_loop_hazard call.
9913 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9914 into hwloop_optimize.
9915 (arc_label_align): Remove ZOL handling.
9916 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9917 * config/arc/arc.md (doloop_begin): Remove pattern.
9918 (doloop_begin_i): Likewise.
9919 (doloop_end_i): Likewise.
9920 (doloop_fallback): Likewise.
9921 (doloop_fallback_m): Likewise.
9922 (doloop_end): Reimplement expand.
9923 (arc_lp): New pattern for LP instruction.
9924 (loop_end): New pattern.
9925 (loop_fail): Likewise.
9926 (decrement_and_branch_until_zero): Likewise.
9927 * config/arc/arc.opt (mlpc-width): New option.
9928 * doc/invoke.texi (mlpc-width): Document option.
9930 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9932 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9933 (arc_ccfsm_advance): Fix checking for delay slots.
9934 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9936 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9938 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9940 (movhi_insn): Update store instruction constraint which are saving
9941 6-bit short immediates.
9942 (movsi_insn): Consider also short scaled load operations.
9943 (zero_extendhisi2_i): Use Usd constraint instead of T.
9944 (extendhisi2_i): Add q constraint.
9945 (arc_clzsi2): Add type and length attributes.
9946 (arc_ctzsi2): Likewise.
9947 * config/arc/constraints.md (Usc): Update constraint, the
9948 assembler can parse two relocations for a single instruction.
9950 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9952 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9953 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9955 2017-08-31 Olivier Hainque <hainque@adacore.com>
9957 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9958 match as powerpc-wrs-vxworks*.
9960 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9962 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9963 register constraint for by-element operand.
9964 (aarch64_mls_elt_merge<mode>): Likewise.
9966 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9968 * config/arc/arc.c (arc_can_follow_jump): Check for short
9971 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9973 * config.gcc: Use g.opt for arc.
9974 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9975 functionality moved to ...
9976 (legitimate_scaled_address_p): New function, ...here.
9977 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9978 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9979 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9981 (arc_override_options): Handle G option.
9982 (arc_output_pic_addr_const): Correct function definition.
9983 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9984 (arc_decl_anon_ns_mem_p): Delete.
9985 (arc_in_small_data_p): Overhaul this function to take into
9986 consideration the value given via G option.
9987 (arc_rewrite_small_data_1): Renamed and corrected old
9988 arc_rewrite_small_data function.
9989 (arc_rewrite_small_data): New function.
9990 (small_data_pattern): Don't use pic_offset_table_rtx.
9991 * config/arc/arc.h (CC1_SPEC): Recognize G option.
9992 * config/arc/simdext.md (movmisalignv2hi): Use
9993 prepare_move_operands function.
9995 (movmisalign*): Likewise.
9996 * doc/invoke.texi (ARC options): Document -G option.
9998 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10000 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10002 * config/arc/arc.c (arc_print_operand): Output scalled address for
10003 sdata whenever is possible.
10004 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10005 load/stores are available.
10006 (compact_sda_memory_operand): Check for the alignment required by
10007 code density instructions.
10008 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10010 * config/arc/constraints.md (Usd): Update constraint.
10011 (Us0): New constraint.
10012 (Usc): Update constraint.
10014 2017-08-31 Richard Biener <rguenther@suse.de>
10016 PR middle-end/82054
10017 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10018 function only once.
10020 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10022 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10023 Resize type_signature.
10025 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10026 Alan Hayward <alan.hayward@arm.com>
10027 David Sherwood <david.sherwood@arm.com>
10029 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10030 subregs whose inner modes can be stored in GPRs.
10031 (aarch64_classify_index): Likewise.
10033 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10034 Alan Hayward <alan.hayward@arm.com>
10035 David Sherwood <david.sherwood@arm.com>
10037 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10038 (V_INT_EQUIV): ...this.
10039 (v_cmp_result): Rename to...
10040 (v_int_equiv): ...this.
10041 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10042 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10043 (copysign<mode>3): Likewise.
10044 (aarch64_simd_bsl<mode>_internal): Likewise.
10045 (aarch64_simd_bsl<mode>): Likewise.
10046 (vec_cmp<mode><mode>): Likewise.
10047 (vcond<mode><mode>): Likewise.
10048 (vcond<v_cmp_mixed><mode>): Likewise.
10049 (vcondu<mode><v_cmp_mixed>): Likewise.
10050 (aarch64_cm<optab><mode>): Likewise.
10051 (aarch64_cmtst<mode>): Likewise.
10052 (aarch64_fac<optab><mode>): Likewise.
10053 (vec_perm_const<mode>): Likewise.
10054 (vcond_mask_<mode><v_cmp_result>): Rename to...
10055 (vcond_mask_<mode><v_int_equiv>): ...this.
10056 (vec_cmp<mode><v_cmp_result>): Rename to...
10057 (vec_cmp<mode><v_int_equiv>): ...this.
10059 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10060 Alan Hayward <alan.hayward@arm.com>
10061 David Sherwood <david.sherwood@arm.com>
10063 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
10065 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10066 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10067 (UNSPEC_LD4_DREG): New unspecs.
10068 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10069 (aarch64_ld2<mode>_dreg_be): Replace with...
10070 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10072 (aarch64_ld3<mode>_dreg_le)
10073 (aarch64_ld3<mode>_dreg_be): Replace with...
10074 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10076 (aarch64_ld4<mode>_dreg_le)
10077 (aarch64_ld4<mode>_dreg_be): Replace with...
10078 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10081 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10083 PR tree-optimization/81987
10084 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10085 insert an initializer in a location not dominated by the stride
10088 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10090 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10091 on the entire header of the finally block in the fallthru case.
10093 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10095 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10097 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10099 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10100 rs6000_emit_move_from_cr and call renamed function.
10101 (rs6000_emit_prologue): Call renamed functions.
10102 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10103 movesi_from_cr, remove volatile CRs.
10105 2017-08-30 Jon Beniston <jon@beniston.com>
10106 Richard Biener <rguenther@suse.de>
10108 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10109 of VECTOR_MODE_P check.
10110 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10111 element vector types.
10113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10115 * df.h (df_read_modify_subreg_p): Remove in favor of...
10116 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10117 const_rtx instead of an rtx.
10118 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10119 * df-problems.c (df_word_lr_mark_ref): Likewise.
10120 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10121 (mark_pseudo_reg_dead): Likewise.
10122 (mark_ref_dead): Likewise.
10123 * reginfo.c (init_subregs_of_mode): Likewise.
10124 * sched-deps.c (sched_analyze_1): Likewise.
10125 * df-scan.c (df_def_record_1): Likewise.
10126 (df_uses_record): Likewise.
10127 (df_read_modify_subreg_p): Remove in favor of...
10128 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10129 const_rtx instead of an rtx.
10131 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10132 Alan Hayward <alan.hayward@arm.com>
10133 David Sherwood <david.sherwood@arm.com>
10135 * rtl.h (partial_subreg_p): New function.
10136 * caller-save.c (save_call_clobbered_regs): Use it.
10137 * calls.c (expand_call): Likewise.
10138 * combine.c (combinable_i3pat): Likewise.
10139 (simplify_set): Likewise.
10140 (make_extraction): Likewise.
10141 (make_compound_operation_int): Likewise.
10142 (gen_lowpart_or_truncate): Likewise.
10143 (force_to_mode): Likewise.
10144 (make_field_assignment): Likewise.
10145 (reg_truncated_to_mode): Likewise.
10146 (record_truncated_value): Likewise.
10147 (move_deaths): Likewise.
10148 * cse.c (record_jump_cond): Likewise.
10149 (cse_insn): Likewise.
10150 * cselib.c (cselib_lookup_1): Likewise.
10151 * expmed.c (extract_bit_field_using_extv): Likewise.
10152 * function.c (assign_parm_setup_reg): Likewise.
10153 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10154 * ira-build.c (create_insn_allocnos): Likewise.
10155 * lra-coalesce.c (merge_pseudos): Likewise.
10156 * lra-constraints.c (match_reload): Likewise.
10157 (simplify_operand_subreg): Likewise.
10158 (curr_insn_transform): Likewise.
10159 * lra-lives.c (process_bb_lives): Likewise.
10160 * lra.c (new_insn_reg): Likewise.
10161 (lra_substitute_pseudo): Likewise.
10162 * regcprop.c (mode_change_ok): Likewise.
10163 (maybe_mode_change): Likewise.
10164 (copyprop_hardreg_forward_1): Likewise.
10165 * reload.c (push_reload): Likewise.
10166 (find_reloads): Likewise.
10167 (find_reloads_subreg_address): Likewise.
10168 * reload1.c (alter_reg): Likewise.
10169 (eliminate_regs_1): Likewise.
10170 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10172 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
10174 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
10175 back to if statements, including unpack.
10177 2017-08-30 Martin Liska <mliska@suse.cz>
10179 PR inline-asm/82001
10180 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
10182 (func_checker::compare_asm_inputs_outputs): ... this function.
10183 (func_checker::compare_gimple_asm): Use the function to compare
10184 also ASM constrains.
10185 * ipa-icf-gimple.h: Rename the function.
10187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10188 Alan Hayward <alan.hayward@arm.com>
10189 David Sherwood <david.sherwood@arm.com>
10191 * coretypes.h (complex_mode): New type.
10192 * gdbhooks.py (build_pretty_printer): Handle it.
10193 * machmode.h (complex_mode): New class.
10194 (complex_mode::includes_p): New function.
10195 (is_complex_int_mode): Likewise.
10196 (is_complex_float_mode): Likewise.
10197 * genmodes.c (get_mode_class): Handle complex mode classes.
10198 * function.c (expand_function_end): Use is_complex_int_mode.
10200 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10201 Alan Hayward <alan.hayward@arm.com>
10202 David Sherwood <david.sherwood@arm.com>
10204 * coretypes.h (scalar_mode_pod): New typedef.
10205 * gdbhooks.py (build_pretty_printer): Handle it.
10206 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
10207 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
10208 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
10209 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
10210 as_a <scalar_mode>.
10212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10213 Alan Hayward <alan.hayward@arm.com>
10214 David Sherwood <david.sherwood@arm.com>
10216 * machmode.h (mode_for_vector): Take a scalar_mode instead
10218 * stor-layout.c (mode_for_vector): Likewise.
10219 * explow.c (promote_mode): Use as_a <scalar_mode>.
10220 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
10222 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10223 Alan Hayward <alan.hayward@arm.com>
10224 David Sherwood <david.sherwood@arm.com>
10226 * target.def (preferred_simd_mode): Take a scalar_mode
10227 instead of a machine_mode.
10228 * targhooks.h (default_preferred_simd_mode): Likewise.
10229 * targhooks.c (default_preferred_simd_mode): Likewise.
10230 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
10231 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
10232 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
10233 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
10234 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
10235 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10236 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10237 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10239 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10240 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10241 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10242 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10243 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10244 * doc/tm.texi: Regenerate.
10245 * optabs-query.c (can_vec_mask_load_store_p): Return false for
10248 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10249 Alan Hayward <alan.hayward@arm.com>
10250 David Sherwood <david.sherwood@arm.com>
10252 * target.def (scalar_mode_supported_p): Take a scalar_mode
10253 instead of a machine_mode.
10254 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10255 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10256 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10257 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10258 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10259 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10260 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10261 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10262 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10263 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10264 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10265 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10266 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10267 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10269 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10270 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10271 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10272 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10273 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10275 * doc/tm.texi: Regenerate.
10277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10278 Alan Hayward <alan.hayward@arm.com>
10279 David Sherwood <david.sherwood@arm.com>
10281 * coretypes.h (opt_scalar_mode): New typedef.
10282 * gdbhooks.py (build_pretty_printers): Handle it.
10283 * machmode.h (mode_iterator::get_2xwider): Add overload for
10285 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10287 * expr.c (convert_mode_scalar): Likewise.
10288 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10289 * optabs.c (expand_float): Likewise.
10290 (expand_fix): Likewise.
10291 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10293 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10294 Alan Hayward <alan.hayward@arm.com>
10295 David Sherwood <david.sherwood@arm.com>
10297 * optabs.c (expand_float): Explicitly check for scalars before
10298 using a branching expansion.
10299 (expand_fix): Likewise.
10301 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10302 Alan Hayward <alan.hayward@arm.com>
10303 David Sherwood <david.sherwood@arm.com>
10305 * expr.c (convert_mode): Split scalar handling out into...
10306 (convert_mode_scalar): ...this new function. Treat the modes
10309 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10310 Alan Hayward <alan.hayward@arm.com>
10311 David Sherwood <david.sherwood@arm.com>
10313 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10315 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10318 Alan Hayward <alan.hayward@arm.com>
10319 David Sherwood <david.sherwood@arm.com>
10321 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10322 rather than a machine_mode.
10323 (fixed_from_string): Likewise.
10324 (fixed_convert): Likewise.
10325 (fixed_convert_from_int): Likewise.
10326 (fixed_convert_from_real): Likewise.
10327 (real_convert_from_fixed): Likewise.
10328 * fixed-value.c (fixed_from_double_int): Likewise.
10329 (fixed_from_string): Likewise.
10330 (fixed_convert): Likewise.
10331 (fixed_convert_from_int): Likewise.
10332 (fixed_convert_from_real): Likewise.
10333 (real_convert_from_fixed): Likewise.
10334 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10337 Alan Hayward <alan.hayward@arm.com>
10338 David Sherwood <david.sherwood@arm.com>
10340 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10341 of separate mode class checks. Do not allow vector modes here.
10342 (immed_wide_int_const): Use as_a <scalar_mode>.
10343 * explow.c (trunc_int_for_mode): Likewise.
10344 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10345 (wi::shwi): Likewise.
10346 (wi::min_value): Likewise.
10347 (wi::max_value): Likewise.
10348 * dwarf2out.c (loc_descriptor): Likewise.
10349 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10350 for CONST_WIDE_INT.
10352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10353 Alan Hayward <alan.hayward@arm.com>
10354 David Sherwood <david.sherwood@arm.com>
10356 * tree.h (SCALAR_TYPE_MODE): New macro.
10357 * expr.c (expand_expr_addr_expr_1): Use it.
10358 (expand_expr_real_2): Likewise.
10359 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10360 (fold_convert_const_fixed_from_int): Likewise.
10361 (fold_convert_const_fixed_from_real): Likewise.
10362 (native_encode_fixed): Likewise
10363 (native_encode_complex): Likewise
10364 (native_encode_vector): Likewise.
10365 (native_interpret_fixed): Likewise.
10366 (native_interpret_real): Likewise.
10367 (native_interpret_complex): Likewise.
10368 (native_interpret_vector): Likewise.
10369 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10370 (simd_clone_adjust_argument_types): Likewise.
10371 (simd_clone_init_simd_arrays): Likewise.
10372 (simd_clone_adjust): Likewise.
10373 * stor-layout.c (layout_type): Likewise.
10374 * tree.c (build_minus_one_cst): Likewise.
10375 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10376 * tree-inline.c (estimate_move_cost): Likewise.
10377 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10378 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10379 (vectorizable_reduction): Likewise.
10380 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10381 (vect_recog_mixed_size_cond_pattern): Likewise.
10382 (check_bool_pattern): Likewise.
10383 (adjust_bool_pattern): Likewise.
10384 (search_type_for_mask_1): Likewise.
10385 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10386 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10387 (vectorizable_load): Likewise.
10388 (vectorizable_store): Likewise.
10389 * ubsan.c (ubsan_encode_value): Likewise.
10390 * varasm.c (output_constant): Likewise.
10392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10393 Alan Hayward <alan.hayward@arm.com>
10394 David Sherwood <david.sherwood@arm.com>
10396 * coretypes.h (scalar_mode): New class.
10397 * machmode.h (scalar_mode): Likewise.
10398 (scalar_mode::includes_p): New function.
10399 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10400 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10401 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10402 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10403 * expmed.c (store_bit_field_1): Likewise.
10404 (extract_bit_field_1): Likewise.
10405 * expr.c (write_complex_part): Likewise.
10406 (read_complex_part): Likewise.
10407 (emit_move_complex_push): Likewise.
10408 (expand_expr_real_2): Likewise.
10409 * function.c (assign_parm_setup_reg): Likewise.
10410 (assign_parms_unsplit_complex): Likewise.
10411 * optabs.c (expand_binop): Likewise.
10412 * rtlanal.c (subreg_get_info): Likewise.
10413 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10414 * varasm.c (output_constant_pool_2): Likewise.
10416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10417 Alan Hayward <alan.hayward@arm.com>
10418 David Sherwood <david.sherwood@arm.com>
10420 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10422 (expmed_mult_highpart_optab): Likewise.
10423 (expmed_mult_highpart): Likewise.
10425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10426 Alan Hayward <alan.hayward@arm.com>
10427 David Sherwood <david.sherwood@arm.com>
10429 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10430 instead of a machine_mode.
10431 (builtin_memset_read_str): Likewise.
10432 * builtins.c (c_readstr): Likewise.
10433 (builtin_memcpy_read_str): Likewise.
10434 (builtin_strncpy_read_str): Likewise.
10435 (builtin_memset_read_str): Likewise.
10436 (builtin_memset_gen_str): Likewise.
10437 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10438 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10439 instead of a machine_mode.
10440 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10442 (make_extraction): Likewise.
10443 (try_widen_shift_mode): Take and return scalar_int_modes instead
10445 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10446 a scalar_int_mode instead of a machine_mode.
10447 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10448 (avr_addr_space_pointer_mode): Likewise.
10449 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10450 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10451 (msp430_unwind_word_mode): Likewise.
10452 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10453 (spu_addr_space_pointer_mode): Likewise.
10454 (spu_addr_space_address_mode): Likewise.
10455 (spu_libgcc_cmp_return_mode): Likewise.
10456 (spu_libgcc_shift_count_mode): Likewise.
10457 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10458 (rl78_addr_space_pointer_mode): Likewise.
10459 (fl78_unwind_word_mode): Likewise.
10460 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10462 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10463 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10464 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10465 (mips_valid_pointer_mode): Likewise.
10466 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10467 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10468 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10470 (ft32_addr_space_address_mode): Likewise.
10471 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10472 scalar_int_mode instead of a machine_mode.
10473 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10475 (m32c_addr_space_address_mode): Likewise.
10476 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10477 (rs6000_eh_return_filter_mode): Likewise.
10478 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10479 (rs6000_eh_return_filter_mode): Likewise.
10480 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10481 (s390_libgcc_shift_count_mode): Likewise.
10482 (s390_unwind_word_mode): Likewise.
10483 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10485 * target.def (mode_rep_extended): Likewise.
10486 (valid_pointer_mode): Likewise.
10487 (addr_space.valid_pointer_mode): Likewise.
10488 (eh_return_filter_mode): Return a scalar_int_mode rather than
10490 (libgcc_cmp_return_mode): Likewise.
10491 (libgcc_shift_count_mode): Likewise.
10492 (unwind_word_mode): Likewise.
10493 (addr_space.pointer_mode): Likewise.
10494 (addr_space.address_mode): Likewise.
10495 * doc/tm.texi: Regenerate.
10496 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10498 (do_jump): Use scalar_int_mode for local variables.
10499 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10500 rather than a machine_mode.
10501 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10502 (scompare_loc_descriptor_wide): Likewise.
10503 (scompare_loc_descriptor_narrow): Likewise.
10504 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10506 * except.c (sjlj_emit_dispatch_table): Likewise.
10507 (expand_builtin_eh_copy_values): Likewise.
10508 * explow.c (convert_memory_address_addr_space_1): Likewise.
10509 Take a scalar_int_mode rather than a machine_mode.
10510 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10511 than a machine_mode.
10512 (memory_address_addr_space): Use scalar_int_mode for local variables.
10513 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10514 rather than a machine_mode.
10515 * expmed.c (mask_rtx): Likewise.
10516 (init_expmed_one_conv): Likewise.
10517 (expand_mult_highpart_adjust): Likewise.
10518 (extract_high_half): Likewise.
10519 (expmed_mult_highpart_optab): Likewise.
10520 (expmed_mult_highpart): Likewise.
10521 (expand_smod_pow2): Likewise.
10522 (expand_sdiv_pow2): Likewise.
10523 (emit_store_flag_int): Likewise.
10524 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10526 (extract_low_bits): Likewise.
10527 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10529 * expr.c (pieces_addr::adjust): Likewise.
10530 (can_store_by_pieces): Likewise.
10531 (store_by_pieces): Likewise.
10532 (clear_by_pieces_1): Likewise.
10533 (expand_expr_addr_expr_1): Likewise.
10534 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10535 (expand_expr_real_1): Likewise.
10536 (try_casesi): Likewise.
10537 * final.c (shorten_branches): Likewise.
10538 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10539 type of "mode" to machine_mode.
10540 * internal-fn.c (expand_arith_overflow_result_store): Take a
10541 scalar_int_mode rather than a machine_mode.
10542 (expand_mul_overflow): Use scalar_int_mode for local variables.
10543 * loop-doloop.c (doloop_modify): Likewise.
10544 (doloop_optimize): Likewise.
10545 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10546 than a machine_mode.
10547 (expand_doubleword_shift_condmove): Likewise.
10548 (expand_doubleword_shift): Likewise.
10549 (expand_doubleword_clz): Likewise.
10550 (expand_doubleword_popcount): Likewise.
10551 (expand_doubleword_parity): Likewise.
10552 (expand_absneg_bit): Use scalar_int_mode for local variables.
10553 (prepare_float_lib_cmp): Likewise.
10554 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10555 rather than a machine_mode.
10556 (convert_memory_address_addr_space): Likewise.
10557 (get_mode_bounds): Likewise.
10558 (get_address_mode): Return a scalar_int_mode rather than a
10560 * rtlanal.c (get_address_mode): Likewise.
10561 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10562 than a machine_mode.
10563 * targhooks.c (default_mode_rep_extended): Likewise.
10564 (default_valid_pointer_mode): Likewise.
10565 (default_addr_space_valid_pointer_mode): Likewise.
10566 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10567 than a machine_mode.
10568 (default_libgcc_cmp_return_mode): Likewise.
10569 (default_libgcc_shift_count_mode): Likewise.
10570 (default_unwind_word_mode): Likewise.
10571 (default_addr_space_pointer_mode): Likewise.
10572 (default_addr_space_address_mode): Likewise.
10573 * targhooks.h (default_eh_return_filter_mode): Likewise.
10574 (default_libgcc_cmp_return_mode): Likewise.
10575 (default_libgcc_shift_count_mode): Likewise.
10576 (default_unwind_word_mode): Likewise.
10577 (default_addr_space_pointer_mode): Likewise.
10578 (default_addr_space_address_mode): Likewise.
10579 (default_mode_rep_extended): Take a scalar_int_mode rather than
10581 (default_valid_pointer_mode): Likewise.
10582 (default_addr_space_valid_pointer_mode): Likewise.
10583 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10585 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10586 rather than a machine_mode.
10587 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10588 for local variables.
10589 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10590 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10591 than a machine_mode.
10593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10594 Alan Hayward <alan.hayward@arm.com>
10595 David Sherwood <david.sherwood@arm.com>
10597 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10598 the mode argument to scalar_int_mode.
10599 (do_jump_by_parts_zero_rtx): Likewise.
10600 (do_jump_by_parts_equality_rtx): Likewise.
10601 (do_jump_by_parts_greater): Take a mode argument.
10602 (do_jump_by_parts_equality): Likewise.
10603 (do_jump_1): Update calls accordingly.
10605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10606 Alan Hayward <alan.hayward@arm.com>
10607 David Sherwood <david.sherwood@arm.com>
10609 * is-a.h (safe_dyn_cast): New function.
10610 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10611 (jump_table_for_label): Likewise.
10612 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10613 instead of an rtx_insn *.
10614 (shorten_branches): Use dyn_cast instead of LABEL_P and
10615 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10616 rtx_jump_table_data::get_data_mode.
10617 (final_scan_insn): Likewise.
10619 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10620 Alan Hayward <alan.hayward@arm.com>
10621 David Sherwood <david.sherwood@arm.com>
10623 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10624 trying to combine a full-register integer set with a subreg
10627 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10628 Alan Hayward <alan.hayward@arm.com>
10629 David Sherwood <david.sherwood@arm.com>
10631 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10632 that is always either address_mode or pointer_mode.
10634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10635 Alan Hayward <alan.hayward@arm.com>
10636 David Sherwood <david.sherwood@arm.com>
10638 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10639 when the two are known to be equal.
10641 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10642 Alan Hayward <alan.hayward@arm.com>
10643 David Sherwood <david.sherwood@arm.com>
10645 * simplify-rtx.c (simplify_const_unary_operation): Use
10646 is_a <scalar_int_mode> instead of checking for a nonzero
10647 precision. Forcibly convert op_mode to a scalar_int_mode
10648 in that case. More clearly differentiate the operand and
10649 result modes and use the former when deciding what the value
10650 of a count-bits operation should be. Use is_int_mode instead
10651 of checking for a MODE_INT. Remove redundant check for whether
10652 this mode has a zero precision.
10654 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10655 Alan Hayward <alan.hayward@arm.com>
10656 David Sherwood <david.sherwood@arm.com>
10658 * optabs.c (widen_leading): Change the type of the mode argument
10659 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10660 (widen_bswap): Likewise.
10661 (expand_parity): Likewise.
10662 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10663 (expand_ffs): Likewise.
10664 (epand_unop): Check for scalar integer modes before calling the
10667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10668 Alan Hayward <alan.hayward@arm.com>
10669 David Sherwood <david.sherwood@arm.com>
10671 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
10673 (expand_expr_real_1): Update call accordingly.
10675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10676 Alan Hayward <alan.hayward@arm.com>
10677 David Sherwood <david.sherwood@arm.com>
10679 * expmed.c (store_bit_field_using_insv): Add op0_mode and
10680 value_mode arguments. Use scalar_int_mode internally.
10681 (store_bit_field_1): Rename the new integer mode from imode
10682 to op0_mode and use it instead of GET_MODE (op0). Update calls
10683 to store_split_bit_field, store_bit_field_using_insv and
10684 store_fixed_bit_field.
10685 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
10686 Use scalar_int_mode internally. Use a bit count rather than a mode
10687 when calculating the largest bit size for get_best_mode.
10688 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10689 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10690 Remove assertion that OP0 has a scalar integer mode.
10691 (store_split_bit_field): Add op0_mode and value_mode arguments.
10692 Update calls to extract_fixed_bit_field.
10693 (extract_bit_field_using_extv): Add an op0_mode argument.
10694 Use scalar_int_mode internally.
10695 (extract_bit_field_1): Rename the new integer mode from imode to
10696 op0_mode and use it instead of GET_MODE (op0). Update calls to
10697 extract_split_bit_field, extract_bit_field_using_extv and
10698 extract_fixed_bit_field.
10699 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10700 to extract_split_bit_field and extract_fixed_bit_field_1.
10701 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10702 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10703 on the target mode.
10704 (extract_split_bit_field): Add an op0_mode argument. Update call
10705 to extract_fixed_bit_field.
10707 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10708 Alan Hayward <alan.hayward@arm.com>
10709 David Sherwood <david.sherwood@arm.com>
10711 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10712 * explow.c (hard_function_value): Likewise.
10713 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10714 convert_to_mode call outside the loop.
10715 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10716 for the mode iterator. Require the mode specified by max_pieces
10718 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10720 (copy_blkmode_to_reg): Likewise.
10721 (set_storage_via_setmem): Likewise.
10722 * optabs.c (prepare_cmp_insn): Likewise.
10723 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10724 * stor-layout.c (finish_bitfield_representative): Likewise.
10726 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10727 Alan Hayward <alan.hayward@arm.com>
10728 David Sherwood <david.sherwood@arm.com>
10730 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10731 * expr.c (convert_move): Use them.
10732 (convert_modes): Likewise.
10733 (store_expr_with_bounds): Likewise.
10735 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10736 Alan Hayward <alan.hayward@arm.com>
10737 David Sherwood <david.sherwood@arm.com>
10739 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10740 parameter for the mode of "x". Remove the "known_x", "known_mode"
10741 and "known_ret" arguments. Change the type of the mode argument
10742 to scalar_int_mode.
10743 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10744 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10745 (reg_num_sign_bit_copies_for_combine): Likewise.
10746 * rtlanal.c (nonzero_bits1): Likewise.
10747 (num_sign_bit_copies1): Likewise.
10748 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10749 (reg_num_sign_bit_copies_general): Likewise.
10750 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10751 (reg_nonzero_bits_general): Likewise.
10753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10754 Alan Hayward <alan.hayward@arm.com>
10755 David Sherwood <david.sherwood@arm.com>
10757 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10758 than in subroutines. Return 1 for non-integer modes.
10759 (cached_num_sign_bit_copies): Change the type of the mode parameter
10760 to scalar_int_mode.
10761 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10762 classes. Handle CONST_INT_P first and then check whether X also
10763 has a scalar integer mode. Check the same thing for inner registers
10764 of a SUBREG and for values that are being extended or truncated.
10766 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10767 Alan Hayward <alan.hayward@arm.com>
10768 David Sherwood <david.sherwood@arm.com>
10770 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10771 in subroutines. Return the mode mask for non-integer modes.
10772 (cached_nonzero_bits): Change the type of the mode parameter
10773 to scalar_int_mode.
10774 (nonzero_bits1): Likewise. Remove early exit for other mode
10775 classes. Handle CONST_INT_P first and then check whether X
10776 also has a scalar integer mode.
10778 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10779 Alan Hayward <alan.hayward@arm.com>
10780 David Sherwood <david.sherwood@arm.com>
10782 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10783 Return a scalar_int_mode and assert that the size is greater than
10785 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10786 (op_by_pieces_d::op_by_pieces_d): Likewise.
10787 (op_by_pieces_d::run): Likewise.
10788 (can_store_by_pieces): Likewise.
10790 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10791 Alan Hayward <alan.hayward@arm.com>
10792 David Sherwood <david.sherwood@arm.com>
10794 * combine.c (extract_left_shift): Add a mode argument and update
10796 (make_compound_operation_int): Change the type of the mode parameter
10797 to scalar_int_mode and update the call to extract_left_shift.
10799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10800 Alan Hayward <alan.hayward@arm.com>
10801 David Sherwood <david.sherwood@arm.com>
10803 * combine.c (simplify_and_const_int): Change the type of the mode
10804 parameter to scalar_int_mode.
10805 (simplify_and_const_int_1): Likewise. Update recursive call.
10807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10808 Alan Hayward <alan.hayward@arm.com>
10809 David Sherwood <david.sherwood@arm.com>
10811 * combine.c (simplify_compare_const): Check that the mode is a
10812 scalar_int_mode (rather than VOIDmode) before testing its
10814 (simplify_comparison): Move COMPARISON_P handling out of the
10815 loop and restrict the latter part of the loop to scalar_int_modes.
10816 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10817 and when considering SUBREG_REGs. Use is_int_mode instead of
10818 checking GET_MODE_CLASS against MODE_INT.
10820 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10821 Alan Hayward <alan.hayward@arm.com>
10822 David Sherwood <david.sherwood@arm.com>
10824 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10825 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10826 shift_unit_mode and for modes involved in scalar shifts.
10828 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10829 Alan Hayward <alan.hayward@arm.com>
10830 David Sherwood <david.sherwood@arm.com>
10832 * combine.c (force_int_to_mode): New function, split out from...
10833 (force_to_mode): ...here. Keep xmode up-to-date and use it
10834 instead of GET_MODE (x).
10836 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10837 Alan Hayward <alan.hayward@arm.com>
10838 David Sherwood <david.sherwood@arm.com>
10840 * optabs-query.h (extraction_insn::struct_mode): Change type to
10841 opt_scalar_int_mode and update comment.
10842 (extraction_insn::field_mode): Change type to scalar_int_mode.
10843 (extraction_insn::pos_mode): Likewise.
10844 * combine.c (make_extraction): Update accordingly.
10845 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10846 (get_optab_extraction_insn): Likewise.
10847 * recog.c (simplify_while_replacing): Likewise.
10848 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10849 parameter to opt_scalar_int_mode.
10851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10852 Alan Hayward <alan.hayward@arm.com>
10853 David Sherwood <david.sherwood@arm.com>
10855 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10856 to a scalar_int_mode instead of a machine_mode.
10857 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10858 (get_best_mode): Return a boolean and use a pointer argument to store
10859 the selected mode. Replace the limit mode parameter with a bit limit.
10860 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10861 for the values returned by bit_field_mode_iterator::next_mode.
10862 (store_bit_field): Update call to get_best_mode.
10863 (store_fixed_bit_field): Likewise.
10864 (extract_fixed_bit_field): Likewise.
10865 * expr.c (optimize_bitfield_assignment_op): Likewise.
10866 * fold-const.c (optimize_bit_field_compare): Likewise.
10867 (fold_truth_andor_1): Likewise.
10868 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10869 Update for new type of m_mode.
10870 (get_best_mode): As above.
10872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10873 Alan Hayward <alan.hayward@arm.com>
10874 David Sherwood <david.sherwood@arm.com>
10876 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10877 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10878 (store_bit_field): Check is_a <scalar_int_mode> before calling
10879 strict_volatile_bitfield_p.
10880 (extract_bit_field): Likewise.
10882 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10883 Alan Hayward <alan.hayward@arm.com>
10884 David Sherwood <david.sherwood@arm.com>
10886 * target.def (cstore_mode): Return a scalar_int_mode.
10887 * doc/tm.texi: Regenerate.
10888 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10889 * targhooks.h (default_cstore_mode): Likewise.
10890 * targhooks.c (default_cstore_mode): Likewise, using a forced
10892 * expmed.c (emit_cstore): Expect the target of the cstore to be
10895 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10896 Alan Hayward <alan.hayward@arm.com>
10897 David Sherwood <david.sherwood@arm.com>
10899 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10901 (niter_desc): Likewise mode.
10902 (iv_analyze): Add a mode parameter.
10904 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10905 and change its type to scalar_int_mode.
10906 * loop-iv.c: Update commentary at head of file.
10907 (iv_constant): Pass the mode paraeter before the rtx it describes
10908 and change its type to scalar_int_mode. Remove VOIDmode handling.
10909 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10910 (iv_extend): Likewise.
10911 (shorten_into_mode): Likewise.
10912 (iv_add): Use scalar_int_mode.
10913 (iv_mult): Likewise.
10914 (iv_shift): Likewise.
10915 (canonicalize_iv_subregs): Likewise.
10916 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10917 it describes and change its mode to scalar_int_mode. Also change
10918 the type of the returned inner_mode to scalar_int_mode.
10919 (get_biv_step): Likewise, turning outer_mode from a pointer
10920 into a direct parameter. Update call to get_biv_step_1.
10921 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10922 iv_constant and get_biv_step.
10923 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10924 and change its type to scalar_int_mode. Don't initialise iv->mode
10925 to VOIDmode and remove later checks for its still being VOIDmode.
10926 Update calls to iv_analyze_op and iv_analyze_expr. Check
10927 is_a <scalar_int_mode> when changing the mode under consideration.
10928 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10929 Update call to iv_analyze_expr.
10930 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10931 inner register is not also a scalar_int_mode. Update call to
10933 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10934 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10935 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10936 separate mode class checks. Update calls to iv_analyze. Remove
10937 fix-up of VOIDmodes after iv_analyze_biv.
10938 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10939 don't have a scalar_int_mode. Update call to biv_p.
10941 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10942 Alan Hayward <alan.hayward@arm.com>
10943 David Sherwood <david.sherwood@arm.com>
10945 * cfgexpand.c (convert_debug_memory_address): Use
10946 as_a <scalar_int_mode>.
10947 * combine.c (expand_compound_operation): Likewise.
10948 (make_extraction): Likewise.
10949 (change_zero_ext): Likewise.
10950 (simplify_comparison): Likewise.
10951 * cse.c (cse_insn): Likewise.
10952 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10953 (mem_loc_descriptor): Likewise.
10954 (loc_descriptor): Likewise.
10955 * expmed.c (init_expmed_one_mode): Likewise.
10956 (synth_mult): Likewise.
10957 (emit_store_flag_1): Likewise.
10958 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10959 of a comparison with size.
10960 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10961 (reduce_to_bit_field_precision): Likewise.
10962 * function.c (expand_function_end): Likewise.
10963 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10964 * loop-doloop.c (doloop_modify): Likewise.
10965 * optabs.c (expand_binop): Likewise.
10966 (expand_unop): Likewise.
10967 (expand_copysign_absneg): Likewise.
10968 (prepare_cmp_insn): Likewise.
10969 (maybe_legitimize_operand): Likewise.
10970 * recog.c (const_scalar_int_operand): Likewise.
10971 * rtlanal.c (get_address_mode): Likewise.
10972 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10973 (simplify_cond_clz_ctz): Likewise.
10974 * tree-nested.c (get_nl_goto_field): Likewise.
10975 * tree.c (build_vector_type_for_mode): Likewise.
10976 * var-tracking.c (use_narrower_mode): Likewise.
10978 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10979 Alan Hayward <alan.hayward@arm.com>
10980 David Sherwood <david.sherwood@arm.com>
10982 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10983 * builtins.c (expand_builtin_signbit): Use it.
10984 * cfgexpand.c (expand_debug_expr): Likewise.
10985 * dojump.c (do_jump): Likewise.
10986 (do_compare_and_jump): Likewise.
10987 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
10988 * expmed.c (make_tree): Likewise.
10989 * expr.c (expand_expr_real_2): Likewise.
10990 (expand_expr_real_1): Likewise.
10991 (try_casesi): Likewise.
10992 * fold-const-call.c (fold_const_call_ss): Likewise.
10993 * fold-const.c (unextend): Likewise.
10994 (extract_muldiv_1): Likewise.
10995 (fold_single_bit_test): Likewise.
10996 (native_encode_int): Likewise.
10997 (native_encode_string): Likewise.
10998 (native_interpret_int): Likewise.
10999 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11000 * internal-fn.c (expand_addsub_overflow): Likewise.
11001 (expand_neg_overflow): Likewise.
11002 (expand_mul_overflow): Likewise.
11003 (expand_arith_overflow): Likewise.
11004 * match.pd: Likewise.
11005 * stor-layout.c (layout_type): Likewise.
11006 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11007 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11008 * tree-ssanames.c (get_range_info): Likewise.
11009 * tree-switch-conversion.c (array_value_type) Likewise.
11010 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11011 (vect_recog_divmod_pattern): Likewise.
11012 (vect_recog_mixed_size_cond_pattern): Likewise.
11013 * tree-vrp.c (extract_range_basic): Likewise.
11014 (simplify_float_conversion_using_ranges): Likewise.
11015 * tree.c (int_fits_type_p): Likewise.
11016 * ubsan.c (instrument_bool_enum_load): Likewise.
11017 * varasm.c (mergeable_string_section): Likewise.
11018 (narrowing_initializer_constant_valid_p): Likewise.
11019 (output_constant): Likewise.
11021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11022 Alan Hayward <alan.hayward@arm.com>
11023 David Sherwood <david.sherwood@arm.com>
11025 * machmode.h (NARROWEST_INT_MODE): New macro.
11026 * expr.c (alignment_for_piecewise_move): Use it instead of
11027 GET_CLASS_NARROWEST_MODE (MODE_INT).
11028 (push_block): Likewise.
11029 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11031 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11033 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11034 Alan Hayward <alan.hayward@arm.com>
11035 David Sherwood <david.sherwood@arm.com>
11037 * postreload.c (move2add_valid_value_p): Change the type of the
11038 mode parameter to scalar_int_mode.
11039 (move2add_use_add2_insn): Add a mode parameter and use it instead
11041 (move2add_use_add3_insn): Likewise.
11042 (reload_cse_move2add): Update accordingly.
11044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11045 Alan Hayward <alan.hayward@arm.com>
11046 David Sherwood <david.sherwood@arm.com>
11048 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11050 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11051 * optabs.c (expand_unop): Likewise.
11053 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11054 Alan Hayward <alan.hayward@arm.com>
11055 David Sherwood <david.sherwood@arm.com>
11057 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11058 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11059 (popcount_loc_descriptor): Likewise.
11060 (bswap_loc_descriptor): Likewise.
11061 (rotate_loc_descriptor): Likewise.
11062 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11063 calling the functions above.
11065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11066 Alan Hayward <alan.hayward@arm.com>
11067 David Sherwood <david.sherwood@arm.com>
11069 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11071 (try_combine): Likewise.
11072 (simplify_if_then_else): Likewise.
11073 * cse.c (cse_insn): Likewise.
11074 * dwarf2out.c (mem_loc_descriptor): Likewise.
11075 * emit-rtl.c (gen_lowpart_common): Likewise.
11076 * simplify-rtx.c (simplify_truncation): Likewise.
11077 (simplify_binary_operation_1): Likewise.
11078 (simplify_const_relational_operation): Likewise.
11079 (simplify_ternary_operation): Likewise.
11080 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11082 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11083 Alan Hayward <alan.hayward@arm.com>
11084 David Sherwood <david.sherwood@arm.com>
11086 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11087 * reload.c (push_reload): Likewise.
11088 (find_reloads): Likewise.
11090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11091 Alan Hayward <alan.hayward@arm.com>
11092 David Sherwood <david.sherwood@arm.com>
11094 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11095 (make_compound_operation_int): Likewise.
11096 (change_zero_ext): Likewise.
11097 * expr.c (convert_move): Likewise.
11098 (convert_modes): Likewise.
11099 * fwprop.c (forward_propagate_subreg): Likewise.
11100 * loop-iv.c (get_biv_step_1): Likewise.
11101 * optabs.c (widen_operand): Likewise.
11102 * postreload.c (move2add_valid_value_p): Likewise.
11103 * recog.c (simplify_while_replacing): Likewise.
11104 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11105 (simplify_binary_operation_1): Likewise. Remove redundant
11106 mode equality check.
11108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11109 Alan Hayward <alan.hayward@arm.com>
11110 David Sherwood <david.sherwood@arm.com>
11112 * combine.c (combine_simplify_rtx): Add checks for
11113 is_a <scalar_int_mode>.
11114 (simplify_if_then_else): Likewise.
11115 (make_field_assignment): Likewise.
11116 (simplify_comparison): Likewise.
11117 * ifcvt.c (noce_try_bitop): Likewise.
11118 * loop-invariant.c (canonicalize_address_mult): Likewise.
11119 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11121 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11122 Alan Hayward <alan.hayward@arm.com>
11123 David Sherwood <david.sherwood@arm.com>
11125 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11126 is_a <scalar_int_mode> instead of != BLKmode.
11128 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11129 Alan Hayward <alan.hayward@arm.com>
11130 David Sherwood <david.sherwood@arm.com>
11132 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11133 instead of != VOIDmode.
11134 * combine.c (if_then_else_cond): Likewise.
11135 (change_zero_ext): Likewise.
11136 * dwarf2out.c (mem_loc_descriptor): Likewise.
11137 (loc_descriptor): Likewise.
11138 * rtlanal.c (canonicalize_condition): Likewise.
11139 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11141 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11142 Alan Hayward <alan.hayward@arm.com>
11143 David Sherwood <david.sherwood@arm.com>
11145 * simplify-rtx.c (simplify_binary_operation_1): Use
11146 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11148 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11149 Alan Hayward <alan.hayward@arm.com>
11150 David Sherwood <david.sherwood@arm.com>
11152 * wide-int.h (int_traits<unsigned char>) New class.
11153 (int_traits<unsigned short>) Likewise.
11154 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11155 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11157 * combine.c (set_nonzero_bits_and_sign_copies): Use
11158 is_a <scalar_int_mode>.
11159 (find_split_point): Likewise.
11160 (combine_simplify_rtx): Likewise.
11161 (simplify_logical): Likewise.
11162 (expand_compound_operation): Likewise.
11163 (expand_field_assignment): Likewise.
11164 (make_compound_operation): Likewise.
11165 (extended_count): Likewise.
11166 (change_zero_ext): Likewise.
11167 (simplify_comparison): Likewise.
11168 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11169 (ucompare_loc_descriptor): Likewise.
11170 (minmax_loc_descriptor): Likewise.
11171 (mem_loc_descriptor): Likewise.
11172 (loc_descriptor): Likewise.
11173 * expmed.c (init_expmed_one_mode): Likewise.
11174 * lra-constraints.c (lra_constraint_offset): Likewise.
11175 * optabs.c (prepare_libcall_arg): Likewise.
11176 * postreload.c (move2add_note_store): Likewise.
11177 * reload.c (operands_match_p): Likewise.
11178 * rtl.h (load_extend_op): Likewise.
11179 * rtlhooks.c (gen_lowpart_general): Likewise.
11180 * simplify-rtx.c (simplify_truncation): Likewise.
11181 (simplify_unary_operation_1): Likewise.
11182 (simplify_binary_operation_1): Likewise.
11183 (simplify_const_binary_operation): Likewise.
11184 (simplify_const_relational_operation): Likewise.
11185 (simplify_subreg): Likewise.
11186 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11187 * var-tracking.c (adjust_mems): Likewise.
11188 (prepare_call_arguments): Likewise.
11190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11191 Alan Hayward <alan.hayward@arm.com>
11192 David Sherwood <david.sherwood@arm.com>
11194 * machmode.h (is_int_mode): New fuction.
11195 * combine.c (find_split_point): Use it.
11196 (combine_simplify_rtx): Likewise.
11197 (simplify_if_then_else): Likewise.
11198 (simplify_set): Likewise.
11199 (simplify_shift_const_1): Likewise.
11200 (simplify_comparison): Likewise.
11201 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
11202 * cse.c (notreg_cost): Likewise.
11203 (cse_insn): Likewise.
11204 * cselib.c (cselib_lookup_1): Likewise.
11205 * dojump.c (do_jump_1): Likewise.
11206 (do_compare_rtx_and_jump): Likewise.
11207 * dse.c (get_call_args): Likewise.
11208 * dwarf2out.c (rtl_for_decl_init): Likewise.
11209 (native_encode_initializer): Likewise.
11210 * expmed.c (emit_store_flag_1): Likewise.
11211 (emit_store_flag): Likewise.
11212 * expr.c (convert_modes): Likewise.
11213 (store_field): Likewise.
11214 (expand_expr_real_1): Likewise.
11215 * fold-const.c (fold_read_from_constant_string): Likewise.
11216 * gimple-ssa-sprintf.c (get_format_string): Likewise.
11217 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
11218 * optabs.c (expand_binop): Likewise.
11219 (expand_unop): Likewise.
11220 (expand_abs_nojump): Likewise.
11221 (expand_one_cmpl_abs_nojump): Likewise.
11222 * simplify-rtx.c (mode_signbit_p): Likewise.
11223 (val_signbit_p): Likewise.
11224 (val_signbit_known_set_p): Likewise.
11225 (val_signbit_known_clear_p): Likewise.
11226 (simplify_relational_operation_1): Likewise.
11227 * tree.c (vector_type_mode): Likewise.
11229 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11230 Alan Hayward <alan.hayward@arm.com>
11231 David Sherwood <david.sherwood@arm.com>
11233 * machmode.h (smallest_mode_for_size): Fix formatting.
11234 (smallest_int_mode_for_size): New function.
11235 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11236 instead of smallest_mode_for_size.
11237 * combine.c (make_extraction): Likewise.
11238 * config/arc/arc.c (arc_expand_movmem): Likewise.
11239 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11240 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11241 * config/s390/s390.c (s390_expand_insv): Likewise.
11242 * config/sparc/sparc.c (assign_int_registers): Likewise.
11243 * config/spu/spu.c (spu_function_value): Likewise.
11244 (spu_function_arg): Likewise.
11245 * coverage.c (get_gcov_type): Likewise.
11246 (get_gcov_unsigned_t): Likewise.
11247 * dse.c (find_shift_sequence): Likewise.
11248 * expmed.c (store_bit_field_1): Likewise.
11249 * expr.c (convert_move): Likewise.
11250 (store_field): Likewise.
11251 * internal-fn.c (expand_arith_overflow): Likewise.
11252 * optabs-query.c (get_best_extraction_insn): Likewise.
11253 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11254 * stor-layout.c (layout_type): Likewise.
11255 (initialize_sizetypes): Likewise.
11256 * targhooks.c (default_get_mask_mode): Likewise.
11257 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11260 Alan Hayward <alan.hayward@arm.com>
11261 David Sherwood <david.sherwood@arm.com>
11263 * machmode.h (opt_mode::else_blk): New function.
11264 (int_mode_for_mode): Declare.
11265 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11266 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11268 * cfgexpand.c (expand_debug_expr): Likewise.
11269 * combine.c (gen_lowpart_or_truncate): Likewise.
11270 (gen_lowpart_for_combine): Likewise.
11271 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11272 * config/avr/avr.c (avr_to_int_mode): Likewise.
11273 (avr_out_plus_1): Likewise.
11274 (avr_out_plus): Likewise.
11275 (avr_out_round): Likewise.
11276 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11277 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11278 (s390_expand_vcond): Likewise.
11279 * config/spu/spu.c (spu_split_immediate): Likewise.
11280 (spu_expand_mov): Likewise.
11281 * dse.c (get_stored_val): Likewise.
11282 * expmed.c (store_bit_field_1): Likewise.
11283 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11285 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11286 (extract_low_bits): Likewise.
11287 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11288 handling rather than repeating the check.
11289 (emit_group_store): Likewise.
11290 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11291 * optabs.c (expand_absneg_bit): Likewise.
11292 (expand_copysign_absneg): Likewise.
11293 (expand_copysign_bit): Likewise.
11294 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11295 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11296 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11297 * var-tracking.c (prepare_call_arguments): Likewise.
11298 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11299 int_mode_for_mode instead of mode_for_size.
11300 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11303 Alan Hayward <alan.hayward@arm.com>
11304 David Sherwood <david.sherwood@arm.com>
11306 * machmode.h (int_mode_for_size): New function.
11307 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11308 instead of mode_for_size.
11309 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11311 * combine.c (expand_field_assignment): Use int_mode_for_size
11312 instead of mode_for_size.
11313 (make_extraction): Likewise.
11314 (simplify_shift_const_1): Likewise.
11315 (simplify_comparison): Likewise.
11316 * dojump.c (do_jump): Likewise.
11317 * dwarf2out.c (mem_loc_descriptor): Likewise.
11318 * emit-rtl.c (init_derived_machine_modes): Likewise.
11319 * expmed.c (flip_storage_order): Likewise.
11320 (convert_extracted_bit_field): Likewise.
11321 * expr.c (copy_blkmode_from_reg): Likewise.
11322 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11323 * internal-fn.c (expand_mul_overflow): Likewise.
11324 * lower-subreg.c (simple_move): Likewise.
11325 * optabs-libfuncs.c (init_optabs): Likewise.
11326 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11327 * tree.c (vector_type_mode): Likewise.
11328 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11329 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11330 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11331 * tree-vect-stmts.c (vectorizable_load): Likewise.
11332 (vectorizable_store): Likewise.
11334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11335 Alan Hayward <alan.hayward@arm.com>
11336 David Sherwood <david.sherwood@arm.com>
11338 * coretypes.h (pod_mode): New type.
11339 (scalar_int_mode_pod): New typedef.
11340 * machmode.h (pod_mode): New class.
11341 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11342 * genmodes.c (emit_mode_int_n): Update accordingly.
11343 * lower-subreg.h (target_lower_subreg): Change type to
11344 scalar_int_mode_pod.
11345 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11346 scalar_int_mode_pod.
11348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11349 Alan Hayward <alan.hayward@arm.com>
11350 David Sherwood <david.sherwood@arm.com>
11352 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11353 machine_mode to scalar_int_mode.
11354 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11355 (rs6000_option_override_internal): Remove cast to int.
11356 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11357 machine_mode to scalar_int_mode.
11358 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11359 (rs6000_option_override_internal): Remove cast to int.
11360 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11361 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11363 * config/s390/s390.c (s390_expand_builtin): Likewise.
11364 * coretypes.h (scalar_int_mode): New type.
11365 (opt_scalar_int_mode): New typedef.
11366 * machmode.h (scalar_int_mode): New class.
11367 (scalar_int_mode::includes_p): New function.
11368 (byte_mode): Change type to scalar_int_mode.
11369 (word_mode): Likewise.
11370 (ptr_mode): Likewise.
11371 * emit-rtl.c (byte_mode): Likewise.
11372 (word_mode): Likewise.
11373 (ptr_mode): Likewise.
11374 (init_derived_machine_modes): Update accordingly.
11375 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11376 and MODE_PARTIAL_INT.
11377 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11378 opt_scalar_int_mode.
11380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11381 Alan Hayward <alan.hayward@arm.com>
11382 David Sherwood <david.sherwood@arm.com>
11384 * target.def (libgcc_floating_mode_supported_p): Take a
11386 * doc/tm.texi: Regenerate.
11387 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11389 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11390 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11394 Alan Hayward <alan.hayward@arm.com>
11395 David Sherwood <david.sherwood@arm.com>
11397 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11398 * doc/tm.texi: Regenerate.
11399 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11400 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11401 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11402 * targhooks.h (default_floatn_mode): Likewise.
11403 * targhooks.c (default_floatn_mode): Likewise.
11404 * tree.c (build_common_tree_nodes): Update accordingly.
11406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11407 Alan Hayward <alan.hayward@arm.com>
11408 David Sherwood <david.sherwood@arm.com>
11410 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11411 (mode_iterator::iterate_p): Likewise.
11412 (mode_iterator::get_wider): Likewise.
11413 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11416 Alan Hayward <alan.hayward@arm.com>
11417 David Sherwood <david.sherwood@arm.com>
11419 * coretypes.h (opt_scalar_float_mode): New typedef.
11420 * machmode.h (float_mode_for_size): New function.
11421 * emit-rtl.c (double_mode): Delete.
11422 (init_emit_once): Use float_mode_for_size.
11423 * stor-layout.c (layout_type): Likewise.
11424 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11426 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11427 Alan Hayward <alan.hayward@arm.com>
11428 David Sherwood <david.sherwood@arm.com>
11430 * output.h (assemble_real): Take a scalar_float_mode.
11431 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11432 * config/arm/arm.md (consttable_4): Likewise.
11433 (consttable_8): Likewise.
11434 (consttable_16): Likewise.
11435 * config/mips/mips.md (consttable_float): Likewise.
11436 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11437 * varasm.c (assemble_real): Take a scalar_float_mode.
11438 (output_constant_pool_2): Update accordingly.
11439 (output_constant): Likewise.
11441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11442 Alan Hayward <alan.hayward@arm.com>
11443 David Sherwood <david.sherwood@arm.com>
11445 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11446 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11447 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11448 (native_encode_real): Likewise.
11449 (native_interpret_real): Likewise.
11450 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11451 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11453 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11454 Alan Hayward <alan.hayward@arm.com>
11455 David Sherwood <david.sherwood@arm.com>
11457 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11458 <scalar_float_mode>. Simplify.
11459 (gen_extend_conv_libfunc): Likewise.
11461 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11462 Alan Hayward <alan.hayward@arm.com>
11463 David Sherwood <david.sherwood@arm.com>
11465 * coretypes.h (scalar_float_mode): New type.
11466 * machmode.h (mode_traits::from_int): Use machine_mode if
11467 USE_ENUM_MODES is defined.
11468 (is_a): New function.
11470 (dyn_cast): Likewise.
11471 (scalar_float_mode): New class.
11472 (scalar_float_mode::includes_p): New function.
11473 (is_float_mode): Likewise.
11474 * gdbhooks.py (MachineModePrinter): New class.
11475 (build_pretty_printer): Use it for scalar_float_mode.
11476 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11477 (format_helper::format_helper): Turn into a template.
11478 * genmodes.c (get_mode_class): New function.
11479 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11480 or machine_mode if none.
11481 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11482 as_a <scalar_float_mode>.
11483 * dwarf2out.c (mem_loc_descriptor): Likewise.
11484 (insert_float): Likewise.
11485 (add_const_value_attribute): Likewise.
11486 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11487 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11488 (expand_unop): Update accordingly.
11489 (expand_abs_nojump): Likewise.
11490 (expand_copysign_absneg): Take a scalar_float_mode.
11491 (expand_copysign_bit): Likewise.
11492 (expand_copysign): Update accordingly.
11494 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11495 Alan Hayward <alan.hayward@arm.com>
11496 David Sherwood <david.sherwood@arm.com>
11498 * coretypes.h (opt_mode): New class.
11499 * machmode.h (opt_mode): Likewise.
11500 (opt_mode::else_void): New function.
11501 (opt_mode::require): Likewise.
11502 (opt_mode::exists): Likewise.
11503 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11504 (GET_MODE_2XWIDER_MODE): Likewise.
11505 (mode_iterator::get_wider): Update accordingly.
11506 (mode_iterator::get_2xwider): Likewise.
11507 (mode_iterator::get_known_wider): Likewise, turning into a template.
11508 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11509 forcing a wider mode to exist.
11510 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11511 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11512 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11513 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11514 * lower-subreg.c (init_lower_subreg): Likewise.
11515 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11516 on the final iteration.
11517 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11518 a wider mode exists before asking for a move pattern.
11519 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11520 forcing a wider mode to exist.
11521 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11522 returning false if no such mode exists.
11523 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11524 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11525 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11526 Avoid checking for a MODE_INT if we already know the mode is not a
11528 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11529 forcing a wider mode to exist.
11530 (expmed_mult_highpart_optab): Likewise.
11531 (expmed_mult_highpart): Likewise.
11532 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11534 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11535 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11536 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11537 * internal-fn.c (expand_mul_overflow): Update use of
11538 GET_MODE_2XWIDER_MODE.
11539 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11540 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11541 GET_MODE_WIDER_MODE.
11542 (convert_plusminus_to_widen): Likewise.
11543 * tree-switch-conversion.c (array_value_type): Likewise.
11544 * var-tracking.c (emit_note_insn_var_location): Likewise.
11545 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11546 Return false inside rather than outside the loop if no wider mode
11548 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11549 and GET_MODE_2XWIDER_MODE
11550 (can_compare_p): Use else_void.
11551 * gdbhooks.py (OptMachineModePrinter): New class.
11552 (build_pretty_printer): Use it for opt_mode.
11554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11555 Alan Hayward <alan.hayward@arm.com>
11556 David Sherwood <david.sherwood@arm.com>
11558 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11559 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11562 Alan Hayward <alan.hayward@arm.com>
11563 David Sherwood <david.sherwood@arm.com>
11565 * machmode.h (mode_traits): New structure.
11566 (get_narrowest_mode): New function.
11567 (mode_iterator::start): Likewise.
11568 (mode_iterator::iterate_p): Likewise.
11569 (mode_iterator::get_wider): Likewise.
11570 (mode_iterator::get_known_wider): Likewise.
11571 (mode_iterator::get_2xwider): Likewise.
11572 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11573 (FOR_EACH_MODE): Likewise.
11574 (FOR_EACH_MODE_FROM): Likewise.
11575 (FOR_EACH_MODE_UNTIL): Likewise.
11576 (FOR_EACH_WIDER_MODE): Likewise.
11577 (FOR_EACH_2XWIDER_MODE): Likewise.
11578 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11579 * combine.c (simplify_comparison): Likewise
11580 * config/i386/i386.c (type_natural_mode): Likewise.
11581 * cse.c (cse_insn): Likewise.
11582 * dse.c (find_shift_sequence): Likewise.
11583 * emit-rtl.c (init_derived_machine_modes): Likewise.
11584 (init_emit_once): Likewise.
11585 * explow.c (hard_function_value): Likewise.
11586 * expmed.c (extract_fixed_bit_field_1): Likewise.
11587 (extract_bit_field_1): Likewise.
11588 (expand_divmod): Likewise.
11589 (emit_store_flag_1): Likewise.
11590 * expr.c (init_expr_target): Likewise.
11591 (convert_move): Likewise.
11592 (alignment_for_piecewise_move): Likewise.
11593 (widest_int_mode_for_size): Likewise.
11594 (emit_block_move_via_movmem): Likewise.
11595 (copy_blkmode_to_reg): Likewise.
11596 (set_storage_via_setmem): Likewise.
11597 (compress_float_constant): Likewise.
11598 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11599 * optabs-query.c (get_best_extraction_insn): Likewise.
11600 * optabs.c (expand_binop): Likewise.
11601 (expand_twoval_unop): Likewise.
11602 (expand_twoval_binop): Likewise.
11603 (widen_leading): Likewise.
11604 (widen_bswap): Likewise.
11605 (expand_parity): Likewise.
11606 (expand_unop): Likewise.
11607 (prepare_cmp_insn): Likewise.
11608 (prepare_float_lib_cmp): Likewise.
11609 (expand_float): Likewise.
11610 (expand_fix): Likewise.
11611 (expand_sfix_optab): Likewise.
11612 * postreload.c (move2add_use_add2_insn): Likewise.
11613 * reg-stack.c (reg_to_stack): Likewise.
11614 * reginfo.c (choose_hard_reg_mode): Likewise.
11615 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11616 * stor-layout.c (mode_for_size): Likewise.
11617 (smallest_mode_for_size): Likewise.
11618 (mode_for_vector): Likewise.
11619 (finish_bitfield_representative): Likewise.
11620 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11621 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11622 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11623 * var-tracking.c (prepare_call_arguments): Likewise.
11625 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11626 Alan Hayward <alan.hayward@arm.com>
11627 David Sherwood <david.sherwood@arm.com>
11629 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11630 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11631 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11632 * machmode.h (mode_size): Move earlier in file.
11633 (mode_precision): Likewise.
11634 (mode_inner): Likewise.
11635 (mode_nunits): Likewise.
11636 (mode_unit_size): Likewise.
11637 (unit_unit_precision): Likewise.
11638 (mode_wider): Likewise.
11639 (mode_2xwider): Likewise.
11640 (machine_mode): New class.
11641 (mode_to_bytes): New function.
11642 (mode_to_bits): Likewise.
11643 (mode_to_precision): Likewise.
11644 (mode_to_inner): Likewise.
11645 (mode_to_unit_size): Likewise.
11646 (mode_to_unit_precision): Likewise.
11647 (mode_to_nunits): Likewise.
11648 (GET_MODE_SIZE): Use mode_to_bytes.
11649 (GET_MODE_BITSIZE): Use mode_to_bits.
11650 (GET_MODE_PRECISION): Use mode_to_precision.
11651 (GET_MODE_INNER): Use mode_to_inner.
11652 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11653 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11654 (GET_MODE_NUNITS): Use mode_to_nunits.
11655 * system.h (ALWAYS_INLINE): New macro.
11656 * config/powerpcspe/powerpcspe-c.c
11657 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11658 int for arg1_mode and arg2_mode.
11660 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11661 Alan Hayward <alan.hayward@arm.com>
11662 David Sherwood <david.sherwood@arm.com>
11664 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11665 Prefix mode names with E_ in case statements.
11666 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11667 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11668 (aarch64_split_simd_move): Likewise.
11669 (aarch64_gen_storewb_pair): Likewise.
11670 (aarch64_gen_loadwb_pair): Likewise.
11671 (aarch64_gen_store_pair): Likewise.
11672 (aarch64_gen_load_pair): Likewise.
11673 (aarch64_get_condition_code_1): Likewise.
11674 (aarch64_constant_pool_reload_icode): Likewise.
11675 (get_rsqrte_type): Likewise.
11676 (get_rsqrts_type): Likewise.
11677 (get_recpe_type): Likewise.
11678 (get_recps_type): Likewise.
11679 (aarch64_gimplify_va_arg_expr): Likewise.
11680 (aarch64_simd_container_mode): Likewise.
11681 (aarch64_emit_load_exclusive): Likewise.
11682 (aarch64_emit_store_exclusive): Likewise.
11683 (aarch64_expand_compare_and_swap): Likewise.
11684 (aarch64_gen_atomic_cas): Likewise.
11685 (aarch64_emit_bic): Likewise.
11686 (aarch64_emit_atomic_swap): Likewise.
11687 (aarch64_emit_atomic_load_op): Likewise.
11688 (aarch64_evpc_trn): Likewise.
11689 (aarch64_evpc_uzp): Likewise.
11690 (aarch64_evpc_zip): Likewise.
11691 (aarch64_evpc_ext): Likewise.
11692 (aarch64_evpc_rev): Likewise.
11693 (aarch64_evpc_dup): Likewise.
11694 (aarch64_gen_ccmp_first): Likewise.
11695 (aarch64_gen_ccmp_next): Likewise.
11696 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11697 (alpha_emit_xfloating_libcall): Likewise.
11698 (emit_insxl): Likewise.
11699 (alpha_arg_type): Likewise.
11700 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11701 (arc_preferred_simd_mode): Likewise.
11702 (arc_secondary_reload): Likewise.
11703 (get_arc_condition_code): Likewise.
11704 (arc_print_operand): Likewise.
11705 (arc_legitimate_constant_p): Likewise.
11706 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11707 * config/arc/arc.md (casesi_load): Likewise.
11708 (casesi_compact_jump): Likewise.
11709 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11710 (cc_use_register): Likewise.
11711 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11712 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11713 (arm_init_iwmmxt_builtins): Likewise.
11714 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11715 (neon_expand_vector_init): Likewise.
11716 (arm_attr_length_move_neon): Likewise.
11717 (maybe_get_arm_condition_code): Likewise.
11718 (arm_emit_vector_const): Likewise.
11719 (arm_preferred_simd_mode): Likewise.
11720 (arm_output_iwmmxt_tinsr): Likewise.
11721 (thumb1_output_casesi): Likewise.
11722 (thumb2_output_casesi): Likewise.
11723 (arm_emit_load_exclusive): Likewise.
11724 (arm_emit_store_exclusive): Likewise.
11725 (arm_expand_compare_and_swap): Likewise.
11726 (arm_evpc_neon_vuzp): Likewise.
11727 (arm_evpc_neon_vzip): Likewise.
11728 (arm_evpc_neon_vrev): Likewise.
11729 (arm_evpc_neon_vtrn): Likewise.
11730 (arm_evpc_neon_vext): Likewise.
11731 (arm_validize_comparison): Likewise.
11732 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11733 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11734 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11735 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11736 (c6x_preferred_simd_mode): Likewise.
11737 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11738 (epiphany_rtx_costs): Likewise.
11739 * config/epiphany/predicates.md (proper_comparison_operator):
11741 * config/frv/frv.c (condexec_memory_operand): Likewise.
11742 (frv_emit_move): Likewise.
11743 (output_move_single): Likewise.
11744 (output_condmove_single): Likewise.
11745 (frv_hard_regno_mode_ok): Likewise.
11746 (frv_matching_accg_mode): Likewise.
11747 * config/h8300/h8300.c (split_adds_subs): Likewise.
11748 (h8300_rtx_costs): Likewise.
11749 (h8300_print_operand): Likewise.
11750 (compute_mov_length): Likewise.
11751 (output_logical_op): Likewise.
11752 (compute_logical_op_length): Likewise.
11753 (compute_logical_op_cc): Likewise.
11754 (h8300_shift_needs_scratch_p): Likewise.
11755 (output_a_shift): Likewise.
11756 (compute_a_shift_length): Likewise.
11757 (compute_a_shift_cc): Likewise.
11758 (expand_a_rotate): Likewise.
11759 (output_a_rotate): Likewise.
11760 * config/i386/i386.c (classify_argument): Likewise.
11761 (function_arg_advance_32): Likewise.
11762 (function_arg_32): Likewise.
11763 (function_arg_64): Likewise.
11764 (function_value_64): Likewise.
11765 (ix86_gimplify_va_arg): Likewise.
11766 (ix86_legitimate_constant_p): Likewise.
11767 (put_condition_code): Likewise.
11768 (split_double_mode): Likewise.
11769 (ix86_avx256_split_vector_move_misalign): Likewise.
11770 (ix86_expand_vector_logical_operator): Likewise.
11771 (ix86_split_idivmod): Likewise.
11772 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11773 (ix86_build_const_vector): Likewise.
11774 (ix86_build_signbit_mask): Likewise.
11775 (ix86_match_ccmode): Likewise.
11776 (ix86_cc_modes_compatible): Likewise.
11777 (ix86_expand_branch): Likewise.
11778 (ix86_expand_sse_cmp): Likewise.
11779 (ix86_expand_sse_movcc): Likewise.
11780 (ix86_expand_int_sse_cmp): Likewise.
11781 (ix86_expand_vec_perm_vpermi2): Likewise.
11782 (ix86_expand_vec_perm): Likewise.
11783 (ix86_expand_sse_unpack): Likewise.
11784 (ix86_expand_int_addcc): Likewise.
11785 (ix86_split_to_parts): Likewise.
11786 (ix86_vectorize_builtin_gather): Likewise.
11787 (ix86_vectorize_builtin_scatter): Likewise.
11788 (avx_vpermilp_parallel): Likewise.
11789 (inline_memory_move_cost): Likewise.
11790 (ix86_tieable_integer_mode_p): Likewise.
11791 (x86_maybe_negate_const_int): Likewise.
11792 (ix86_expand_vector_init_duplicate): Likewise.
11793 (ix86_expand_vector_init_one_nonzero): Likewise.
11794 (ix86_expand_vector_init_one_var): Likewise.
11795 (ix86_expand_vector_init_concat): Likewise.
11796 (ix86_expand_vector_init_interleave): Likewise.
11797 (ix86_expand_vector_init_general): Likewise.
11798 (ix86_expand_vector_set): Likewise.
11799 (ix86_expand_vector_extract): Likewise.
11800 (emit_reduc_half): Likewise.
11801 (ix86_emit_i387_round): Likewise.
11802 (ix86_mangle_type): Likewise.
11803 (ix86_expand_round_sse4): Likewise.
11804 (expand_vec_perm_blend): Likewise.
11805 (canonicalize_vector_int_perm): Likewise.
11806 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11807 (expand_vec_perm_1): Likewise.
11808 (expand_vec_perm_interleave3): Likewise.
11809 (expand_vec_perm_even_odd_pack): Likewise.
11810 (expand_vec_perm_even_odd_1): Likewise.
11811 (expand_vec_perm_broadcast_1): Likewise.
11812 (ix86_vectorize_vec_perm_const_ok): Likewise.
11813 (ix86_expand_vecop_qihi): Likewise.
11814 (ix86_expand_mul_widen_hilo): Likewise.
11815 (ix86_expand_sse2_abs): Likewise.
11816 (ix86_expand_pextr): Likewise.
11817 (ix86_expand_pinsr): Likewise.
11818 (ix86_preferred_simd_mode): Likewise.
11819 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11820 * config/i386/sse.md (*andnot<mode>3): Likewise.
11821 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11822 (*<code><mode>3): Likewise.
11823 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11824 (ia64_expand_atomic_op): Likewise.
11825 (ia64_arg_type): Likewise.
11826 (ia64_mode_to_int): Likewise.
11827 (ia64_scalar_mode_supported_p): Likewise.
11828 (ia64_vector_mode_supported_p): Likewise.
11829 (expand_vec_perm_broadcast): Likewise.
11830 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11831 (iq2000_function_arg_advance): Likewise.
11832 (iq2000_function_arg): Likewise.
11833 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11834 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11835 (m68k_libcall_value): Likewise.
11836 (m68k_function_value): Likewise.
11837 (sched_attr_op_type): Likewise.
11838 * config/mcore/mcore.c (mcore_output_move): Likewise.
11839 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11841 (microblaze_function_arg): Likewise.
11842 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11843 (mips_print_operand): Likewise.
11844 (mips_mode_ok_for_mov_fmt_p): Likewise.
11845 (mips_vector_mode_supported_p): Likewise.
11846 (mips_preferred_simd_mode): Likewise.
11847 (mips_expand_vpc_loongson_even_odd): Likewise.
11848 (mips_expand_vec_unpack): Likewise.
11849 (mips_expand_vi_broadcast): Likewise.
11850 (mips_expand_vector_init): Likewise.
11851 (mips_expand_vec_reduc): Likewise.
11852 (mips_expand_msa_cmp): Likewise.
11853 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11854 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11855 (cc_flags_for_mode): Likewise.
11856 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11857 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11858 (nds32_output_casesi_pc_relative): Likewise.
11859 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11860 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11861 (nvptx_gen_unpack): Likewise.
11862 (nvptx_gen_pack): Likewise.
11863 (nvptx_gen_shuffle): Likewise.
11864 (nvptx_gen_wcast): Likewise.
11865 (nvptx_preferred_simd_mode): Likewise.
11866 * config/pa/pa.c (pa_secondary_reload): Likewise.
11867 * config/pa/predicates.md (base14_operand): Likewise.
11868 * config/powerpcspe/powerpcspe-c.c
11869 (altivec_resolve_overloaded_builtin): Likewise.
11870 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11872 (rs6000_preferred_simd_mode): Likewise.
11873 (output_vec_const_move): Likewise.
11874 (rs6000_expand_vector_extract): Likewise.
11875 (rs6000_split_vec_extract_var): Likewise.
11876 (reg_offset_addressing_ok_p): Likewise.
11877 (rs6000_legitimate_offset_address_p): Likewise.
11878 (rs6000_legitimize_address): Likewise.
11879 (rs6000_emit_set_const): Likewise.
11880 (rs6000_const_vec): Likewise.
11881 (rs6000_emit_move): Likewise.
11882 (spe_build_register_parallel): Likewise.
11883 (rs6000_darwin64_record_arg_recurse): Likewise.
11884 (swap_selector_for_mode): Likewise.
11885 (spe_init_builtins): Likewise.
11886 (paired_init_builtins): Likewise.
11887 (altivec_init_builtins): Likewise.
11888 (do_load_for_compare): Likewise.
11889 (rs6000_generate_compare): Likewise.
11890 (rs6000_expand_float128_convert): Likewise.
11891 (emit_load_locked): Likewise.
11892 (emit_store_conditional): Likewise.
11893 (rs6000_output_function_epilogue): Likewise.
11894 (rs6000_handle_altivec_attribute): Likewise.
11895 (rs6000_function_value): Likewise.
11896 (emit_fusion_gpr_load): Likewise.
11897 (emit_fusion_p9_load): Likewise.
11898 (emit_fusion_p9_store): Likewise.
11899 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11900 (fusion_gpr_mem_load): Likewise.
11901 (fusion_addis_mem_combo_load): Likewise.
11902 (fusion_addis_mem_combo_store): Likewise.
11903 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11904 (fusion_gpr_mem_load): Likewise.
11905 (fusion_addis_mem_combo_load): Likewise.
11906 (fusion_addis_mem_combo_store): Likewise.
11907 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11909 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11910 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11911 (rs6000_preferred_simd_mode): Likewise.
11912 (output_vec_const_move): Likewise.
11913 (rs6000_expand_vector_extract): Likewise.
11914 (rs6000_split_vec_extract_var): Likewise.
11915 (reg_offset_addressing_ok_p): Likewise.
11916 (rs6000_legitimate_offset_address_p): Likewise.
11917 (rs6000_legitimize_address): Likewise.
11918 (rs6000_emit_set_const): Likewise.
11919 (rs6000_const_vec): Likewise.
11920 (rs6000_emit_move): Likewise.
11921 (rs6000_darwin64_record_arg_recurse): Likewise.
11922 (swap_selector_for_mode): Likewise.
11923 (paired_init_builtins): Likewise.
11924 (altivec_init_builtins): Likewise.
11925 (rs6000_expand_float128_convert): Likewise.
11926 (emit_load_locked): Likewise.
11927 (emit_store_conditional): Likewise.
11928 (rs6000_output_function_epilogue): Likewise.
11929 (rs6000_handle_altivec_attribute): Likewise.
11930 (rs6000_function_value): Likewise.
11931 (emit_fusion_gpr_load): Likewise.
11932 (emit_fusion_p9_load): Likewise.
11933 (emit_fusion_p9_store): Likewise.
11934 * config/rx/rx.c (rx_gen_move_template): Likewise.
11935 (flags_from_mode): Likewise.
11936 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11937 (s390_slb_comparison): Likewise.
11938 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11939 (s390_vector_mode_supported_p): Likewise.
11940 (s390_cc_modes_compatible): Likewise.
11941 (s390_match_ccmode_set): Likewise.
11942 (s390_canonicalize_comparison): Likewise.
11943 (s390_emit_compare_and_swap): Likewise.
11944 (s390_branch_condition_mask): Likewise.
11945 (s390_rtx_costs): Likewise.
11946 (s390_secondary_reload): Likewise.
11947 (__SECONDARY_RELOAD_CASE): Likewise.
11948 (s390_expand_cs): Likewise.
11949 (s390_preferred_simd_mode): Likewise.
11950 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11951 * config/sh/sh.c (sh_print_operand): Likewise.
11952 (dump_table): Likewise.
11953 (sh_secondary_reload): Likewise.
11954 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11955 * config/sh/sh.md (casesi_worker_1): Likewise.
11956 (casesi_worker_2): Likewise.
11957 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11958 (fcc_comparison_operator): Likewise.
11959 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11960 (emit_soft_tfmode_cvt): Likewise.
11961 (sparc_preferred_simd_mode): Likewise.
11962 (output_cbranch): Likewise.
11963 (sparc_print_operand): Likewise.
11964 (sparc_expand_vec_perm_bmask): Likewise.
11965 (vector_init_bshuffle): Likewise.
11966 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11967 (spu_vector_mode_supported_p): Likewise.
11968 (spu_expand_insv): Likewise.
11969 (spu_emit_branch_or_set): Likewise.
11970 (spu_handle_vector_attribute): Likewise.
11971 (spu_builtin_splats): Likewise.
11972 (spu_builtin_extract): Likewise.
11973 (spu_builtin_promote): Likewise.
11974 (spu_expand_sign_extend): Likewise.
11975 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11976 (tilegx_simd_int): Likewise.
11977 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11978 (tilepro_simd_int): Likewise.
11979 * config/v850/v850.c (const_double_split): Likewise.
11980 (v850_print_operand): Likewise.
11981 (ep_memory_offset): Likewise.
11982 * config/vax/vax.c (vax_rtx_costs): Likewise.
11983 (vax_output_int_move): Likewise.
11984 (vax_output_int_add): Likewise.
11985 (vax_output_int_subtract): Likewise.
11986 * config/visium/predicates.md (visium_branch_operator): Likewise.
11987 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
11988 (visium_print_operand_address): Likewise.
11989 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11990 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
11991 (xtensa_expand_conditional_branch): Likewise.
11992 (xtensa_copy_incoming_a7): Likewise.
11993 (xtensa_output_literal): Likewise.
11994 * dfp.c (decimal_real_maxval): Likewise.
11995 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11997 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11998 Alan Hayward <alan.hayward@arm.com>
11999 David Sherwood <david.sherwood@arm.com>
12001 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12002 (mode_nunits_inline): Likewise.
12003 (mode_inner_inline): Likewise.
12004 (mode_unit_size_inline): Likewise.
12005 (mode_unit_precision_inline): Likewise.
12006 (emit_insn_modes_h): Likewise. Also emit a #define of the
12008 (emit_mode_wider): Add an E_ prefix to mode names.
12009 (emit_mode_complex): Likewise.
12010 (emit_mode_inner): Likewise.
12011 (emit_mode_adjustments): Likewise.
12012 (emit_mode_int_n): Likewise.
12013 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12014 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12015 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12016 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12017 (CRC32_BUILTIN, ENTRY): Likewise.
12018 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12019 (aarch64_pop_regs): Likewise.
12020 (aarch64_process_components): Likewise.
12021 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12022 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12023 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12024 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12025 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12026 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12028 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12029 (pcmpistr): Likewise.
12030 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12031 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12032 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12034 * config/rl78/rl78.c (mduc_regs): Likewise.
12035 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12036 (htm_expand_builtin): Likewise.
12037 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12038 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12039 * config/xtensa/xtensa.c (print_operand): Likewise.
12040 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12041 (NUM_MODE_VECTOR_INT): Likewise.
12042 * genoutput.c (null_operand): Likewise.
12043 (output_operand_data): Likewise.
12044 * genrecog.c (print_parameter_value): Likewise.
12045 * lra.c (debug_operand_data): Likewise.
12047 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12049 * dwarf2out.c (scompare_loc_descriptor_wide)
12050 (scompare_loc_descriptor_narrow): New functions, split out from...
12051 (scompare_loc_descriptor): ...here.
12052 * expmed.c (emit_store_flag_int): New function, split out from...
12053 (emit_store_flag): ...here.
12055 2017-08-30 Richard Biener <rguenther@suse.de>
12057 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12058 (dwarf2out_early_finish): Move setting of AT_pubnames from
12059 early debug output to early finish.
12061 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12063 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12064 and -mdata-region to the assembler.
12066 2017-08-30 Richard Biener <rguenther@suse.de>
12068 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12070 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12071 (dwarf2out_early_global_decl): Only generate a DIE for the
12072 abstract origin if it doesn't already exist or is a declaration DIE.
12073 (resolve_addr): Do not add the linkage name twice when
12074 generating a stub DIE for the DW_TAG_GNU_call_site target.
12076 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12078 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12079 Use machine_mode rather than int for arg1_mode.
12081 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12084 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12085 that the second argument of the built-in functions to unpack
12086 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12087 switch statement instead a lot of if statements.
12088 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12089 Allow 64-bit values to be in Altivec registers as well as
12090 traditional floating point registers.
12091 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12093 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12095 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12096 MAX_REGS_PER_ADDRESS == 1.
12098 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12100 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12101 * config/i386/i386.c (ix86_option_override_internal): Simplify
12102 setting of opts->x_flag_entry.
12104 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12105 Jakub Jelinek <jakub@redhat.com>
12106 Richard Biener <rguenther@suse.de>
12108 PR tree-optimization/81503
12109 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12110 folded constant fits in the target type; reorder tests for clarity.
12112 2017-08-29 Martin Liska <mliska@suse.cz>
12114 * passes.def: Include pass_lower_switch.
12115 * stmt.c (dump_case_nodes): Remove and move to
12116 tree-switch-conversion.
12117 (case_values_threshold): Likewise.
12118 (expand_switch_as_decision_tree_p): Likewise.
12119 (emit_case_decision_tree): Likewise.
12120 (expand_case): Likewise.
12121 (balance_case_nodes): Likewise.
12122 (node_has_low_bound): Likewise.
12123 (node_has_high_bound): Likewise.
12124 (node_is_bounded): Likewise.
12125 (emit_case_nodes): Likewise.
12126 (struct simple_case_node): New struct.
12127 (add_case_node): Remove.
12128 (emit_case_dispatch_table): Use vector instead of case_list.
12129 (reset_out_edges_aux): Remove.
12130 (compute_cases_per_edge): Likewise.
12131 (expand_case): Build list of simple_case_node.
12132 (expand_sjlj_dispatch_table): Use it.
12133 * tree-switch-conversion.c (struct case_node): Moved from
12134 stmt.c and adjusted.
12135 (emit_case_nodes): Likewise.
12136 (node_has_low_bound): Likewise.
12137 (node_has_high_bound): Likewise.
12138 (node_is_bounded): Likewise.
12139 (case_values_threshold): Likewise.
12140 (reset_out_edges_aux): Likewise.
12141 (compute_cases_per_edge): Likewise.
12142 (add_case_node): Likewise.
12143 (dump_case_nodes): Likewise.
12144 (balance_case_nodes): Likewise.
12145 (expand_switch_as_decision_tree_p): Likewise.
12146 (emit_jump): Likewise.
12147 (emit_case_decision_tree): Likewise.
12148 (try_switch_expansion): Likewise.
12149 (do_jump_if_equal): Likewise.
12150 (emit_cmp_and_jump_insns): Likewise.
12151 (fix_phi_operands_for_edge): New function.
12152 (record_phi_operand_mapping): Likewise.
12153 (class pass_lower_switch): New pass.
12154 (pass_lower_switch::execute): New function.
12155 (make_pass_lower_switch): Likewise.
12156 (conditional_probability):
12157 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12158 * tree-pass.h: Add make_pass_lower_switch.
12160 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12163 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12166 2017-08-29 Richard Biener <rguenther@suse.de>
12168 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12169 we do not add a DW_AT_inline attribute twice.
12170 (gen_subprogram_die): Remove code setting DW_AT_inline on
12171 DECL_ABSTRACT_P nodes.
12173 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
12175 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
12176 calls to internal functions.
12177 (gimplify_modify_expr): Likewise.
12178 * tree-call-cdce.c (use_internal_fn): Likewise.
12179 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
12180 (convert_to_divmod): Set the nothrow flag.
12181 * tree-if-conv.c (predicate_mem_writes): Likewise.
12182 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12183 (vectorizable_call): Likewise.
12184 (vectorizable_store): Likewise.
12185 (vectorizable_load): Likewise.
12186 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
12187 (vect_recog_mask_conversion_pattern): Likewise.
12189 2017-08-29 Martin Liska <mliska@suse.cz>
12192 * gcc.c (driver_handle_option): Add new argument.
12193 * opts-common.c (handle_option): Pass
12194 target_option_override_hook.
12195 * opts-global.c (lang_handle_option): Add new option.
12196 (set_default_handlers): Add new argument.
12197 (decode_options): Likewise.
12198 * opts.c (target_handle_option): Likewise.
12199 (common_handle_option): Call target_option_override_hook.
12200 * opts.h (struct cl_option_handler_func): Add hook for
12201 target option override.
12202 (struct cl_option_handlers): Likewise.
12203 (set_default_handlers): Add new argument.
12204 (decode_options): Likewise.
12205 (common_handle_option): Likewise.
12206 (target_handle_option): Likewise.
12207 * toplev.c (toplev::main): Pass targetm.target_option.override
12210 2017-08-29 Richard Biener <rguenther@suse.de>
12211 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
12213 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
12214 life to the active subtree.
12216 2017-08-28 Jeff Law <law@redhat.com>
12218 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
12219 derive_equivalences.
12220 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
12221 Code moved into....
12222 (edge_info::derive_equivalences): New private member function
12224 * tree-ssa-dom.c (class edge_info): Changed from a struct
12225 to a class. Add ctor/dtor, methods and data members.
12226 (edge_info::edge_info): Renamed from allocate_edge_info.
12227 Initialize additional members.
12228 (edge_info::~edge_info): New.
12229 (free_dom_edge_info): Delete the edge info.
12230 (record_edge_info): Use new class & associated member functions.
12231 Tighten forms for testing for edge equivalences.
12232 (record_temporary_equivalences): Iterate over the simple
12233 equivalences rather than assuming there's only one per edge.
12234 (cprop_into_successor_phis): Iterate over the simple
12235 equivalences rather than assuming there's only one per edge.
12236 (optimize_stmt): Use operand_equal_p rather than pointer
12237 equality for mini-DSE code.
12239 2017-08-28 Nathan Sidwell <nathan@acm.org>
12241 * gcc.c (execute): Fold SIGPIPE handling into switch
12242 statement. Adjust internal error message.
12244 2017-08-28 Richard Biener <rguenther@suse.de>
12247 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12248 Do nothing for removed DIEs.
12250 2017-08-28 Richard Biener <rguenther@suse.de>
12252 PR tree-optimization/81977
12253 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12256 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12259 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12260 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12261 using targetm.gen_mem_thread_fence.
12263 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12266 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12267 predicate to register_operand. Reorder operands.
12268 (*btr<mode>): Ditto.
12269 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12270 (*btr<mode>_mask): Ditto.
12272 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12274 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12275 * config/rs6000/xmmintrin.h: New file.
12276 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12278 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12281 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12282 parameter and_insn and return it.
12283 (recombine_lvx_pattern): Insert a copy to ensure availability of
12284 the base register of the copied masking operation at the point of
12285 the instruction replacement.
12286 (recombine_stvx_pattern): Likewise.
12288 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12290 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12291 undocumented switches.
12292 (-mpower9-dform-vector): Likewise.
12293 (-mpower9-dform): Likewise.
12294 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12295 comments to delete references to -mpower9-dform* switches.
12296 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12297 Delete reference to -mpower9-dform* switches, test for
12298 -mpower9-vector instead.
12299 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12300 (OTHER_P9_VECTOR_MASKS): Likewise.
12301 (POWERPC_MASKS): Likewise.
12302 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12303 tests against -mpower9-dform* to -mpower9-vector. Delete code
12304 that checked for -mpower9-dform* consistancy with other options.
12305 Add test for -mpower9-misc to enable other power9 switches.
12306 (rs6000_init_hard_regno_mode_ok): Likewise.
12307 (rs6000_option_override_internal): Likewise.
12308 (rs6000_emit_prologue): Likewise.
12309 (rs6000_emit_epilogue): Likewise.
12310 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12311 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12312 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12314 (emit_fusion_p9_store): Likewise.
12315 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12316 resetting these macros if the assembler does not support ISA 3.0
12318 (TARGET_P9_DFORM_VECTOR): Likewise.
12319 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12320 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12322 2017-08-25 Alan Modra <amodra@gmail.com>
12324 PR rtl-optimization/81747
12325 * cse.c (cse_extended_basic_block): Don't attempt to record
12326 equivalences for degenerate conditional branches that branch
12327 to their fall-through.
12329 2017-08-24 Martin Sebor <msebor@redhat.com>
12331 PR middle-end/81908
12332 * gimple-fold.c (size_must_be_zero_p): New function.
12333 (gimple_fold_builtin_memory_op): Call it.
12335 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12337 * config/rs6000/mm_malloc.h: New file.
12339 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12341 PR tree-optimization/81913
12342 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12343 analysis when either IVs in condition can wrap.
12345 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12347 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12348 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12350 2017-08-24 Richard Biener <rguenther@suse.de>
12353 * targhooks.c (default_target_can_inline_p): Properly
12354 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12355 is present and always compare.
12356 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12357 infer -mfpmath=sse from TARGET_SSE_P.
12358 (ix86_can_inline_p): Properly use target_option_default_node when
12359 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12361 2017-08-24 Richard Biener <rguenther@suse.de>
12364 * dwarf2out.c (output_die): Handle flag_generate_offload like
12366 (output_comp_unit): Likewise.
12367 (gen_array_type_die): Likewise.
12368 (dwarf2out_early_finish): Likewise.
12369 (note_variable_value_in_expr): Likewise.
12370 (dwarf2out_finish): Likewise. Adjust assert.
12371 * cgraphunit.c (symbol_table::compile): Move setting of
12372 flag_generate_offload earlier ...
12373 (symbol_table::finalize_compilation_unit): ... here, before
12374 early debug finalization.
12376 2017-08-24 Richard Biener <rguenther@suse.de>
12378 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12379 and ipa-fnsummary.h.
12380 (ix86_can_inline_p): When ix86_fpmath flags do not match
12381 check whether the callee uses FP math at all.
12383 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12385 PR middle-end/81931
12386 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12387 instead of TYPE_PRECISION.
12389 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12390 Alan Hayward <alan.hayward@arm.com>
12391 David Sherwood <david.sherwood@arm.com>
12393 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12395 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12397 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12399 (ix86_frame::stack_realign_allocate): New field.
12400 (struct machine_frame_state): Modify comments.
12401 (machine_frame_state::sp_realigned_fp_end): New field.
12402 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12403 layout calculation.
12404 (sp_valid_at): Add assertion to assure no attempt to access invalid
12405 offset of a realigned stack.
12406 (fp_valid_at): Likewise.
12407 (choose_baseaddr): Modify comments.
12408 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12409 ix86_expand_prologue.
12410 (ix86_expand_prologue): Modify stack realignment and allocation.
12411 (ix86_expand_epilogue): Modify comments.
12412 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12413 avx2_runtime, avx512f, and avx512f_runtime.
12415 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12417 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12418 (mstackrealign): Do not init to -1.
12419 * config/i386/i386.c (ix86_option_override_internal):
12420 Check opts_set, not opts when setting default value of
12421 opts->x_ix86_force_align_arg_pointer.
12423 2017-08-23 Richard Biener <rguenther@suse.de>
12425 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12426 lang_hooks.decl_printable_name.
12427 * print-rtl-function.c (print_rtx_function): Likewise.
12428 * tree-pretty-print.c (dump_function_header): Likewise.
12430 2017-08-23 Richard Biener <rguenther@suse.de>
12433 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12434 -g0 at compile-time.
12436 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12438 PR middle-end/19706
12439 * doc/sourcebuild.texi (Other hardware attributes):
12442 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12444 PR middle-end/19706
12445 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12446 Fix single-use check.
12448 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12450 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12452 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12454 * doc/install.texi: Modify to add more details on running selected
12457 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12459 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12460 is combined with -mabi=ms.
12461 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12464 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12466 PR tree-optimization/81488
12467 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12468 and cached_basis fields.
12469 (MAX_SPREAD): New constant.
12470 (alloc_cand_and_find_basis): Initialize new fields.
12471 (clear_visited): New function.
12472 (create_phi_basis_1): Rename from create_phi_basis, set visited
12473 and cached_basis fields.
12474 (create_phi_basis): New wrapper function.
12475 (phi_add_costs_1): Rename from phi_add_costs, add spread
12476 parameter, set visited field, short-circuit when limits reached.
12477 (phi_add_costs): New wrapper function.
12478 (record_phi_increments_1): Rename from record_phi_increments, set
12480 (record_phi_increments): New wrapper function.
12481 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12482 (phi_incr_cost): New wrapper function.
12483 (all_phi_incrs_profitable_1): Rename from
12484 all_phi_incrs_profitable, set visited field.
12485 (all_phi_incrs_profitable): New wrapper function.
12487 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12488 Alan Hayward <alan.hayward@arm.com>
12489 David Sherwood <david.sherwood@arm.com>
12491 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12492 that takes the outer and inner modes.
12493 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12494 comparison as the canonical test for a paradoxical subreg.
12495 * combine.c (simplify_set): Use paradoxical_subreg_p.
12496 (make_extraction): Likewise.
12497 (force_to_mode): Likewise.
12498 (rtx_equal_for_field_assignment_p): Likewise.
12499 (gen_lowpart_for_combine): Likewise.
12500 (simplify_comparison): Likewise.
12501 * cse.c (equiv_constant): Likewise.
12502 * expmed.c (store_bit_field_1): Likewise.
12503 * final.c (alter_subreg): Likewise.
12504 * fwprop.c (propagate_rtx): Likewise.
12505 (forward_propagate_subreg): Likewise.
12506 * ira-conflicts.c (ira_build_conflicts): Likewise.
12507 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12508 * lra-constraints.c (curr_insn_transform): Likewise.
12509 (split_reg): Likewise.
12510 * lra-eliminations.c (move_plus_up): Likewise.
12511 (lra_eliminate_regs_1): Likewise.
12512 * recog.c (general_operand): Likewise.
12513 * ree.c (combine_reaching_defs): Likewise.
12514 * reload.c (push_reload): Likewise.
12515 (find_reloads): Likewise.
12516 * reload1.c (elimination_effects): Likewise.
12517 (compute_reload_subreg_offset): Likewise.
12518 (choose_reload_regs): Likewise.
12519 * rtlanal.c (subreg_lsb_1): Likewise.
12520 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12521 (simplify_subreg): Likewise.
12522 * var-tracking.c (track_loc_p): Likewise.
12523 * emit-rtl.c (byte_lowpart_offset): Likewise.
12524 (paradoxical_subreg_p): Delete out-of-line definition.
12526 2017-08-22 Jeff Law <law@redhat.com>
12528 PR tree-optimization/81741
12529 PR tree-optimization/71947
12530 * tree-ssa-dom.c: Include tree-inline.h.
12531 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12532 equivalences if one is more expensive to compute than the other.
12533 * tree-ssa-scopedtables.h (class const_or_copies): Make
12534 record_const_or_copy_raw method private.
12535 (class avail_exprs_stack): New method simplify_binary_operation.
12536 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12537 avail_exprs_stack::simplify_binary_operation as needed.
12538 (avail_exprs_stack::simplify_binary_operation): New function.
12540 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12542 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12543 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12544 (DOT_SYMBOLS): Likewise.
12545 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12546 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12547 (RS6000_ABI_NAME): Likewise.
12548 (TARGET_CMODEL): Likewise.
12549 (TOC_SECTION_ASM_OP): Likewise.
12550 (SET_CMODEL): New macro.
12551 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12553 2017-08-22 Richard Biener <rguenther@suse.de>
12555 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12556 to allow for free-lang-data replacements similar to verify_type_variant.
12558 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12559 Alan Hayward <alan.hayward@arm.com>
12560 David Sherwood <david.sherwood@arm.com>
12562 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12563 VOIDmode for the LABEL_REF.
12565 2017-08-22 Richard Biener <rguenther@suse.de>
12567 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12569 2017-08-22 Richard Biener <rguenther@suse.de>
12571 * common.opt (feliminate-dwarf2-dups): Ignore.
12572 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12573 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12574 same_die_p_wrap, compute_section_prefix,
12575 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12576 (comdat_symbol_id, comdat_symbol_number): Likewise.
12577 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12579 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12580 (output_die): Mark unreachable path unreachable.
12581 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12582 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12583 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12584 (dwarf2out_early_finish): Likewise.
12586 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12588 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12590 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12593 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12594 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12595 (avr_attribute_table) <io, io_low, address>: Initialize
12596 .decl_required with true.
12598 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12600 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12601 undocumented debugging options.
12602 (-mvsx-scalar-double): Likewise.
12603 (-mallow-df-permute): Likewise.
12604 (-mvectorize-builtins): Likewise.
12605 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12606 (rs6000_builtin_vectorized_function): Likewise.
12607 (rs6000_builtin_md_vectorized_function): Likewise.
12608 (rs6000_opt_vars): Likewise.
12610 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12613 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12614 (*btrq_imm): Rename from *btrq.
12615 (*btcq_imm): Rename from *btcq.
12616 (btsc): New code attribute.
12617 (*<btsc><mode>): New insn pattern.
12618 (*btr<mode>): Ditto.
12619 (*<btsc><mode>_mask): New insn_and_split pattern.
12620 (*btr<mode>_mask): Ditto.
12622 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12623 Alan Hayward <alan.hayward@arm.com>
12624 David Sherwood <david.sherwood@arm.com>
12626 * function.c (pad_below): Simplify padding calculation.
12628 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12629 Alan Hayward <alan.hayward@arm.com>
12630 David Sherwood <david.sherwood@arm.com>
12632 * target.def (function_prologue): Remove frame size argument.
12633 (function_epilogue): Likewise.
12634 * doc/tm.texi: Regenerate.
12635 * final.c (final_start_function): Update call to function_prologue.
12636 (final_end_function): Update call to function_epilogue.
12637 (default_function_pro_epilogue): Remove frame size argument.
12638 * output.h (default_function_pro_epilogue): Likewise.
12639 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12640 (arm_output_function_prologue): Likewise.
12641 * config/frv/frv.c (frv_function_prologue): Likewise.
12642 (frv_function_epilogue): Likewise.
12643 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12644 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12645 (ia64_output_function_epilogue): Likewise.
12646 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12647 (m32r_output_function_epilogue): Likewise.
12648 * config/microblaze/microblaze.c (microblaze_function_prologue)
12649 (microblaze_function_epilogue): Likewise.
12650 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12651 (mips_output_function_epilogue): Likewise.
12652 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12653 (mmix_target_asm_function_epilogue): Likewise.
12654 * config/msp430/msp430.c (msp430_start_function): Likewise.
12655 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12656 (nds32_asm_function_epilogue): Likewise.
12657 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12658 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12659 (pa_output_function_epilogue): Likewise.
12660 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12661 (rs6000_output_function_epilogue): Likewise.
12662 * config/rl78/rl78.c (rl78_start_function): Likewise.
12663 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12664 (rs6000_output_function_epilogue): Likewise.
12665 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12666 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12667 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12668 (sparc_asm_function_epilogue): Likewise.
12670 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12672 * tree.h (type_has_mode_precision_p): New function.
12673 * convert.c (convert_to_integer_1): Use it.
12674 * expr.c (expand_expr_real_2): Likewise.
12675 (expand_expr_real_1): Likewise.
12676 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
12677 * match.pd: Likewise.
12678 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
12679 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
12680 * tree-tailcall.c (process_assignment): Likewise.
12681 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12682 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
12683 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
12684 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12685 (vectorizable_assignment): Likewise.
12686 (vectorizable_shift): Likewise.
12687 (vectorizable_operation): Likewise.
12688 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12690 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12692 * match.pd: Add pow (C, x) simplification.
12694 2017-08-21 Richard Biener <rguenther@suse.de>
12696 PR tree-optimization/81900
12697 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12698 for blocks with abnormal predecessors.
12699 (compute_antic): Do not set visited flag prematurely.
12701 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12704 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12706 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12708 * stor-layout.h (vector_type_mode): Move to...
12709 * tree.h (vector_type_mode): ...here.
12710 * stor-layout.c (vector_type_mode): Move to...
12711 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12713 2017-08-21 Richard Biener <rguenther@suse.de>
12715 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12716 register_external_die hooks.
12717 (debug_false_tree_charstarstar_uhwistar): Declare.
12718 (debug_nothing_tree_charstar_uhwi): Likewise.
12719 * debug.c (do_nothing_debug_hooks): Adjust.
12720 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12721 (debug_nothing_tree_charstar_uhwi): Likewise.
12722 * dbxout.c (dbx_debug_hooks): Adjust.
12723 (xcoff_debug_hooks): Likewise.
12724 * sdbout.c (sdb_debug_hooks): Likewise.
12725 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12726 * dwarf2out.c (macinfo_label_base): New global.
12727 (dwarf2out_register_external_die): New function for the
12728 register_external_die hook.
12729 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12730 (dwarf2_debug_hooks): Use them.
12731 (dwarf2_lineno_debug_hooks): Adjust.
12732 (struct die_struct): Add with_offset flag.
12733 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12734 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12735 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12736 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12737 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12738 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12739 defining section names for the early LTO debug variants.
12740 (reset_indirect_string): New helper.
12741 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12742 (print_dw_val): Add support for offsetted symbol references.
12743 (get_ultimate_context): Split out from is_cxx.
12744 (is_cxx): Use get_ultimate_context.
12745 (is_fortran): Add decl overload.
12746 (compute_comp_unit_symbol): Split out worker from
12747 compute_section_prefix.
12748 (compute_section_prefix): Call compute_comp_unit_symbol and
12749 set comdat_type_p here.
12750 (output_die): Skip DIE symbol output for the LTO added one.
12751 Handle DIE symbol references with offset.
12752 (output_comp_unit): Guard section name mangling properly.
12753 For LTO debug sections emit a symbol at the section beginning
12754 which we use to refer to its DIEs.
12755 (add_abstract_origin_attribute): For DIEs registered via
12756 dwarf2out_register_external_die directly refer to the early
12757 DIE rather than indirectly through the shadow one we created.
12758 Remove obsolete call to dwarf2out_abstract_function for
12759 non-function/block origins.
12760 (gen_array_type_die): When generating early LTO debug do
12761 not emit DW_AT_string_length.
12762 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12763 late when in LTO. As suggested place a gcc_unreachable for
12764 the DECL_ABSTRACT_P case.
12765 (gen_subprogram_die): Avoid another specification DIE
12766 for early built declarations/definitions for the late LTO case.
12767 (gen_variable_die): Add type references for late duplicated VLA dies
12769 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12770 we have the abstract instance already.
12771 (process_scope_var): Adjust decl DIE contexts in LTO which
12772 first puts them in limbo.
12773 (gen_decl_die): Do not generate type DIEs late apart from
12774 types for VLAs or for decls we do not yet have a DIE. Do not
12775 call dwarf2out_abstract_function late.
12776 (dwarf2out_early_global_decl): Make sure to create DIEs
12777 for abstract instances of a decl first.
12778 (dwarf2out_late_global_decl): Adjust comment.
12779 (output_macinfo_op): With multiple macro sections use
12780 macinfo_label_base to distinguish labels.
12781 (output_macinfo): Likewise. Update macinfo_label_base.
12782 Pass in the line info label.
12783 (note_variable_value_in_expr): When generating LTO resolve
12784 all variable values here by generating DIEs as needed.
12785 (init_sections_and_labels): Add early LTO debug flag parameter
12786 and generate different sections and names if set. Add generation
12787 counter for the labels so we can have multiple of them.
12788 (reset_dies): Helper to allow DIEs to be output multiple times.
12789 (dwarf2out_finish): When outputting DIEs to the fat part of an
12790 LTO object first reset DIEs.
12791 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12792 (modified_type_die): Check for decl_ultimate_origin being self
12794 (gen_type_die_with_usage): Likewise.
12795 (gen_typedef_die): Allow decl_ultimate_origin being self.
12796 (set_decl_abstract_flags): Remove.
12797 (set_block_abstract_flags): Likewise.
12798 (dwarf2out_abstract_function): Treat the early generated DIEs
12799 as the abstract copy and only add DW_AT_inline and
12800 DW_AT_artificial here and call set_decl_origin_self.
12801 If the DIE has an abstract origin don't do anything.
12802 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12803 if we have none yet (Go fails to build one, PR78628).
12804 (variably_modified_type_p): Prevent endless recursion for Ada
12805 cyclic pointer types.
12806 * lto-streamer-in.c: Include debug.h.
12807 (dref_queue): New global.
12808 (lto_read_tree_1): Stream in DIE references.
12809 (lto_input_tree): Register DIE references.
12810 (input_function): Stream DECL_DEBUG_ARGS.
12811 * lto-streamer-out.c: Include debug.h.
12812 (lto_write_tree_1): Output DIE references.
12813 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12814 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12815 (output_function): Stream DECL_DEBUG_ARGS.
12816 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12817 Stream DECL_ABSTRACT_ORIGIN.
12818 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12819 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12820 DECL_CONTEXT for file-scope decls.
12821 * lto-streamer.h (struct dref_entry): Declare.
12822 (dref_queue): Likewise.
12823 * cfgexpand.c (pass_expand::execute): Do not call the
12824 outlining_inline_function hook here.
12825 * lto-wrapper.c (debug_obj): New global.
12826 (tool_cleanup): Unlink it if required.
12827 (debug_objcopy): New function.
12828 (run_gcc): Handle early debug sections in the IL files by
12829 extracting them to separate files, partially linkin them and
12830 feeding the result back as result to the linker.
12831 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12832 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12833 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12834 sections into a separate segment.
12835 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12837 (darwin_asm_dwarf_section): Likewise.
12838 (darwin_asm_output_dwarf_offset): Likewise.
12839 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12841 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12842 Alan Hayward <alan.hayward@arm.com>
12843 David Sherwood <david.sherwood@arm.com>
12845 * read-md.h (md_reader::record_potential_iterator_use): Replace
12846 pointer argument with an rtx and an index.
12847 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12848 (apply_mode_iterator): Likewise.
12849 (apply_code_iterator): Likewise.
12850 (apply_int_iterator): Likewise.
12851 (apply_subst_iterator): Likewise.
12852 (record_iterator_use): Likewise.
12853 (record_attribute_use): Likewise.
12854 (md_reader::record_potential_iterator_use): Likewise. Update calls
12855 to record_iterator_use and apply_iterator.
12856 (iterator_use): Replace ptr with x and index.
12857 (attribute_use): Likewise.
12858 (apply_attribute_uses): Update calls to apply_iterator.
12859 (apply_iterators): Likewise. Update initialization of iterator_use.
12860 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12861 and record_potential_iterator_use.
12862 (rtx_reader::read_rtx_operand): Likewise.
12864 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12865 Alan Hayward <alan.hayward@arm.com>
12866 David Sherwood <david.sherwood@arm.com>
12868 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12871 2017-08-21 Richard Biener <rguenther@suse.de>
12873 PR middle-end/81884
12874 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12875 at struct end conservatively when comparing common bases.
12877 2017-08-21 Richard Biener <rguenther@suse.de>
12879 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12880 (mem_ref_in_stmt): Remove.
12881 (determine_max_movement): Use ref index to get at the reference.
12882 (invariantness_dom_walker::before_dom_children): Deal with
12883 lim data already initialized.
12884 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12886 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12888 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12889 (TARGET_ISA_ROUND): Ditto.
12890 (TARGET_ROUND): Ditto.
12891 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12892 * config/i386/i386.md: Ditto.
12893 * config/i386/sse.md: Ditto.
12894 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12895 with OPTION_MASK_ISA_SSE4_1.
12897 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12900 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12901 __builtin_ia32_lzcnt_u16.
12903 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12906 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12907 (rs6000_set_current_function): Rewrite function to use it.
12909 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12912 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12913 and TYPE_WARN_IF_NOT_ALIGN.
12914 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12915 (handle_warn_if_not_align): New.
12916 (place_union_field): Call handle_warn_if_not_align.
12917 (place_field): Call handle_warn_if_not_align.
12918 Copy TYPE_WARN_IF_NOT_ALIGN.
12919 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12920 (layout_type): Likewise.
12921 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12923 (tree_decl_common): Add warn_if_not_align.
12924 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12925 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12926 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12927 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12928 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12929 * doc/extend.texi: Document warn_if_not_aligned attribute.
12930 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12932 2017-08-17 Martin Liska <mliska@suse.cz>
12935 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12936 (get_data_dependence): Use it as pointer type.
12937 (distribute_loop): Likewise.
12939 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12941 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12942 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12943 (p8_vmrgow_<mode>_direct): New define_insn.
12944 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12945 handle endianness for vmrgew and vmrgow permute patterns.
12947 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12949 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12950 * config/rs6000/rs6000-cpus.def: Remove comment.
12951 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12952 (POWERPC_MASKS): Likewise.
12953 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12954 use of TARGET_VSX_TIMODE.
12955 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12956 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12957 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12958 (rs6000_option_override_internal): Remove dead code.
12959 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12960 (rs6000_legitimize_reload_address): Likewise.
12961 (rs6000_legitimate_address_p): Likewise.
12962 (rs6000_opt_masks): Delete "vsx-timode".
12963 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12964 from function comment.
12965 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12966 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12967 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12969 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12970 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12971 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12972 (VSX_TI): Likewise.
12974 (define_peephole2): Likewise.
12976 2017-08-17 Martin Sebor <msebor@redhat.com>
12979 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12980 past the end of an array.
12981 (test_pp_format): Add test cases.
12983 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
12985 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
12986 missing ECF_NOTHROW flags.
12988 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12991 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
12992 operands residing in integer registers.
12993 (*vsx_le_perm_load_<mode>): Likewise.
12994 (*vsx_le_perm_store_<mode>): Likewise.
12995 (define_peephole2): Add peepholes to optimize the above.
12997 2017-08-17 Marek Polacek <polacek@redhat.com>
12999 PR middle-end/81814
13000 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13001 to mimic what shorten_compare did. Change the return type to bool.
13002 (fold_cond_expr_with_comparison): Update call to
13003 operand_equal_for_comparison_p.
13004 (fold_ternary_loc): Likewise.
13006 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13008 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13010 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13012 2017-08-17 Richard Biener <rguenther@suse.de>
13014 * tree-ssa-structalias.c (solve_graph): When propagating
13015 to successors update the graphs succ edges and avoid duplicate work.
13017 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13020 * config/i386/i386.c (ix86_option_override_internal): Save target
13021 specific options after ix86_stack_protector_guard_reg was changed.
13023 2017-08-17 Richard Biener <rguenther@suse.de>
13025 PR tree-optimization/81827
13026 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13027 (new_var_info): Initialize it conservatively.
13028 (get_call_vi): Mark register vars.
13029 (new_scalar_tmp_constraint_exp): Likewise.
13030 (handle_rhs_call): Likewise.
13031 (handle_const_call): Likewise.
13032 (create_function_info_for): Likewise.
13033 (solve_constraints): Sort varinfos to separate register from
13034 non-register vars to pack points-to solution bitmaps during
13037 2017-08-17 Marek Polacek <polacek@redhat.com>
13039 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13041 2017-08-17 Richard Biener <rguenther@suse.de>
13043 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13044 to true when overflow is undefined and we saturated the result.
13046 2017-08-17 Alan Modra <amodra@gmail.com>
13049 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13050 Don't use store multiple if only one reg needs saving.
13051 (interesting_frame_related_regno): New function.
13052 (rs6000_frame_related): Don't emit frame info for regs that
13054 (rs6000_emit_epilogue): Likewise.
13056 2017-08-16 Nathan Sidwell <nathan@acm.org>
13058 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13059 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13060 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13061 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13062 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13063 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13064 (verify_type): Adjust for TYPE_BINFO move.
13065 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13066 process TYPE_BINFO directly.
13067 (hash_tree): Likewise.
13068 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13070 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13073 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13075 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13077 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13080 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13081 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13082 Add ix86_binary_operator_ok to insn constraint.
13083 (*iordi_1_bts): Ditto.
13084 (*xordi_1_btc): Ditto.
13085 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13086 Update corresponding peephole2 pattern.
13090 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13092 PR tree-optimization/81832
13093 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13094 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13096 2017-08-16 Marek Polacek <polacek@redhat.com>
13099 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13100 regarding size_zero_node.
13102 2017-08-16 Martin Liska <mliska@suse.cz>
13105 * config.gcc: Respect previously set extra_objs in case
13108 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13110 PR tree-optimization/81835
13111 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13112 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13113 not depend on the phi.
13115 2017-08-16 Alan Modra <amodra@gmail.com>
13117 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13120 2017-08-16 Alan Modra <amodra@gmail.com>
13122 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13123 (save_reg_p): ..into this. Update all callers.
13124 (first_reg_to_save): Simplify.
13126 2017-08-16 Alan Modra <amodra@gmail.com>
13128 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13131 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13135 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13136 constant count if that count is less than 32.
13138 2017-08-15 Nathan Sidwell <nathan@acm.org>
13140 * gcc.c (execute): Emit friendlier message if inferior is killed
13141 by an external cause.
13143 2017-08-15 Richard Biener <rguenther@suse.de>
13145 PR tree-optimization/81790
13146 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13147 CONSTRUCTORs from simplifying and VN.
13149 2017-08-14 Martin Sebor <msebor@redhat.com>
13151 * builtin-attrs.def: Add comments.
13153 2017-08-14 Martin Sebor <msebor@redhat.com>
13156 * doc/extend.texi (attribute nonstring): Document new attribute.
13158 2017-08-14 Martin Sebor <msebor@redhat.com>
13161 * tree-diagnostic.c (default_tree_printer): Handle %G.
13162 * gimple-pretty-print.h (percent_G_format): Declare new function.
13163 * gimple-pretty-print.c (percent_G_format): Define.
13164 * tree-pretty-print.c (percent_K_format): Add argument.
13166 2017-08-14 Martin Sebor <msebor@redhat.com>
13168 PR translation/79998
13169 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13170 Remove a stray space.
13172 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
13175 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
13176 (*iordi_1_bts): Ditto.
13177 (*xordi_1_btc): Ditto.
13179 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13182 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
13183 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13185 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
13186 quoted strings, and make more translator-friendly.
13187 (darwin_rs6000_override_options): Likewise.
13188 (rs6000_option_override_internal): Likewise.
13189 (rs6000_return_in_memory): Fix overlong line.
13190 (init_cmulative_args): Use quoted strings, and make more
13191 translator-friendly.
13192 (rs6000_pass_by_reference): Fix overlong line.
13193 (def_builtin): Use quoted strings.
13194 (altivec_expand_predicate_builtin): Use quoted strings, and make
13195 more translator-friendly.
13196 (htm_expand_builtin): Use quoted strings.
13197 (cpu_expand_builtin): Use quoted strings, and make more
13198 translator-friendly.
13199 (altivec_expand_builtin): Likewise.
13200 (paired_expand_predicate_builtin): Likewise.
13201 (rs6000_invalid_builtin): Likewise.
13202 (builtin_function_type): Use quoted strings.
13203 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
13204 more translator-friendly.
13205 (rs6000_trampoline_init): Likewise.
13206 (rs6000_handle_altivec_attribute): Likewise.
13207 (rs6000_inner_target_options): Use quoted strings.
13208 (rs6000_disable_incompatible_switches): Likewise.
13209 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
13210 strings, and make more translator-friendly.
13211 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
13213 2017-08-14 Bin Cheng <bin.cheng@arm.com>
13215 PR tree-optimization/81799
13216 * tree-loop-distribution.c (version_loop_by_alias_check): Force
13217 cond_expr to simple gimple operand.
13219 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
13221 PR middle-end/46932
13222 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
13224 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
13228 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
13229 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
13230 TARGET_GASISR_PROLOGUES.
13231 * config/avr/avr.c (avr_option_override): Same.
13232 (avr_pass_pre_proep::execute): Same.
13234 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13237 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13238 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13240 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13242 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13243 define_insn_and_split. Split to a memory load from 0 in
13244 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13245 using PTR mode iterator.
13246 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13247 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13248 (*add_tp_<mode>): Redefine as define_insn_and_split.
13249 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13250 address space. Merge with *add_tp_x32 using PTR mode iterator.
13251 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13252 Split to an add with a memory load from 0 in
13253 DEFAULT_TLS_SEG_REG address space.
13255 2017-08-12 Andrew Pinski <apinski@cavium.com>
13257 * config/aarch64/aarch64-option-extensions.def (rdma):
13258 Fix feature string to what Linux prints out in /proc/cpuinfo.
13260 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13263 * dwarf2out.c (modified_type_die): For C typedef types that have
13264 an ultimate origin, process the ultimate origin instead of the
13266 (gen_typedef_die): Assert that input DECLs have no ultimate
13268 (gen_type_die_with_usage): For typedef variants that have an
13269 ultimate origin, just call gen_decl_die on the original DECL.
13270 (process_scope_var): Avoid creating DIEs for local typedefs and
13271 concrete static variables.
13273 2017-08-12 Alan Modra <amodra@gmail.com>
13277 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13278 match gnu-user.h startfile.
13279 (ENDFILE_LINUX_SPEC): Similarly.
13281 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13284 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13286 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13288 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13289 * config/aarch64/aarch64.md (mov<mode>): Change.
13290 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13291 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13292 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13294 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13296 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13297 for storage order barriers.
13299 2017-08-11 Martin Liska <mliska@suse.cz>
13302 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13303 variables of void type.
13305 2017-08-11 Martin Liska <mliska@suse.cz>
13307 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13308 TARGET_SUPPORTS_ALIASES.
13309 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13310 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13311 (optimize_weakref): Likewise.
13312 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13313 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13314 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13316 2017-08-11 Martin Liska <mliska@suse.cz>
13319 * config/i386/i386.c (make_resolver_func): Do complete
13320 refactoring of the function.
13322 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13325 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13326 * config/i386/i386.c (ix86_stack_protect_guard): Use
13327 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13328 * doc/invoke.texi (x86 Options): Document
13329 -mstack-protector-guard-symbol= option.
13331 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13333 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13334 * config/i386/i386.c (ix86_split_stack_guard): New function.
13335 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13336 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13337 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13338 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13339 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13340 (split_stack_space_check): Call ix86_split_stack_guard.
13342 2017-08-10 Martin Sebor <msebor@redhat.com>
13344 * print-tree.c (print_node): Print location using the established
13346 Replace spaces with colons.
13347 (debug_raw, debug): Ditto.
13349 2017-08-10 Martin Sebor <msebor@redhat.com>
13352 * pretty-print.c (pp_format): Correct the handling of %s precision.
13354 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13357 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13359 (ix86_finalize_stack_frame_flags): This. Also clear
13360 frame_pointer_needed if -fno-omit-frame-pointer is used without
13362 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13363 with ix86_finalize_stack_frame_flags.
13364 (ix86_expand_epilogue): Likewise.
13365 (ix86_expand_split_stack_prologue): Likewise.
13366 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13368 2017-08-10 Martin Liska <mliska@suse.cz>
13371 * c-attribs.c (handle_target_attribute):
13372 Report warning for an empty string argument of target attribute.
13374 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13377 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13379 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13380 or DECL_NONLOCAL labels.
13381 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13382 or DECL_NONLOCAL labels here.
13384 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13386 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13387 to indicate when early gimple folding has been disabled.
13388 (rs6000_gimple_fold_builtin): Add debug content.
13389 (rs6000_invalid_builtin): Fix whitespace.
13390 (rs6000_expand_builtin): Fix whitespace.
13391 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13393 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13396 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13397 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13399 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13401 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13402 (qdf24xx): Likewise.
13403 * config/aarch64/aarch64.md: Include falkor.md.
13404 * config/aarch64/falkor.md: New.
13406 2017-08-09 Marek Polacek <polacek@redhat.com>
13409 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13410 * diagnostic.c (emit_diagnostic): Add a comment.
13411 (emit_diagnostic_valist): New function.
13413 2017-08-09 Marek Polacek <polacek@redhat.com>
13416 * input.c (make_location): New overload.
13417 * input.h (make_location): Declare.
13419 2017-08-08 Alan Modra <amodra@gmail.com>
13420 H.J. Lu <hongjiu.lu@intel.com>
13423 * gcc.c (NO_PIE_SPEC): Delete.
13424 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13426 (LINK_PIE_SPEC): ..to here.
13427 (LINK_COMMAND_SPEC): Support -no-pie.
13428 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13429 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13431 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13432 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13433 (ENDFILE_CRTEND_SPEC): Similarly.
13435 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13438 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13439 (mstack-protector-guard-offset=): Ditto.
13440 * config/i386/i386.c (ix86_option_override): Handle
13441 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13443 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13444 ix86_stack_protect_guard_offset variables.
13445 (TARGET_STACK_PROTECT_GUARD): Always define.
13446 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13447 and -mstack-protector-guard-offset= options.
13449 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13451 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13452 boundary case for the last candidate.
13454 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13456 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13457 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13459 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13461 PR middle-end/19706
13462 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13463 * config/aarch64/aarch64-builtins.c
13464 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13465 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13466 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13468 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13469 Andrew Pinski <pinskia@gmail.com>
13471 PR middle-end/19706
13472 * internal-fn.def (XORSIGN): New.
13473 * optabs.def (xorsign_optab): New.
13474 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13475 (convert_expand_mult_copysign): New.
13476 (pass_optimize_widening_mul::execute): Call
13477 convert_expand_mult_copysign.
13479 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13481 PR tree-optimization/81354
13482 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13483 Insert on edges rather than explicitly creating landing pads.
13484 (analyze_candidates_and_replace): Commit edge inserts.
13486 2017-08-08 Richard Biener <rguenther@suse.de>
13488 PR middle-end/81719
13489 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13490 (expand_simple_operations): Also look through ADDR_EXPRs with
13491 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13493 2017-08-08 Richard Biener <rguenther@suse.de>
13495 PR tree-optimization/81723
13496 * tree-vect-slp.c (struct bst_traits): New hash traits.
13497 (bst_fail): New global.
13498 (vect_build_slp_tree_2): New worker, split out from ...
13499 (vect_build_slp_tree): ... this now wrapping it with using
13500 bst_fail set to cache SLP tree build fails. Properly handle
13502 (vect_analyze_slp_instance): Allocate and free bst_fail.
13504 2017-08-08 Martin Liska <mliska@suse.cz>
13507 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13508 LABEL_DECLs that can be from a different function.
13510 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13512 PR tree-optimization/81744
13513 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13514 loop's number of iterations.
13516 2017-08-08 Martin Liska <mliska@suse.cz>
13518 * asan.c: Include header files.
13519 * attribs.c (build_decl_attribute_variant): New function moved
13521 (build_type_attribute_qual_variant): Likewise.
13522 (cmp_attrib_identifiers): Likewise.
13523 (simple_cst_list_equal): Likewise.
13524 (omp_declare_simd_clauses_equal): Likewise.
13525 (attribute_value_equal): Likewise.
13526 (comp_type_attributes): Likewise.
13527 (build_type_attribute_variant): Likewise.
13528 (lookup_ident_attribute): Likewise.
13529 (remove_attribute): Likewise.
13530 (merge_attributes): Likewise.
13531 (merge_type_attributes): Likewise.
13532 (merge_decl_attributes): Likewise.
13533 (merge_dllimport_decl_attributes): Likewise.
13534 (handle_dll_attribute): Likewise.
13535 (attribute_list_equal): Likewise.
13536 (attribute_list_contained): Likewise.
13537 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13538 (lookup_attribute_by_prefix): Likewise.
13539 * bb-reorder.c: Include header files.
13540 * builtins.c: Likewise.
13541 * calls.c: Likewise.
13542 * cfgexpand.c: Likewise.
13543 * cgraph.c: Likewise.
13544 * cgraphunit.c: Likewise.
13545 * convert.c: Likewise.
13546 * dwarf2out.c: Likewise.
13547 * final.c: Likewise.
13548 * fold-const.c: Likewise.
13549 * function.c: Likewise.
13550 * gimple-expr.c: Likewise.
13551 * gimple-fold.c: Likewise.
13552 * gimple-pretty-print.c: Likewise.
13553 * gimple.c: Likewise.
13554 * gimplify.c: Likewise.
13555 * hsa-common.c: Likewise.
13556 * hsa-gen.c: Likewise.
13557 * internal-fn.c: Likewise.
13558 * ipa-chkp.c: Likewise.
13559 * ipa-cp.c: Likewise.
13560 * ipa-devirt.c: Likewise.
13561 * ipa-fnsummary.c: Likewise.
13562 * ipa-inline.c: Likewise.
13563 * ipa-visibility.c: Likewise.
13565 * lto-cgraph.c: Likewise.
13566 * omp-expand.c: Likewise.
13567 * omp-general.c: Likewise.
13568 * omp-low.c: Likewise.
13569 * omp-offload.c: Likewise.
13570 * omp-simd-clone.c: Likewise.
13571 * opts-global.c: Likewise.
13572 * passes.c: Likewise.
13573 * predict.c: Likewise.
13574 * sancov.c: Likewise.
13575 * sanopt.c: Likewise.
13576 * symtab.c: Likewise.
13577 * toplev.c: Likewise.
13578 * trans-mem.c: Likewise.
13579 * tree-chkp.c: Likewise.
13580 * tree-eh.c: Likewise.
13581 * tree-into-ssa.c: Likewise.
13582 * tree-object-size.c: Likewise.
13583 * tree-parloops.c: Likewise.
13584 * tree-profile.c: Likewise.
13585 * tree-ssa-ccp.c: Likewise.
13586 * tree-ssa-live.c: Likewise.
13587 * tree-ssa-loop.c: Likewise.
13588 * tree-ssa-sccvn.c: Likewise.
13589 * tree-ssa-structalias.c: Likewise.
13590 * tree-ssa.c: Likewise.
13591 * tree-streamer-in.c: Likewise.
13592 * tree-vectorizer.c: Likewise.
13593 * tree-vrp.c: Likewise.
13594 * tsan.c: Likewise.
13595 * ubsan.c: Likewise.
13596 * varasm.c: Likewise.
13597 * varpool.c: Likewise.
13598 * tree.c: Remove functions moved to attribs.[ch].
13599 * tree.h: Likewise.
13600 * config/aarch64/aarch64.c: Add attrs.h header file.
13601 * config/alpha/alpha.c: Likewise.
13602 * config/arc/arc.c: Likewise.
13603 * config/arm/arm.c: Likewise.
13604 * config/avr/avr.c: Likewise.
13605 * config/bfin/bfin.c: Likewise.
13606 * config/c6x/c6x.c: Likewise.
13607 * config/cr16/cr16.c: Likewise.
13608 * config/cris/cris.c: Likewise.
13609 * config/darwin.c: Likewise.
13610 * config/epiphany/epiphany.c: Likewise.
13611 * config/fr30/fr30.c: Likewise.
13612 * config/frv/frv.c: Likewise.
13613 * config/ft32/ft32.c: Likewise.
13614 * config/h8300/h8300.c: Likewise.
13615 * config/i386/winnt.c: Likewise.
13616 * config/ia64/ia64.c: Likewise.
13617 * config/iq2000/iq2000.c: Likewise.
13618 * config/lm32/lm32.c: Likewise.
13619 * config/m32c/m32c.c: Likewise.
13620 * config/m32r/m32r.c: Likewise.
13621 * config/m68k/m68k.c: Likewise.
13622 * config/mcore/mcore.c: Likewise.
13623 * config/microblaze/microblaze.c: Likewise.
13624 * config/mips/mips.c: Likewise.
13625 * config/mmix/mmix.c: Likewise.
13626 * config/mn10300/mn10300.c: Likewise.
13627 * config/moxie/moxie.c: Likewise.
13628 * config/msp430/msp430.c: Likewise.
13629 * config/nds32/nds32-isr.c: Likewise.
13630 * config/nds32/nds32.c: Likewise.
13631 * config/nios2/nios2.c: Likewise.
13632 * config/nvptx/nvptx.c: Likewise.
13633 * config/pa/pa.c: Likewise.
13634 * config/pdp11/pdp11.c: Likewise.
13635 * config/powerpcspe/powerpcspe.c: Likewise.
13636 * config/riscv/riscv.c: Likewise.
13637 * config/rl78/rl78.c: Likewise.
13638 * config/rx/rx.c: Likewise.
13639 * config/s390/s390.c: Likewise.
13640 * config/sh/sh.c: Likewise.
13641 * config/sol2.c: Likewise.
13642 * config/sparc/sparc.c: Likewise.
13643 * config/spu/spu.c: Likewise.
13644 * config/stormy16/stormy16.c: Likewise.
13645 * config/tilegx/tilegx.c: Likewise.
13646 * config/tilepro/tilepro.c: Likewise.
13647 * config/v850/v850.c: Likewise.
13648 * config/vax/vax.c: Likewise.
13649 * config/visium/visium.c: Likewise.
13650 * config/xtensa/xtensa.c: Likewise.
13652 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13655 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13656 constraints since the -mupper-regs-* switches have been
13658 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13659 into a vector from a double word element that was extracted from
13660 another vector, and eliminate extra XXPERMDI instructions.
13661 (vsx_concat_<mode>_2): Likewise.
13662 (vsx_concat_<mode>_3): Likewise.
13663 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13664 concat to allow optimizing inserts from previous extracts.
13666 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13668 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13669 memory reference to a SSP offset in TLS address space.
13670 (ix86_print_operand) <case '@'>: Remove.
13671 (ix86_print_operand_punct_valid_p): Remove '@' code.
13672 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
13673 UNSPEC_SP_TLS_TEST.
13674 (stack_tls_protect_set_<mode>): Remove.
13675 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
13676 (stack_tls_protect_test_<mode>): Remove.
13677 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
13679 2017-08-07 Olivier Hainque <hainque@adacore.com>
13682 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
13684 2017-08-07 Douglas Rupp <rupp@adacore.com>
13686 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
13687 variable was referenced as multidir in command.
13689 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13692 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13695 2017-08-07 Martin Liska <mliska@suse.cz>
13697 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13698 * config/rl78/rl78.c: Add include of attribs.h.
13699 * config/sh/sh.c: Likewise.
13700 * config/v850/v850.c: Likewise.
13702 2017-08-07 Tom de Vries <tom@codesourcery.com>
13704 PR middle-end/78266
13705 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13707 2017-08-07 Martin Liska <mliska@suse.cz>
13709 * config/mips/mips.c: Include attribs.h.
13711 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13714 * doc/invoke.texi: Document change in behvaior for -Ofast for
13717 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13719 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13721 (aarch64_pop_regs): Likewise.
13722 (aarch64_gen_load_pair): Likewise.
13723 (aarch64_save_callee_saves): Likewise.
13724 (aarch64_restore_callee_saves): Likewise.
13726 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13728 * config/i386/i386.c: Revert the last change.
13730 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13733 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13735 (ix86_finalize_stack_frame_flags): This. Also clear
13736 frame_pointer_needed if -fno-omit-frame-pointer is used without
13738 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13739 with ix86_finalize_stack_frame_flags.
13740 (ix86_expand_epilogue): Likewise.
13741 (ix86_expand_split_stack_prologue): Likewise.
13743 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13746 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13747 to P_AES for Westmere.
13749 2017-08-07 Jonathan Yong <10walls@gmail.com>
13751 * config/i386/mingw.opt (fset-stack-executable): Removed.
13752 * config/i386/cygming.opt (fset-stack-executable): Moved
13754 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13756 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13758 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13760 2017-08-07 Marek Polacek <polacek@redhat.com>
13762 PR middle-end/81737
13763 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13765 2017-08-07 Martin Liska <mliska@suse.cz>
13767 * attribs.h (canonicalize_attr_name): New function.
13768 (cmp_attribs): Move from c-format.c and adjusted.
13769 (is_attribute_p): Moved from tree.h.
13770 * tree-inline.c: Add new includes.
13771 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13772 (private_is_attribute_p): Remove.
13773 (private_lookup_attribute): Likewise.
13774 (private_lookup_attribute_by_prefix): Simplify.
13775 (remove_attribute): Use is_attribute_p.
13776 * tree.h: Remove removed declarations.
13778 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13780 PR middle-end/81698
13781 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13782 instead of computing it in the function. Formatting fix.
13783 (expand_case): Don't rely on default_edge being the first edge,
13784 clear it if removing it, pass default_edge to
13785 emit_case_dispatch_table.
13786 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13789 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13791 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13792 insn in the function, emit NOP after the insn.
13794 2017-08-06 Tom de Vries <tom@codesourcery.com>
13796 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13799 2017-08-06 Tom de Vries <tom@codesourcery.com>
13801 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13804 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13806 PR tree-optimization/57371
13807 * match.pd: New pattern.
13809 2017-08-04 Marek Polacek <polacek@redhat.com>
13811 PR middle-end/81695
13812 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13813 perform the computation in offset_int.
13815 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13817 PR tree-optimization/81136
13818 * tree-vectorizer.h: Include tree-hash-traits.h.
13819 (vec_base_alignments): New typedef.
13820 (vec_info): Add a base_alignments field.
13821 (vect_record_base_alignments): Declare.
13822 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13824 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13825 (create_data_ref): Add an is_conditional_in_stmt argument.
13826 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13827 the is_conditional_in_stmt field.
13828 (data_ref_loc): Add an is_conditional_in_stmt field.
13829 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13830 (find_data_references_in_stmt): Update call to create_data_ref.
13831 (graphite_find_data_references_in_stmt): Likewise.
13832 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13833 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13834 (vect_record_base_alignment): New function.
13835 (vect_record_base_alignments): Likewise.
13836 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13837 for nested statements even if we fail to compute a misalignment.
13838 Use pooled base alignments for unconditional references.
13839 (vect_find_same_alignment_drs): Compare base addresses instead
13841 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13842 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13844 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13846 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13847 Add an explicit name for the enum. Use auto_vec for slp_instances
13848 and grouped_stores.
13849 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13851 (_bb_vec_info): Add a constructor and destructor.
13852 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13853 (destroy_loop_vec_info): Delete.
13854 (vect_destroy_datarefs): Likewise.
13855 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13856 (vec_info::vec_info): New function.
13857 (vec_info::~vec_info): Likewise.
13858 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13859 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13860 destroy_loop_vec_info.
13861 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13862 (_loop_vec_info::_loop_vec_info): ...this.
13863 (destroy_loop_vec_info): Replace with...
13864 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13865 the stmt_vec_infos. Leave handling of vec_info information to its
13866 destructor. Remove explicit vector releases.
13867 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13868 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13869 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13870 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13871 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13872 (destroy_bb_vec_info): Replace with...
13873 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13874 information to its destructor.
13875 (vect_slp_analyze_bb_1): Use new and delete instead of
13876 new_bb_vec_info and destroy_bb_vec_info.
13877 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13880 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13882 * tree-data-ref.h (subscript): Add access_fn field.
13883 (data_dependence_relation): Add could_be_independent_p.
13884 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13885 (same_access_functions): Move to tree-data-ref.c.
13886 * tree-data-ref.c (ref_contains_union_access_p): New function.
13887 (access_fn_component_p): Likewise.
13888 (access_fn_components_comparable_p): Likewise.
13889 (dr_analyze_indices): Add a reference to access_fn_component_p.
13890 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13892 (constant_access_functions): Likewise.
13893 (add_other_self_distances): Likewise.
13894 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13895 (initialize_data_dependence_relation): Use XCNEW and remove
13896 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13897 of access functions that have the same type. Allow the
13898 subsequence to end with different bases in some circumstances.
13899 Record the chosen access functions in SUB_ACCESS_FN.
13900 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13901 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13902 (subscript_dependence_tester_1): Likewise dra and drb.
13903 (build_classic_dist_vector): Update calls accordingly.
13904 (subscript_dependence_tester): Likewise.
13905 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13906 DDR_COULD_BE_INDEPENDENT_P.
13907 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13908 comp_alias_ddrs instead of may_alias_ddrs.
13909 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13911 (vect_analyze_data_ref_dependence): Use it if
13912 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13913 distance vectors if that fails.
13914 (dependence_distance_ge_vf): New function.
13915 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13916 LOOP_VINFO_MAY_ALIAS_DDRS.
13918 2017-08-04 Richard Biener <rguenther@suse.de>
13920 PR middle-end/81705
13921 * fold-const.c (fold_binary_loc): Properly restrict
13922 minus_var0 && minus_var1 case when associating undefined overflow
13925 2017-08-04 Tom de Vries <tom@codesourcery.com>
13927 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13929 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13931 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13932 Don't start diagnostic messages with a capital letter.
13933 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13935 (rs6000_invalid_builtin): Likewise.
13936 (rs6000_trampoline_init): Likewise.
13938 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13941 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13942 after setting changeable df flags.
13944 2017-08-03 Richard Biener <rguenther@suse.de>
13946 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13947 up if the use is in USE - X.
13949 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13951 * toplev.c (dumpfile.h): New include.
13952 (internal_error_reentered): New static function. Use it...
13953 (internal_error_function): ...here to handle reentered internal_error.
13955 2017-08-03 Richard Biener <rguenther@suse.de>
13957 PR middle-end/81148
13958 * fold-const.c (split_tree): Add minus_var and minus_con
13959 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13960 here but always use minus_*.
13961 (associate_trees): Assert we never associate with MINUS_EXPR
13962 and NULL first operand. Do not recurse for PLUS_EXPR operands
13963 when associating as MINUS_EXPR either.
13964 (fold_binary_loc): Track minus_var and minus_con.
13966 2017-08-03 Tom de Vries <tom@codesourcery.com>
13969 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13970 ACCEL_COMPILER, apply finish_options on
13971 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13973 2017-08-03 Tom de Vries <tom@codesourcery.com>
13976 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13977 function_entry_patch_area_size > 0.
13979 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13982 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13983 instead of 10??LU, perform unit multiplication in wide_int,
13984 don't change alloc_object_size_limit if the limit is larger
13987 PR tree-optimization/81655
13988 PR tree-optimization/81588
13989 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
13990 the case when ranges[i].low and high are 1 for unsigned type with
13993 PR middle-end/81052
13994 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
13995 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
13997 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13999 * tree-vrp.h: Add include guard.
14001 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14004 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14005 (ud2): New insn pattern.
14006 * config/i386/i386.c (ix86_expand_epilogue):
14007 For naked functions, generate ud2 instead of trap insn.
14009 2017-08-02 Marek Polacek <polacek@redhat.com>
14012 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14014 2017-08-02 Tom de Vries <tom@codesourcery.com>
14015 Cesar Philippidis <cesar@codesourcery.com>
14017 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14018 Add missing edge probabilities.
14020 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14022 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14023 Correct endianness.
14025 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14027 PR middle-end/79499
14028 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14029 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14030 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14031 prologue_seq sequences - if any.
14033 2017-08-02 Richard Biener <rguenther@suse.de>
14035 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14036 via vectors if supported, integer extracts via punning if supported
14037 or otherwise vector extracts.
14039 2017-08-02 Richard Biener <rguenther@suse.de>
14041 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14043 (bitmap_insert_into_set): ... this.
14045 2017-08-02 Richard Biener <rguenther@suse.de>
14047 PR tree-optimization/81633
14049 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14051 PR tree-optimization/71752
14052 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14054 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14056 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14057 (machine_function::call_ms2sysv_pad_out): Remove field.
14058 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14059 (ix86_compute_frame_layout): Likewise.
14061 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14064 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14065 attribute with interrupt attribute.
14067 2017-08-01 Andrew Pinski <apinski@cavium.com>
14069 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14070 BIT_INSERT_EXPR's operand 1
14071 to see if the types precision matches.
14073 2017-08-01 Martin Liska <mliska@suse.cz>
14075 PR middle-end/70140
14076 * builtins.c (expand_builtin_memcpy_args): Remove.
14077 (expand_builtin_memcpy): Call newly added function
14078 expand_builtin_memory_copy_args.
14079 (expand_builtin_memcpy_with_bounds): Likewise.
14080 (expand_builtin_mempcpy): Remove last argument.
14081 (expand_builtin_mempcpy_with_bounds): Likewise.
14082 (expand_builtin_memory_copy_args): New function created from
14083 expand_builtin_mempcpy_args with small modifications.
14084 (expand_builtin_mempcpy_args): Remove.
14085 (expand_builtin_stpcpy): Remove unused argument.
14086 (expand_builtin): Likewise.
14087 (expand_builtin_with_bounds): Likewise.
14089 2017-08-01 Martin Liska <mliska@suse.cz>
14092 Make mempcpy more optimal (PR middle-end/70140).
14094 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14097 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14098 __builtin_vec_cmpne verify both arguments are compatible vectors
14099 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14100 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14101 move computation of aligned to after checking the argument types.
14105 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14106 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14109 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14110 Doug Rupp <rupp@adacore.com>
14111 Olivier Hainque <hainque@adacore.com>
14113 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14114 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14116 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14117 for TARGET_OS_CPP_BUILTIN.
14118 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14119 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14121 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14122 passing required abi options to the assembler for EABI configurations.
14123 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14124 of .text.hot and .text.unlikely sections for kernel modules when
14125 using ARM style exceptions.
14126 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14127 options. Add EXTRA_CC1_SPEC.
14128 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14130 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14132 (ARM_TARGET2_DWARF_FORMAT): Define.
14133 * config/arm/t-vxworks: Adjust multilib control to removal of the
14134 Diab command line options.
14136 2017-08-01 Martin Liska <mliska@suse.cz>
14138 PR gcov-profile/81561
14139 * gcov.c (unblock): Make unblocking safe as we need to preserve
14140 index correspondence of blocks and block_lists.
14142 2017-08-01 Richard Biener <rguenther@suse.de>
14144 PR tree-optimization/81181
14145 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14146 (compute_antic): ... end of iteration here.
14148 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14150 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14151 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14152 (ftree-slp-vectorize): Likewise.
14153 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14154 can no longer be set independent of flag_tree_loop_vectorize.
14155 * omp-general.c (emp_max_vf): Likewise.
14156 * opts.c (enable_fdo_optimizations): Remove references to
14157 flag_tree_vectorize, these are now implicit.
14158 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14159 and leave it for the options machinery.
14161 2017-08-01 Martin Liska <mliska@suse.cz>
14163 PR middle-end/70140
14164 * builtins.c (expand_builtin_memcpy_args): Remove.
14165 (expand_builtin_memcpy): Call newly added function
14166 expand_builtin_memory_copy_args.
14167 (expand_builtin_memcpy_with_bounds): Likewise.
14168 (expand_builtin_mempcpy): Remove last argument.
14169 (expand_builtin_mempcpy_with_bounds): Likewise.
14170 (expand_builtin_memory_copy_args): New function created from
14171 expand_builtin_mempcpy_args with small modifications.
14172 (expand_builtin_mempcpy_args): Remove.
14173 (expand_builtin_stpcpy): Remove unused argument.
14174 (expand_builtin): Likewise.
14175 (expand_builtin_with_bounds): Likewise.
14177 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14180 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
14181 print "ds:" only for immediates in generic address space.
14183 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14186 * config/i386/i386.c (ix86_funciton_naked): New prototype.
14187 (ix86_function_ok_for_sibcall): Return false for naked functions.
14189 2017-08-01 Richard Biener <rguenther@suse.de>
14191 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
14192 (compute_antic): Seed worklist with exit block predecessors.
14193 * cfganal.c (dfs_find_deadend): For a cycle return the source
14194 of the edge closing it.
14196 2017-08-01 Tamar Christina <tamar.christina@arm.com>
14198 * config/aarch64/aarch64.c
14199 (aarch64_can_const_movi_rtx_p): Move 0 check.
14201 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14203 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
14204 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
14206 * match.pd: Ditto in address comparison pattern.
14208 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14210 PR tree-optimization/81627
14211 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
14212 closed ssa form for store-store chain.
14214 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14216 PR tree-optimization/81620
14217 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
14218 for store-store chain.
14220 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14222 PR tree-optimization/81588
14223 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
14224 ranges[i].in_p, invert comparison code ccode. For >/>=,
14225 swap rhs1 and rhs2 and comparison code unconditionally,
14226 for </<= don't do that. Don't swap rhs1/rhs2 again if
14227 ranges[i].in_p, instead invert comparison code ccode if
14228 opcode or oe->rank is BIT_IOR_EXPR.
14231 * optabs.def (vec_extract_optab, vec_init_optab): Change from
14232 a direct optab to conversion optab.
14233 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
14234 with GET_MODE_INNER as last argument instead of optab_handler.
14235 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14236 vector extraction if possible and optab is available.
14237 * expr.c (store_constructor): Use convert_optab_handler instead
14238 of optab_handler. Use vector initialization from smaller
14239 vectors if possible and optab is available.
14240 * tree-vect-stmts.c (vectorizable_load): Likewise.
14241 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14242 now have two modes.
14243 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14244 of vec_init from half-sized vectors with the same element mode.
14245 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14246 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14247 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14248 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14249 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14250 after mode in gen_vec_extract* calls.
14251 (vec_extract<mode>): Renamed to ...
14252 (vec_extract<mode><ssescalarmodelower>): ... this.
14253 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14254 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14255 element mode after mode in gen_vec_init* calls.
14256 (VEC_INIT_HALF_MODE): New mode iterator.
14257 (vec_init<mode>): Renamed to ...
14258 (vec_init<mode><ssescalarmodelower>): ... this.
14259 (vec_init<mode><ssehalfvecmodelower>): New expander.
14260 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14261 (vec_extractv2sfsf): ... this.
14262 (vec_initv2sf): Renamed to ...
14263 (vec_initv2sfsf): ... this.
14264 (vec_extractv2si): Renamed to ...
14265 (vec_extractv2sisi): ... this.
14266 (vec_initv2si): Renamed to ...
14267 (vec_initv2sisi): ... this.
14268 (vec_extractv4hi): Renamed to ...
14269 (vec_extractv4hihi): ... this.
14270 (vec_initv4hi): Renamed to ...
14271 (vec_initv4hihi): ... this.
14272 (vec_extractv8qi): Renamed to ...
14273 (vec_extractv8qiqi): ... this.
14274 (vec_initv8qi): Renamed to ...
14275 (vec_initv8qiqi): ... this.
14276 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14277 (vec_init<mode>): Renamed to ...
14278 (vec_init<mode><VEC_base_l>): ... this.
14279 (vec_extract<mode>): Renamed to ...
14280 (vec_extract<mode><VEC_base_l>): ... this.
14281 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14282 (vec_initv2sfsf): ... this.
14283 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14284 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14285 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14286 element mode after mode in gen_vec_init* calls.
14287 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14288 (vec_init<mode><Vel>): ... this.
14289 (vec_extract<mode>): Renamed to ...
14290 (vec_extract<mode><Vel>): ... this.
14291 * config/aarch64/iterators.md (Vel): New mode attribute.
14292 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14293 Add element mode after mode in gen_vec_extract* calls.
14294 * config/s390/vector.md (non_vec_l): New mode attribute.
14295 (vec_extract<mode>): Renamed to ...
14296 (vec_extract<mode><non_vec_l>): ... this.
14297 (vec_init<mode>): Renamed to ...
14298 (vec_init<mode><non_vec_l>): ... this.
14299 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14300 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14302 * config/arm/iterators.md (V_elem_l): New mode attribute.
14303 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14304 (vec_extract<mode><V_elem_l>): ... this.
14305 (vec_extractv2di): Renamed to ...
14306 (vec_extractv2didi): ... this.
14307 (vec_init<mode>): Renamed to ...
14308 (vec_init<mode><V_elem_l>): ... this.
14309 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14310 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14311 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14312 Add element mode after gen_vec_extract* calls.
14313 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14314 (vec_init<mode><unitmode>): ... this.
14315 (vec_extract<mode>): Renamed to ...
14316 (vec_extract<mode><unitmode>): ... this.
14317 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14318 (vec_init<mode><unitmode>): ... this.
14319 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14320 (vec_initv2sfsf): ... this.
14321 (vec_extractv2sf): Renamed to ...
14322 (vec_extractv2sfsf): ... this.
14323 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14324 Add element mode after gen_vec_extract* calls.
14325 * config/mips/mips.md (unitmode): New mode iterator.
14326 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14327 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14328 * config/spu/spu.md (inner_l): New mode attribute.
14329 (vec_init<mode>): Renamed to ...
14330 (vec_init<mode><inner_l>): ... this.
14331 (vec_extract<mode>): Renamed to ...
14332 (vec_extract<mode><inner_l>): ... this.
14333 * config/sparc/sparc.md (veltmode): New mode iterator.
14334 (vec_init<VMALL:mode>): Renamed to ...
14335 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14336 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14337 (vec_initv2sisi): ... this.
14338 (vec_initv2sf): Renamed to ...
14339 (vec_initv2sfsf): ... this.
14340 (vec_extractv2sf): Renamed to ...
14341 (vec_extractv2sfsf): ... this.
14342 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14343 (vec_init<mode>): Renamed to ...
14344 (vec_init<mode><VEC_base_l>): ... this.
14345 (vec_extract<mode>): Renamed to ...
14346 (vec_extract<mode><VEC_base_l>): ... this.
14347 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14348 (vec_initv2sfsf): ... this.
14349 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14350 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14351 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14352 gen_vec_init* calls.
14354 2017-08-01 Richard Biener <rguenther@suse.de>
14356 PR tree-optimization/81297
14357 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14358 TREE_OVERFLOW from INTEGER_CSTs.
14360 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14362 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14364 2017-07-31 Carl Love <cel@us.ibm.com>
14366 * config/rs6000/rs6000-c: Add support for built-in functions
14367 vector signed char vec_xl_be (signed long long, signed char *);
14368 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14369 vector signed int vec_xl_be (signed long long, signed int *);
14370 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14371 vector signed long long vec_xl_be (signed long long, signed long long *);
14372 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14373 vector signed short vec_xl_be (signed long long, signed short *);
14374 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14375 vector double vec_xl_be (signed long long, double *);
14376 vector float vec_xl_be (signed long long, float *);
14377 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14378 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14379 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14381 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14382 (altivec_expand_builtin): Add switch statement to call
14383 altivec_expand_xl_be for each builtin.
14384 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14385 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14386 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14387 __builtin_vsx_le_be_v16qi.
14388 * doc/extend.texi: Update the built-in documentation file for the
14389 new built-in functions.
14391 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14394 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14396 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14398 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14400 * config.gcc: Add z14.
14401 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14402 CPU model numbers for z13s and z14.
14403 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14405 * config/s390/s390-opts.h (enum processor_type): Rename
14406 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14407 * config/s390/s390.c (processor_table): Add field for CPU name to
14408 be passed to Binutils.
14409 (s390_asm_output_machine_for_arch): Use the new field in
14410 processor_table for Binutils.
14411 (s390_expand_builtin): Replace arch12 with z14.
14412 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14413 (s390_get_sched_attrmask): Likewise.
14414 (s390_get_unit_mask): Likewise.
14415 * config/s390/s390.opt: Add z14 to processor_type enum.
14417 2017-07-31 Martin Jambor <mjambor@suse.cz>
14420 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14421 regardless of optimization level.
14423 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14424 Martin Liska <mliska@suse.cz>
14426 * predict.def: Remove old comment and adjust probability.
14427 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14428 PREDICT statements.
14430 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14433 * config/i386/i386.c (ix86_function_naked): New function.
14434 (ix86_can_use_return_insn_p): Return false for naked functions.
14435 (ix86_expand_prologue): Skip prologue for naked functions.
14436 (ix86_expand_epilogue): Skip epilogue for naked functions
14437 and emit trap instruction.
14438 (ix86_warn_func_return): New function.
14439 (ix86_attribute_table): Add "naked" attribute specification.
14440 (TARGET_WARN_FUNC_RETURN): Define.
14441 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14443 2017-07-31 Martin Liska <mliska@suse.cz>
14445 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14446 (dump_gimple_bb_header): Always dump BB info.
14447 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14449 2017-07-31 Martin Liska <mliska@suse.cz>
14452 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14453 also with current_function_decl non-null equality.
14455 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14458 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14459 change type to the element type, instead add eltype variable and
14460 use it where we are interested in the element type.
14462 PR tree-optimization/81603
14463 * ipa-polymorphic-call.c
14464 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14465 offset arithmetic in offset_int, bail out if the resulting bit offset
14466 doesn't fit into shwi.
14468 2017-07-31 Martin Liska <mliska@suse.cz>
14470 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14471 (gimplify_save_expr): Fix comment.
14473 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14476 * config/i386/i386.c (ix86_function_arg): Update arguments for
14478 (ix86_compute_frame_layout): Set the initial stack offset to
14479 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14480 INCOMING_FRAME_SP_OFFSET.
14481 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14482 stack before exception handler returns.
14483 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14484 the 'ERROR_CODE' for exception handler.
14486 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14488 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14489 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14490 (ASM_OUTPUT_REG_POP): Ditto.
14491 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14492 instead of asm_fprintf to output pure string.
14494 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14496 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14497 to imported_module_or_decl hook.
14498 (debug_nothing_tree_tree_tree_bool): Remove.
14499 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14500 * debug.c (do_nothing_debug_hooks): Use
14501 debug_nothing_tree_tree_tree_bool_bool instead of
14502 debug_nothing_tree_tree_tree_bool.
14503 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14504 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14505 * sdbout.c (sdb_debug_hooks): Likewise.
14506 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14507 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14509 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14510 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14511 attribute, don't add anything.
14513 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14515 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14516 (fold_build2_stat_loc): Likewise.
14517 (fold_build3_stat_loc): Likewise.
14518 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14519 (fold_build1_loc): Remove macro.
14520 (fold_build2_loc): Likewise.
14521 (fold_build3_loc): Likewise.
14523 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14525 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14526 (gimple_build_debug_bind_source_stat): Likewise.
14527 * gimple.h (gimple_build_debug_bind): Remove macro.
14528 (gimple_build_debug_bind_source): Likewise.
14530 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14532 * bitmap.c (bitmap_alloc): Adjust.
14533 (bitmap_gc_alloc): Likewise.
14534 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14536 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14538 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14539 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14540 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14541 (bitmap_gc_alloc_stat): Likewise.
14542 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14544 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14546 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14547 * rtl.h (shallow_copy_rtx): Remove macro.
14549 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14551 * emit-rtl.c (gen_raw_REG): Adjust.
14552 * gengenrtl.c (gendef): Likewise.
14553 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14554 * rtl.h (rtx_alloc): Remove macro.
14556 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14558 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14559 (build_tree_list_stat): Likewise.
14560 * tree.h (build_tree_list): Remove macro.
14561 (build_tree_list_vec): Likewise.
14563 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14565 * tree.c (make_vector_stat): Remove _stat from name.
14566 (build_vector_stat): Likewise.
14567 * tree.h (make_vector_stat): Remove macro.
14568 (build_vector_stat): Likewise.
14570 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14572 * tree.h (build_var_debug_value): Remove prototype.
14574 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14576 * tree.c (tree_cons_stat): Remove _stat from name.
14577 * tree.h (tree_cons): Remove macro.
14579 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14581 * tree.c (build_vl_exp_stat): Remove _stat from name.
14582 * tree.h (build_vl_exp): Remove macro.
14584 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14586 * tree.c (build_decl_stat): Remove _stat from name.
14587 * tree.h (build_decl): Remove macro.
14589 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14591 * gimple.c (gimple_build_with_ops_stat): Adjust.
14592 (gimple_alloc_stat): Remove _stat from name.
14593 * gimple.h (gimple_alloc): Remove macro.
14595 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14597 * tree.c (make_tree_vec_stat): Remove _stat from name.
14598 (grow_tree_vec_stat): Likewise.
14599 * tree.h (make_tree_vec_stat): Adjust prototype.
14600 (grow_tree_vec_stat): Likewise.
14601 (make_tree_vec): Remove macro.
14602 (grow_tree_vec): Likewise.
14604 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14606 * fold-const.c (fold_build1_stat_loc): Adjust.
14607 (fold_build2_stat_loc): Likewise.
14608 (fold_build3_stat_loc): Likewise.
14609 * tree.c (build0_stat): Remove _stat from name.
14610 (build1_stat): Likewise.
14611 (build2_stat): Likewise.
14612 (build3_stat): Likewise.
14613 (build4_stat): Likewise.
14614 (build5_stat): Likewise.
14615 * tree.h (build1_loc): Remove macro, and rename _stat function
14617 (build2_loc): Likewise.
14618 (build3_loc): Likewise.
14619 (build4_loc): Likewise.
14620 (build5_loc): Likewise.
14622 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14624 * tree.c (make_int_cst_stat): Remove _stat from name.
14625 * tree.h (make_int_cst_stat): Adjust prototype.
14626 (make_int_cst): Remove macro.
14628 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14630 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14631 * tree.h (make_tree_binfo_stat): Adjust prototype.
14632 (make_tree_binfo): Remove.
14634 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14636 * tree.c (copy_node_stat): Rename to copy_node.
14637 (build_distinct_type_copy): Adjust.
14638 * tree.h (copy_node_stat): Adjust prototype.
14639 (copy_node): Remove macro.
14641 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14643 * tree.c (make_node_stat): rename to make_node.
14644 (build_tree_list_stat): Adjust.
14645 (build0_stat): Likewise.
14646 (build2_stat): Likewise.
14647 (build3_stat): Likewise.
14648 (build4_stat): Likewise.
14649 (build5_stat): Likewise.
14650 (build_decl_stat): Likewise.
14651 * tree.h (make_node_stat): Adjust prototype.
14652 (make_node): remove macro.
14654 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14656 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14657 (PPC_FEATURE2_SCV): Likewise.
14658 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14660 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14662 * config/aarch64/aarch64.c
14663 (aarch64_internal_mov_immediate): Add new special pattern.
14664 * config/aarch64/aarch64.md (*movdi_aarch64):
14665 Add reg/32bit const mov case.
14667 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14668 Richard Sandiford <richard.sandiford@linaro.org>
14670 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14671 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14672 Add integer and movi cases.
14673 (movi-split-hf-df-sf split, fp16): New.
14674 (enabled): Added TARGET_FP_F16INST.
14675 * config/aarch64/iterators.md (GPF_HF): New.
14676 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
14678 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14680 * config/aarch64/aarch64.c
14681 (aarch64_simd_container_mode): Add prototype.
14682 (aarch64_expand_mov_immediate): Add HI support.
14683 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
14684 (aarch64_can_const_movi_rtx_p): New.
14685 (aarch64_preferred_reload_class):
14686 Remove restrictions of using FP registers for certain SIMD operations.
14687 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
14688 (aarch64_valid_floating_const): Add integer move validation.
14689 (aarch64_simd_imm_scalar_p): Remove.
14690 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14691 (aarch64_legitimate_constant_p): Expand list of supported cases.
14692 * config/aarch64/aarch64-protos.h
14693 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14694 (aarch64_reinterpret_float_as_int): New.
14695 (aarch64_simd_imm_scalar_p): Remove.
14696 * config/aarch64/constraints.md (Uvi): New.
14697 (Dd): Split into Ds and new Dd.
14698 * config/aarch64/aarch64.md (*movsi_aarch64):
14700 (*movdi_aarch64): Add SIMD mov case.
14702 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14704 * tree-predcom.c: (struct chain): Handle store-store chain in which
14705 stores for elimination only store loop invariant values.
14706 (execute_pred_commoning_chain): Ditto.
14707 (prepare_initializers_chain_store_elim): Ditto.
14708 (prepare_finalizers): Ditto.
14709 (is_inv_store_elimination_chain): New function.
14710 (initialize_root_vars_store_elim_1): New function.
14712 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14714 * tree-predcom.c: Revise general description of the pass.
14715 (enum chain_type): New enum type for store elimination.
14716 (struct chain): New field supporting store elimination.
14717 (struct component): Ditto.
14718 (dump_chain): Dump store-stores chain.
14719 (release_chain): Release resources.
14720 (split_data_refs_to_components): Compute and create component
14721 contains only stores for elimination.
14722 (get_chain_last_ref_at): New function.
14723 (make_invariant_chain): Initialization.
14724 (make_rooted_chain): Specify chain type in parameter and record it.
14725 (add_looparound_copies): Skip for store-stores chain.
14726 (determine_roots_comp): Compute type of chain and pass it to
14728 (initialize_root_vars_store_elim_2): New function.
14729 (finalize_eliminated_stores): New function.
14730 (remove_stmt): Handle store for elimination.
14731 (execute_pred_commoning_chain): Execute predictive commoning on
14732 store-store chains.
14733 (determine_unroll_factor): Skip unroll for store-stores chain.
14734 (prepare_initializers_chain_store_elim): New function.
14735 (prepare_initializers_chain): Hanlde store-store chain.
14736 (prepare_finalizers_chain, prepare_finalizers): New function.
14737 (tree_predictive_commoning_loop): Return integer value indicating
14738 if loop is unrolled or lcssa form is corrupted.
14739 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14741 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14743 * tree-predcom.c (initialize_root): Delete.
14744 (execute_pred_commoning_chain): Initialize root vars and replace
14745 reference of non-combined chain directly, rather than call above
14748 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14750 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14751 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14753 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14755 * tree-predcom.c (struct chain): New field init_seq.
14756 (release_chain): Release init_seq.
14757 (prepare_initializers_chain): Record intialization stmts in above
14759 (insert_init_seqs): New function.
14760 (tree_predictive_commoning_loop): Call insert_init_seqs.
14762 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14764 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14766 2017-07-28 Richard Biener <rguenther@suse.de>
14768 * match.pd: Remove superfluous :c.
14769 * genmatch.c (simplify::id): Add member.
14770 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14772 (current_id): New global.
14773 (dt_node::parent): Move from ...
14774 (dt_operand::parent): ... here. Add for_id member.
14775 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14776 (decision_tree::find_node): Relax order requirement when
14777 merging DT_TRUE nodes to ones inbetween the current simplify
14778 and the one we try to merge with. Add diagnostic whenever
14779 we need to enforce pattern order by not merging.
14780 (decision_tree::insert): Set current_id.
14781 (decision_tree::print_node): Dump parent node and for_id.
14782 (parser::last_id): Add member.
14783 (parser::push_simplify): Assign unique id.
14784 (parser::parser): Initialize last_id.
14786 2017-07-28 Martin Liska <mliska@suse.cz>
14789 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14790 gimple_build_debug_bind.
14792 2017-07-28 Richard Biener <rguenther@suse.de>
14794 PR tree-optimization/81502
14795 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14797 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14799 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14800 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14801 for BIT_FIELD_REF args.
14802 * fold-const.c (make_bit_field_ref): Likewise.
14803 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14805 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14808 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14809 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14810 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14811 Or it into SANITIZER_UNDEFINED.
14812 * ubsan.c: Include gimple-fold.h and varasm.h.
14813 (ubsan_expand_ptr_ifn): New function.
14814 (instrument_pointer_overflow): New function.
14815 (maybe_instrument_pointer_overflow): New function.
14816 (instrument_object_size): Formatting fix.
14817 (pass_ubsan::execute): Call instrument_pointer_overflow
14818 and maybe_instrument_pointer_overflow.
14819 * internal-fn.c (expand_UBSAN_PTR): New function.
14820 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14821 * sanitizer.def (__ubsan_handle_pointer_overflow,
14822 __ubsan_handle_pointer_overflow_abort): New builtins.
14823 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14824 * internal-fn.def (UBSAN_PTR): New internal function.
14825 * opts.c (sanitizer_opts): Add pointer-overflow.
14826 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14827 * fold-const.c (build_range_check): Compute pointer range check in
14828 integral type if pointer arithmetics would be needed. Formatting
14831 2017-07-28 Martin Liska <mliska@suse.cz>
14834 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14835 parameters that are of a variable-length.
14837 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14839 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14841 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14842 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14843 (CRT_CALL_STATIC_FUNCTION): Likewise.
14844 (ASM_DEFAULT_SPEC): New define.
14845 (ASM_SPEC32): Likewise.
14846 (ASM_SPEC64): Likewise.
14847 (ASM_SPEC_COMMON): Likewise.
14848 (ASM_SPEC): Likewise.
14849 (INVALID_64BIT): Likewise.
14850 (LINK_OS_DEFAULT_SPEC): Likewise.
14851 (LINK_OS_SPEC32): Likewise.
14852 (LINK_OS_SPEC64): Likewise.
14853 (POWERPC_LINUX): Likewise.
14854 (PTRDIFF_TYPE): Likewise.
14855 (RESTORE_FP_PREFIX): Likewise.
14856 (RESTORE_FP_SUFFIX): Likewise.
14857 (SAVE_FP_PREFIX): Likewise.
14858 (SAVE_FP_SUFFIX): Likewise.
14859 (SIZE_TYPE): Likewise.
14860 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14861 (TARGET_64BIT): Likewise.
14862 (TARGET_64BIT): Likewise.
14863 (TARGET_AIX): Likewise.
14864 (WCHAR_TYPE_SIZE): Likewise.
14865 (WCHAR_TYPE): Undefine.
14866 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14867 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14868 (CPP_OS_RTEMS_SPEC): Delete.
14869 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14870 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14872 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14874 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14876 PR tree-optimization/81578
14877 * tree-parloops.c (build_new_reduction): Bail out if
14878 reduction_code isn't one of the standard OpenMP reductions.
14879 Move the details printing after that decision.
14881 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14883 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14884 related to reload_in_progress.
14885 (splat_input_operand): Likewise.
14886 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14888 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14890 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14891 (TARGET_INSTANTIATE_DECLS): Likewise.
14892 (legitimate_indexed_address_p): Delete reload_in_progress code.
14893 (rs6000_debug_legitimate_address_p): Likewise.
14894 (rs6000_eliminate_indexed_memrefs): Likewise.
14895 (rs6000_emit_le_vsx_store): Likewise.
14896 (rs6000_emit_move_si_sf_subreg): Likewise.
14897 (rs6000_emit_move): Likewise.
14898 (register_to_reg_type): Likewise.
14899 (rs6000_pre_atomic_barrier): Likewise.
14900 (rs6000_machopic_legitimize_pic_address): Likewise.
14901 (rs6000_allocate_stack_temp): Likewise.
14902 (rs6000_address_for_fpconvert): Likewise.
14903 (rs6000_address_for_altivec): Likewise.
14904 (rs6000_secondary_memory_needed_rtx): Delete function.
14905 (rs6000_check_sdmode): Likewise.
14906 (rs6000_alloc_sdmode_stack_slot): Likewise.
14907 (rs6000_instantiate_decls): Likewise.
14908 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14909 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14910 Delete reload_in_progress.
14911 (*vec_reload_and_plus_<mptrsize>): Likewise.
14912 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14913 (vsx_div_v2di): Likewise.
14914 (vsx_udiv_v2di): Likewise.
14916 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14918 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14919 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14920 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14921 (rs6000_debug_reg_global): Delete print of LRA status.
14922 (rs6000_option_override_internal): Delete dead LRA related code.
14923 (rs6000_lra_p): Delete function.
14924 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14926 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14928 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14929 * config/riscv/rtems.h: New file.
14931 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14932 Sudakshina Das <sudi.das@arm.com>
14934 * config/aarch64/aarch64.md
14935 (define_split for and<mode>3nr_compare): Move
14936 non aarch64_logical_operand to a register.
14937 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14938 register immediate operand to a register.
14939 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14941 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14943 PR middle-end/81564
14944 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14946 2017-07-27 Richard Biener <rguenther@suse.de>
14948 PR tree-optimization/81573
14949 PR tree-optimization/81494
14950 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14951 multi defuse cycle case.
14953 2017-07-27 Richard Biener <rguenther@suse.de>
14955 PR tree-optimization/81571
14956 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14959 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14961 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14962 earlier and only if MASK_FPU is set. Adjust formatting.
14964 2017-07-27 Martin Liska <mliska@suse.cz>
14966 * opt-functions.awk: Add validation of value of Init.
14967 * optc-gen.awk: Pass new argument.
14969 2017-07-27 Martin Liska <mliska@suse.cz>
14971 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14972 Fix wrong condition.
14974 2017-07-27 Martin Liska <mliska@suse.cz>
14976 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14977 BBs and edges seen by autoFDO.
14979 2017-07-27 Richard Biener <rguenther@suse.de>
14981 PR tree-optimization/81502
14982 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14983 with incompatible but same sized type.
14984 (execute_update_addresses_taken): Likewise.
14986 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
14988 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
14989 flag_tree_loop_vectorize rather than flag_tree_vectorize.
14991 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14994 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
14995 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
14996 Change s_operand to memory_operand.
14998 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15000 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15001 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15002 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15003 Emit instructions rather than returning an expression. Handle TFmode
15004 and KFmode by casting to TImode.
15005 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15006 (rs6000_emit_le_vsx_store): Likewise.
15007 * config/rs6000/vsx.md (VSX_TI): New iterator.
15008 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15009 (*vsx_le_undo_permute_<mode>): Likewise.
15010 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15011 emit the split sequence.
15012 (*vsx_le_perm_store_<mode>): Likewise.
15014 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15016 PR tree-optimization/81555
15017 PR tree-optimization/81556
15018 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15019 if true, force CHANGED for the recursive invocation.
15020 (reassociate_bb): Remember original length of ops array, pass
15021 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15023 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15024 attributes for noipa attribute. For naked attribute use
15025 lookup_attribute first before lookup_attribute_spec.
15026 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15028 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15029 for functions with noipa attribute.
15030 (cgraph_externally_visible_p): Return true for functions with noipa
15032 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15033 for functions with noipa attribute.
15034 * doc/extend.texi: Document noipa function attribute.
15035 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15036 also for functions with noipa attribute.
15037 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15039 2017-07-26 Andrew Pinski <apinski@cavium.com>
15041 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15042 vec_unalign_load_cost and vec_unalign_store_cost.
15044 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15046 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15047 -mvsx-small-integer option.
15048 (ISA_3_0_MASKS_IEEE): Likewise.
15049 (OTHER_VSX_VECTOR_MASKS): Likewise.
15050 (POWERPC_MASKS): Likewise.
15051 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15052 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15053 code, only testing for DImode being allowed in non-VSX floating
15055 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15056 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15058 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15059 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15060 TARGET_P8_VECTOR test.
15061 (rs6000_secondary_reload_simple_move): Likewise.
15062 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15063 since TARGET_P9_VECTOR was already tested.
15064 (rs6000_opt_masks): Remove -mvsx-small-integer.
15065 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15066 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15068 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15069 test for TARGET_VEXTRACTUB was used, and that uses
15071 (p9 extract splitter): Likewise.
15072 (vsx_extract_<mode>_di_p9): Likewise.
15073 (vsx_extract_<mode>_store_p9): Likewise.
15074 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15075 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15076 the elimination of TARGET_VSX_SMALL_INTEGER.
15077 (vsx_extract_<mode>_p8): Likewise.
15078 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15079 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15080 (vsx_set_<mode>_p9): Likewise.
15081 (vsx_set_v4sf_p9): Likewise.
15082 (vsx_set_v4sf_p9_zero): Likewise.
15083 (vsx_insert_extract_v4sf_p9): Likewise.
15084 (vsx_insert_extract_v4sf_p9_2): Likewise.
15085 * config/rs6000/rs6000.md (sign extend splitter): Change
15086 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15087 (floatsi<mode>2_lfiwax_mem): Likewise.
15088 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15089 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15090 since a test for TARGET_P9_VECTOR was used.
15091 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15092 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15093 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15094 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15095 TARGET_P8_VECTOR test.
15096 (fix_trunc<mode>si2_stfiwx): Likewise.
15097 (fix_trunc<mode>si2_internal): Likewise.
15098 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15099 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15101 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15102 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15103 TARGET_P8_VECTOR test.
15104 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15105 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15106 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15108 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15109 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15110 since a test for TARGET_P9_VECTOR was used.
15111 (splitter for loading small constants): Likewise.
15113 2017-07-26 Andrew Pinski <apinski@cavium.com>
15115 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15118 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15121 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15122 (fp_valid_at): Likewise.
15124 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15126 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15127 (qdf24xx_addrcost_table): Likewise.
15128 (cortexa57_tunings): Update to use generic_branch_cost.
15129 (cortexa72_tunings): Likewise.
15130 (cortexa73_tunings): Likewise.
15131 (qdf24xx_tunings): Likewise.
15133 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15135 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15136 (thunderx2t99_branch_cost): Likewise.
15137 (cortexa35_tunings): Update to use generic_branch_cost.
15138 (cortexa53_tunings): Likewise.
15139 (cortexa57_tunings): Likewise.
15140 (cortexa72_tunings): Likewise.
15141 (cortexa73_tunings): Likewise.
15142 (thunderx2t99_tunings): Likewise.
15144 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15146 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15147 (sparc_option_override): Honour MASK_FSMULD.
15148 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15149 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15150 * config/sparc/sparc.opt (mfsmuld): New option.
15151 * doc/invoke.texi (mfsmuld): Document option.
15153 2017-07-26 Marek Polacek <polacek@redhat.com>
15155 PR middle-end/70992
15156 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15158 2017-07-26 Richard Biener <rguenther@suse.de>
15160 * gimple-match-head.c (do_valueize): Return OP if valueize
15162 (get_def): New helper to get at the def stmt of a SSA name
15163 if valueize allows.
15164 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15165 do_valueize to get at the def stmt.
15166 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15168 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15170 PR middle-end/46932
15171 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
15173 2017-07-26 Martin Liska <mliska@suse.cz>
15176 * function.c (expand_function_start): Make expansion of
15177 nonlocal_goto_save_area after parm_birth_insn.
15179 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15181 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
15182 from all CPU target flags enable members.
15184 2017-07-26 Richard Biener <rguenther@suse.de>
15186 * genmatch.c (dt_simplify::gen): Make iterator vars const.
15187 (decision_tree::gen): Make 'type' const.
15188 (write_predicate): Likewise.
15190 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15192 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15193 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
15194 (rs6000_option_override_internal): Likewise.
15195 (rs6000_expand_vector_set): Likewise.
15196 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
15197 (TARGET_UPPER_REGS_SF): Likewise.
15198 (TARGET_UPPER_REGS_DI): Likewise.
15199 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
15200 (TARGET_DIRECT_MOVE_64BIT): Likewise.
15201 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
15202 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15203 (Splitters for DI constants in Altivec registers): Likewise.
15204 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
15205 (vsx_set_v4sf_p9): Likewise.
15206 (vsx_set_v4sf_p9_zero): Likewise.
15207 (vsx_insert_extract_v4sf_p9): Likewise.
15208 (vsx_insert_extract_v4sf_p9_2): Likewise.
15210 2017-07-25 Carl Love <cel@us.ibm.com>
15212 * doc/extend.texi: Update the built-in documentation file for the
15213 existing built-in functions
15214 vector signed char vec_cnttz (vector signed char);
15215 vector unsigned char vec_cnttz (vector unsigned char);
15216 vector signed short vec_cnttz (vector signed short);
15217 vector unsigned short vec_cnttz (vector unsigned short);
15218 vector signed int vec_cnttz (vector signed int);
15219 vector unsigned int vec_cnttz (vector unsigned int);
15220 vector signed long long vec_cnttz (vector signed long long);
15221 vector unsigned long long vec_cnttz (vector unsigned long long);
15223 2017-07-25 Andrew Pinski <apinski@cavium.com>
15225 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
15226 accesses where the use is for the first operand of a BIT_INSERT.
15228 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15231 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15232 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15234 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15236 * config/i386/gstabs.h: Delete.
15237 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15239 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15241 * config/i386/i386.c (ix86_decompose_address): Do not check for
15242 register RTX when looking at index_reg or base_reg.
15243 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15245 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15247 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15248 to update EH info here.
15250 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15252 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15254 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15256 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15258 2017-07-25 Torsten Duwe <duwe@suse.de>
15260 * common.opt: Introduce -fpatchable-function-entry
15261 command line option, and its variables function_entry_patch_area_size
15262 and function_entry_patch_area_start.
15263 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15264 including a two-value parser.
15265 * target.def (print_patchable_function_entry): New target hook.
15266 * targhooks.h (default_print_patchable_function_entry): New function.
15267 * targhooks.c (default_print_patchable_function_entry): Likewise.
15268 * toplev.c (process_options): Switch off IPA-RA if
15269 patchable function entries are being generated.
15270 * varasm.c (assemble_start_function): Look at the
15271 patchable-function-entry command line switch and current
15272 function attributes and maybe generate NOP instructions by
15273 calling the print_patchable_function_entry hook.
15274 * doc/extend.texi: Document patchable_function_entry attribute.
15275 * doc/invoke.texi: Document -fpatchable_function_entry
15276 command line option.
15277 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15279 * doc/tm.texi: Re-generate.
15281 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15284 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15285 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15287 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15289 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15291 2017-07-25 Richard Biener <rguenther@suse.de>
15293 PR tree-optimization/81455
15294 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15295 not walk in cycles when looking for guards.
15297 2017-07-25 Richard Biener <rguenther@suse.de>
15299 PR tree-optimization/81529
15300 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15301 when optimizing backedge uses.
15303 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15305 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15307 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15308 to dl_section_ref. On AIX, append an expression to subtract
15309 the size of the section length to dl_section_ref.
15311 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15313 * configure.ac: If any of the config.* scripts fail, exit 1.
15314 * configure: Regenerate.
15316 2017-07-25 Richard Biener <rguenther@suse.de>
15318 PR middle-end/81546
15319 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15320 of immediate uses, be more verbose on errors.
15322 2017-07-25 Richard Biener <rguenther@suse.de>
15324 PR tree-optimization/81510
15325 * tree-vect-loop.c (vect_is_simple_reduction): When the
15326 reduction stmt is not inside the loop bail out.
15328 2017-07-25 Richard Biener <rguenther@suse.de>
15330 PR tree-optimization/81303
15331 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15332 profitability check against LOOP_VINFO_NITERSM1.
15334 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15336 * domwalk.c (cmp_bb_postorder): Simplify.
15337 (sort_bbs_postorder): New function. Use it...
15338 (dom_walker::walk): ...here to optimize common cases.
15340 2017-07-25 Martin Liska <mliska@suse.cz>
15343 * ipa-visibility.c (function_and_variable_visibility): Make the
15344 redirection just on target that supports aliasing.
15345 Fix GNU coding style.
15347 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15350 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15352 * config/arm/rtems.h: Likewise.
15353 * config/bfin/rtems.h: Likewise.
15354 * config/i386/rtemself.h: Likewise.
15355 * config/lm32/rtems.h: Likewise.
15356 * config/m32c/rtems.h: Likewise.
15357 * config/m68k/rtemself.h: Likewise.
15358 * config/microblaze/rtems.h: Likewise.
15359 * config/mips/rtems.h: Likewise.
15360 * config/moxie/rtems.h: Likewise.
15361 * config/nios2/rtems.h: Likewise.
15362 * config/powerpcspe/rtems.h: Likewise.
15363 * config/rs6000/rtems.h: Likewise.
15364 * config/rtems.h: Likewise.
15365 * config/sh/rtems.h: Likewise.
15366 * config/sh/rtemself.h: Likewise.
15367 * config/sparc/rtemself.h: Likewise.
15369 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15372 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15373 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15374 * tree-ssa-structalias.c (alias_get_name): Same.
15376 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15379 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15380 instructions if no du chain is found.
15382 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15384 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15386 2017-07-25 Richard Biener <rguenther@suse.de>
15388 PR middle-end/81505
15389 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15392 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15394 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15395 upper-regs options.
15396 (ISA_2_7_MASKS_SERVER): Likewise.
15397 (ISA_3_0_MASKS_IEEE): Likewise.
15398 (OTHER_P8_VECTOR_MASKS): Likewise.
15399 (OTHER_VSX_VECTOR_MASKS): Likewise.
15400 (POWERPC_MASKS): Likewise.
15401 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15402 duplicate list of options.
15403 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15404 explicit -mupper-regs options.
15405 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15406 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15407 alias for -mupper-regs-df.
15408 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15409 (rs6000_init_hard_regno_mode_ok): Likewise.
15410 (rs6000_option_override_internal): Likewise.
15411 (rs6000_opt_masks): Likewise.
15412 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15413 options in terms of whether -mvsx or -mpower8-vector was used.
15414 (TARGET_UPPER_REGS_DI): Likewise.
15415 (TARGET_UPPER_REGS_SF): Likewise.
15416 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15417 -mupper-regs-* options.
15419 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15421 * passes.c (emergency_dump_function): Print some empty lines and a
15422 header before the RTL dump.
15424 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15426 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15428 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15431 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15432 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15434 2017-07-24 Carl Love <cel@us.ibm.com>
15436 * config/rs6000/rs6000-c.c: Add support for built-in functions
15437 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15438 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15439 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15440 vec_extract_fp_from_shortl): Add defines for the two builtins.
15441 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15442 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15444 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15445 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15446 * doc/extend.texi: Update the built-in documentation file for the
15447 new built-in function.
15449 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15452 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15454 * doc/generic.texi: Likewise.
15455 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15456 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15458 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15460 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15461 (aarch64_mls_elt_merge<mode>): Likewise.
15463 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15465 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15466 having __cxa_atexit.
15468 2017-07-23 Michael Collison <michael.collison@arm.com>
15470 * config/arm/arm.c (arm_option_override): Deprecate
15471 use of -mstructure-size-boundary.
15472 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15473 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15475 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15478 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15479 Reduce cost estimate for direct moves.
15481 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15484 * config/i386/i386.c (ix86_option_override_internal): Disable
15485 BMI, BMI2 and TBM instructions for -m16.
15487 2017-07-21 Carl Love <cel@us.ibm.com>
15489 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15490 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15491 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15492 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15493 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15494 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15495 VMULOSW): New enum "unspec" values.
15496 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15497 altivec_vmulosw): New patterns.
15498 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15499 VMULOSW): Add definitions.
15501 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15503 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15504 (qdf24xx): Likewise.
15505 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15506 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15507 (AARCH64_FL_V8_1): Renumber.
15508 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15509 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15510 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15511 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15512 rdma to feature modifiers list.
15514 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15516 PR middle-end/56727
15517 * ipa-visibility (function_and_variable_visibility): Convert
15518 recursive PLT call to direct call if appropriate.
15520 2017-07-21 Andrew Pinski <apinski@cavium.com>
15522 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15523 operand 1 to see if the types precision matches.
15524 * fold-const.c (operand_equal_p): Likewise.
15526 2017-07-21 Richard Biener <rguenther@suse.de>
15528 PR tree-optimization/81303
15529 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15530 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15531 (vect_peeling_hash_get_lowest_cost): Adjust.
15532 (vect_enhance_data_refs_alignment): Likewise. Use
15533 vect_get_peeling_costs_all_drs to compute the penalty for no
15534 peeling to match up costs.
15536 2017-07-21 Richard Biener <rguenther@suse.de>
15538 PR tree-optimization/81500
15539 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15540 we didn't identify a reduction path.
15542 2017-07-21 Tom de Vries <tom@codesourcery.com>
15543 Cesar Philippidis <cesar@codesourcery.com>
15545 PR gcov-profile/81442
15546 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15549 2017-07-21 Tom de Vries <tom@codesourcery.com>
15552 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15554 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15555 nvptx_override_options_after_change.
15557 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15559 * dwarf2out.c (output_file_names): Avoid double testing for
15560 dwarf_version >= 5.
15562 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15564 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15566 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15568 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15570 (try_crossjump_to_edge): Do not punt on partitioned functions.
15572 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15574 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15575 Put all BBs reachable only via paths crossing cold region to cold
15577 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15579 2016-07-21 Richard Biener <rguenther@suse.de>
15581 PR tree-optimization/81303
15582 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15583 into account prologue and epilogue iterations when raising
15584 min_profitable_iters to sth at least covering one vector iteration.
15586 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15588 * config/arm/arm.c (arm_test_cpu_arch_dat):
15591 2017-07-20 Nathan Sidwell <nathan@acm.org>
15593 Remove TYPE_METHODS.
15594 * tree.h (TYPE_METHODS): Delete.
15595 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15596 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15597 (dbxout_type_methods): Scan TYPE_FIELDS.
15598 (dbxout_type): Don't check TYPE_METHODS here.
15599 * function.c (use_register_for_decl): Always ignore register for
15600 class types when not optimizing.
15601 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15602 * tree.c (free_lang_data_in_type): Stitch out member functions and
15603 templates from TYPE_FIELDS.
15604 (build_distinct_type_copy, verify_type_variant,
15605 verify_type): Member fns are on TYPE_FIELDS.
15606 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15607 * tree-pretty-print.c (dump_generic_node): Likewise.
15609 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15612 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15613 V2TImode and V4TImode.
15614 (ix86_expand_vector_extract): Likewise.
15615 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15616 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15617 (ssescalarmode): Handle V4TImode and V2TImode.
15618 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15619 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15620 (VEXTRACTI128_MODE): New mode iterator.
15621 (splitter for *vec_extractv?ti first element): New.
15622 (VEC_INIT_MODE): New mode iterator.
15623 (vec_init<mode>): Consolidate 3 expanders into one using
15624 VEC_INIT_MODE mode iterator.
15626 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15628 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15629 non_spilled_static_chain_regno_p.
15631 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15633 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15635 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15637 * bb-reorder.c (connect_traces): Allow copying of blocks within
15640 2017-07-20 Richard Biener <rguenther@suse.de>
15642 * gimple.h (gimple_phi_result): Add gphi * overload.
15643 (gimple_phi_result_ptr): Likewise.
15644 (gimple_phi_arg): Likewise. Adjust index assert to only
15645 allow actual argument accesses rather than all slots available
15647 (gimple_phi_arg_def): Add gphi * overload.
15648 * tree-phinodes.c (make_phi_node): Initialize only actual
15650 (resize_phi_node): Clear memory not covered by old node,
15651 do not initialize excess argument slots.
15652 (reserve_phi_args_for_new_edge): Initialize new argument slot
15655 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15657 PR tree-optimization/81388
15659 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15661 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15662 by removing computation of may_be_zero.
15664 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15665 Tom de Vries <tom@codesourcery.com>
15667 PR middle-end/81030
15668 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15669 when gimple level profile disagrees with what RTL expander did.
15671 2017-07-20 Richard Biener <rguenther@suse.de>
15673 PR tree-optimization/61171
15674 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
15675 (vect_analyze_stmt): Add slp instance parameter.
15676 (vectorizable_reduction): Likewise.
15677 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
15678 (vect_is_simple_reduction): Deal with chains not detected
15679 as SLP reduction chain, specifically not properly associated
15680 chains containing a mix of plus/minus.
15681 (get_reduction_op): Remove.
15682 (get_initial_defs_for_reduction): Simplify, pass in whether
15683 this is a reduction chain, pass in the SLP node for the PHIs.
15684 (vect_create_epilog_for_reduction): Get the SLP instance as
15686 (vectorizable_reduction): Get the SLP instance as arg.
15687 During analysis remember the SLP node with the PHIs in the
15688 instance. Simplify getting at the vectorized reduction PHIs.
15689 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15690 through SLP instance.
15691 (vect_slp_analyze_operations): Likewise.
15692 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15693 (vect_transform_stmt): Likewise.
15695 2017-07-20 Tom de Vries <tom@codesourcery.com>
15697 PR tree-optimization/81489
15698 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15699 read of phi arg location to before loop that modifies phi.
15701 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15703 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15706 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15708 PR middle-end/81331
15709 * except.c (execute): Fix ordering issue.
15711 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15713 PR rtl-optimization/81423
15714 * combine.c (make_compound_operation_int): Don't try to optimize
15715 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15717 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15719 PR rtl-optimization/81423
15720 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15721 with a constant that is -1 in the truncated to mode.
15723 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15725 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15726 (determine_unlikely_bbs): ... here.
15727 * predict.h (propagate_unlikely_bbs_forward): Declare.
15728 * cfgexpand.c (pass_expand::execute): Use it.
15729 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15731 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15732 propagate_unlikely_bbs_forward.
15734 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15736 PR middle-end/81331
15737 * except.c (maybe_add_nop_after_section_switch): New function.
15740 2017-07-19 Tom de Vries <tom@codesourcery.com>
15742 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15744 2017-07-19 Tom de Vries <tom@codesourcery.com>
15746 * gimple.h (gimple_phi_arg): Make assert more strict.
15748 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15750 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15751 * config/rs6000/mmintrin.h: New file.
15752 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15754 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15756 PR tree-optimization/81346
15757 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15759 2017-07-19 Tom de Vries <tom@codesourcery.com>
15761 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15763 2017-07-19 Tom de Vries <tom@codesourcery.com>
15765 * config/nvptx/nvptx-modes.def: Add V2DImode.
15766 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15767 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15768 (nvptx_output_mov_insn): Handle lack of mov.b128.
15769 (nvptx_print_operand): Handle 'H' and 'L' codes.
15770 (nvptx_vector_mode_supported): Allow V2DImode.
15771 (nvptx_preferred_simd_mode): New function.
15772 (nvptx_data_alignment): New function.
15773 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15774 nvptx_preferred_simd_mode.
15775 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15777 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15779 2017-07-19 Tom de Vries <tom@codesourcery.com>
15781 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15782 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15783 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15784 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15785 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15786 (mov<VECIM>_insn): New define_insn.
15787 (define_expand "mov<VECIM>): New define_expand.
15789 2017-07-19 Tom de Vries <tom@codesourcery.com>
15791 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15793 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15795 PR tree-optimization/81346
15796 * fold-const.h (fold_div_compare, range_check_type): Declare.
15797 * fold-const.c (range_check_type): New function.
15798 (build_range_check): Use range_check_type.
15799 (fold_div_compare): No longer static, rewritten into
15800 a match.pd helper function.
15801 (fold_comparison): Don't call fold_div_compare here.
15802 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15803 as helper function.
15805 2017-07-19 Nathan Sidwell <nathan@acm.org>
15807 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15808 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15809 * tree.c (find_decls_types_r, verify_type): Use
15810 TYPE_{MIN,MAX}_VALUE_RAW.
15811 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15812 (hash_tree): Likewise.
15813 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15815 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15818 2017-07-18 Tom de Vries <tom@codesourcery.com>
15820 PR middle-end/81464
15821 * omp-expand.c (expand_omp_for_static_chunk): Handle
15822 equal-argument loop exit phi.
15824 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15827 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15828 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15829 operand 2 predicate.
15830 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15831 operand 2 predicate.
15832 (ror,rol -> rorx splitters): Use const_int_operand as
15833 operand 2 predicate.
15835 2017-06-18 Richard Biener <rguenther@suse.de>
15837 PR tree-optimization/81410
15838 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15839 the gap in the ! slp_perm SLP case after each group.
15841 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15843 PR middle-end/81463
15844 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15847 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15849 PR middle-end/81462
15850 * predict.c (set_even_probabilities): Cleanup; do not affect
15851 probabilities that are already known.
15852 (combine_predictions_for_bb): Call even when count is set.
15854 2017-07-18 Nathan Sidwell <nathan@acm.org>
15856 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15859 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15862 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15863 optimization for loop niter analysis.
15865 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15868 * config/avr/avr.c (avr_optimize_casesi): Don't use
15869 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15871 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15873 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15874 body_cost_vec from _vect_peel_extended_info.
15875 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15876 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15879 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15881 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15883 2017-07-18 Richard Biener <rguenther@suse.de>
15885 PR tree-optimization/80620
15886 PR tree-optimization/81403
15887 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15888 info when re-using a VN table entry.
15890 2017-07-18 Richard Biener <rguenther@suse.de>
15892 PR tree-optimization/81418
15893 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15894 vectype_in. Verify that with lane-reducing reduction operations
15895 we have a single def-use cycle.
15897 2017-07-17 Carl Love <cel@us.ibm.com>
15899 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15901 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15902 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15903 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15904 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15905 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15906 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15907 VMULOSW): New enum "unspec" values.
15908 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15909 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15910 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15911 altivec_vmulosw): New patterns.
15912 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15913 VMULOSW): Add definitions.
15914 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15916 * config/alpha/alpha.c: Include predict.h.
15918 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15920 * tree-vrp.c (compare_assert_loc): Fix comparison function
15921 to return predictable results.
15923 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15925 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15927 (subdi3): Likewise.
15928 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15929 * doc/invoke.texi (mexpand-adddi): Update text.
15931 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15933 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15934 that also clobbers the CC register. The old expand code is moved
15936 (*arc_clzsi2): ... here.
15937 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15938 the CC register. The old expand code is moved to ...
15939 (arc_ctzsi2): ... here.
15941 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15943 * config/arc/arc.opt (mindexed-loads): Use initial value
15944 TARGET_INDEXED_LOADS_DEFAULT.
15945 (mauto-modify-reg): Use initial value
15946 TARGET_AUTO_MODIFY_REG_DEFAULT.
15947 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15948 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15949 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15950 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15952 2017-07-17 Martin Liska <mliska@suse.cz>
15955 * opts.c (finish_options): Do not allow -fgnu-tm
15956 w/ -fsanitize={kernel-,}address. Say sorry.
15958 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15961 * tree-loop-distribution.c (classify_partition): Only assert on
15962 numer of iterations.
15963 (merge_dep_scc_partitions): Delete prameter. Update function call.
15964 (distribute_loop): Remove code handling loop with unknown niters.
15965 (pass_loop_distribution::execute): Skip loop with unknown niters.
15967 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15970 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15971 function sort_partitions_by_post_order.
15973 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15975 PR tree-optimization/81374
15976 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15977 the max index of basic blocks, rather than number of basic blocks.
15979 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15981 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15983 (arc_legitimate_pic_operand_p): Likewise.
15984 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
15986 (arc_needs_pcl_p): Likewise.
15987 (arc_legitimate_pc_offset_p): Likewise.
15988 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
15989 function is also used in constrains.md.
15990 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
15991 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
15992 PLUS. Only return true/false in known cases, otherwise assert.
15993 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
15994 is already called in arc_legitimate_constant_p.
15995 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
15997 (LEGITIMATE_PIC_OPERAND_P): Use
15998 arc_raw_symbolic_reference_mentioned_p function.
15999 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16004 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16005 Andrew Burgess <andrew.burgess@embecosm.com>
16007 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16008 (arc_return_address_register): New function.
16009 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16010 (arc_handle_fndecl_attribute): Add naked attribute.
16011 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16012 (TARGET_WARN_FUNC_RETURN): Likewise.
16013 (arc_allocate_stack_slots_for_args): New function.
16014 (arc_warn_func_return): Likewise.
16015 (machine_function): Change type fn_type.
16016 (arc_compute_function_type): Consider new naked function type,
16017 change function return type.
16018 (arc_must_save_register): Adapt to handle new
16019 arc_compute_function_type's return type.
16020 (arc_expand_prologue): Likewise.
16021 (arc_expand_epilogue): Likewise.
16022 (arc_return_address_regs): Delete.
16023 (arc_return_address_register): New function.
16024 (arc_epilogue_uses): Use above function.
16025 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16026 (arc_function_type): Change encoding, add naked type.
16027 (ARC_INTERRUPT_P): Change to handle the new encoding.
16028 (ARC_FAST_INTERRUPT_P): Likewise.
16029 (ARC_NORMAL_P): Define.
16030 (ARC_NAKED_P): Likewise.
16031 (arc_compute_function_type): Delete prototype.
16032 * config/arc/arc.md (in_ret_delay_slot): Use
16033 arc_return_address_register function.
16034 (simple_return): Likewise.
16035 (p_return_i): Likewise.
16037 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16039 PR tree-optimization/81428
16040 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16041 can't be built for those types.
16043 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16045 Remove stuff dead since r239246.
16047 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16048 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16049 (avr_inform_devices): Remove dead stuff.
16051 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16053 * config/arm/arm_neon.h: Fix softp typo.
16055 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16057 PR tree-optimization/81365
16058 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16059 aggregate moves onto bb predecessor edges, make sure there are no
16060 loads that could alias the lhs in between the start of bb and the
16063 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16066 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16067 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16068 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16070 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16072 PR tree-optimization/81396
16073 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16074 (init_symbolic_number): Initialize it to 1.
16075 (perform_symbolic_merge): Add n_ops from both operands into the new
16077 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16078 without base_addr as useless if they need more than one operation.
16079 (bswap_replace): Handle !bswap case for NULL base_addr.
16081 2017-07-17 Tom de Vries <tom@codesourcery.com>
16084 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16087 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16089 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16090 conditional builtin define __FIX_LEON3FT_B2BST.
16092 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16094 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16095 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16098 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16100 PR rtl-optimization/81424
16101 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16102 to remove potential trapping from operands if -fnon-call-exceptions.
16104 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16106 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16107 profile_proability for scalling.
16108 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16110 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16112 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16114 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16116 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16117 fixpoint arithmetics.
16119 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16121 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16122 fixpoint arithmetics.
16124 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16126 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16127 fixpoint arithmetics.
16129 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16131 * profile-count.h (profile_probability::from_reg_br_prob_note,
16132 profile_probability::to_reg_br_prob_note): New functions.
16133 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16134 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16135 * predict.c (probability_reliable_p): Update.
16136 (edge_probability_reliable_p): Update.
16137 (br_prob_note_reliable_p): Update.
16138 (invert_br_probabilities): Update.
16139 (add_reg_br_prob_note): New function.
16140 (combine_predictions_for_insn): Update.
16141 * asan.c (asan_clear_shadow): Update.
16142 * cfgbuild.c (compute_outgoing_frequencies): Update.
16143 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16144 (update_br_prob_note): Update.
16145 (rtl_verify_edges): Update.
16146 (purge_dead_edges): Update.
16147 (fixup_reorder_chain): Update.
16148 * emit-rtl.c (try_split): Update.
16149 * ifcvt.c (cond_exec_process_insns): Update.
16150 (cond_exec_process_if_block): Update.
16151 (dead_or_predicable): Update.
16152 * internal-fn.c (expand_addsub_overflow): Update.
16153 (expand_neg_overflow): Update.
16154 (expand_mul_overflow): Update.
16155 * loop-doloop.c (doloop_modify): Update.
16156 * loop-unroll.c (compare_and_jump_seq): Update.
16157 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16158 * predict.h: Update.
16159 * reorg.c (mostly_true_jump): Update.
16161 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16162 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16163 * config/arc/arc.c: (emit_unlikely_jump): Update.
16164 * config/arm/arm.c: (emit_unlikely_jump): Update.
16165 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16166 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16167 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16168 (ix86_print_operand): Update.
16169 (ix86_split_fp_branch): Update.
16170 (predict_jump): Update.
16171 * config/ia64/ia64.c (ia64_print_operand): Update.
16172 * config/mmix/mmix.c (mmix_print_operand): Update.
16173 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
16174 (rs6000_expand_split_stack_prologue): Update.
16175 * config/rs6000/rs6000.c: Update.
16176 * config/s390/s390.c (s390_expand_vec_strlen): Update.
16177 (s390_expand_vec_movstr): Update.
16178 (s390_expand_cs_tdsi): Update.
16179 (s390_expand_split_stack_prologue): Update.
16180 * config/sh/sh.c (sh_print_operand): Update.
16181 (expand_cbranchsi4): Update.
16182 (expand_cbranchdi4): Update.
16183 * config/sparc/sparc.c (output_v9branch): Update.
16184 * config/spu/spu.c (get_branch_target): Update.
16185 (ea_load_store_inline): Update.
16186 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
16187 * config/tilepro/tilepro.c: Update.
16189 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16191 * gimplify.c (mostly_copy_tree_r): Revert latest change.
16192 (gimplify_save_expr): Likewise.
16194 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16196 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
16198 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16200 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
16202 * timevar.def (TV_IPA_FNSUMMARY): Define.
16204 2017-07-16 Daniel Cederman <cederman@gaisler.com>
16206 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
16207 to back store errata sensitive sequence from being generated.
16208 (sqrtdf2_fix): Likewise.
16210 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16212 * tree-ssa-threadupdate.c (compute_path_counts,
16213 update_joiner_offpath_counts): Use profile_probability.
16215 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
16218 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16220 * config/arm/arm-c.c (arm_cpu_builtins): Define
16221 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16223 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16225 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16226 array entries to represent __ieee128 versions of the
16227 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
16228 scalar_extract_sig, and scalar_insert_exp built-in functions.
16229 (altivec_resolve_overloaded_builtin): Add special case handling
16230 for the __builtin_scalar_insert_exp function, as represented by
16231 the P9V_BUILTIN_VEC_VSIEDP constant.
16232 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
16233 exponent support for __ieee128 argument.
16234 (VSESQP): Add scalar extract signature support for __ieee128
16236 (VSTDCNQP): Add scalar test negative support for __ieee128
16238 (VSIEQP): Add scalar insert exponent support for __int128 argument
16239 with __ieee128 result.
16240 (VSIEQPF): Add scalar insert exponent support for __ieee128
16241 argument with __ieee128 result.
16242 (VSTDCQP): Add scalar test data class support for __ieee128
16244 (VSTDCNQP): Add overload support for scalar test negative with
16245 __ieee128 argument.
16246 (VSTDCQP): Add overload support for scalar test data class
16247 __ieee128 argument.
16248 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16249 UNSPEC_VSX_SXSIGDP.
16250 (UNSPEC_VSX_SIEXPQP): New constant.
16251 (xsxexpqp): New insn for VSX scalar extract exponent quad
16253 (xsxsigqp): New insn for VSX scalar extract significand quad
16255 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16256 precision with floating point argument.
16257 (xststdcqp): New expand for VSX scalar test data class quad
16259 (xststdcnegqp): New expand for VSX scalar test negative quad
16261 (xststdcqp): New insn to match expansions for VSX scalar test data
16262 class quad precision and VSX scalar test negative quad precision.
16263 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16264 special case operand checking to enforce that second operand of
16265 VSX scalar test data class with quad precision argument is a 7-bit
16267 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16268 prototypes and descriptions of __ieee128 versions of
16269 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16270 scalar_test_data_class, and scalar_test_neg built-in functions.
16272 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16274 PR tree-optimization/81162
16275 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16276 replace a negate with an add.
16278 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16280 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16282 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16284 * config/arm/arm-c.c (arm_cpu_builtins): Define
16285 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16287 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16289 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16290 (armv8-r): Set ARM Cortex-R52 as default CPU.
16291 * config/arm/arm-tables.opt: Regenerate.
16292 * config/arm/arm-tune.md: Regenerate.
16293 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16295 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16296 extension for -mcpu=cortex-r52.
16298 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16300 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16301 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16302 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16303 (fp-armv8): Define it as FP_ARMv8 only.
16304 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16305 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16307 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16308 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16309 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16310 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16311 than TARGET_FPU_ARMV8.
16312 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16313 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16314 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16316 * config/arm/neon.md (neon_vrint): Likewise.
16317 (neon_vcvt): Likewise.
16318 (neon_<fmaxmin_op><mode>): Likewise.
16319 (<fmaxmin><mode>3): Likewise.
16320 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16321 * config/arm/predicates.md (arm_cond_move_operator): Check against
16322 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16324 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16326 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16327 to top of function.
16329 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16331 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16332 loop in comment with memset.
16334 2017-07-14 Martin Liska <mliska@suse.cz>
16336 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16337 * dwarf2out.c (is_java): Remove the function.
16338 (output_pubname): Remove usage of the function.
16339 (lower_bound_default): Remove usage of DW_LANG_Java.
16340 (gen_compile_unit_die): Likewise.
16341 * gcc.c: Remove compiler defaults for .java and .zip files.
16342 * gimple-expr.c (remove_suffix): Change as there's no longer
16343 extension than 4-letter one.
16344 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16345 (gimplify_save_expr): Likewise.
16346 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16347 as it's possible even for other languages than Java.
16348 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16349 * lto-opts.c (lto_write_options): Remove reference to Java.
16350 * opts.c (strip_off_ending): Update file extension handling.
16351 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16352 * tree-eh.c (lower_resx): Likewise.
16353 * tree.c (free_lang_data_in_type): Remove dead code.
16354 (find_decls_types_r): Likewise.
16355 (build_common_builtin_nodes): Remove Java from a comment.
16356 (verify_type): Remove dead code.
16357 * varasm.c (assemble_external): Remove Java from a comment.
16359 2017-07-14 Martin Liska <mliska@suse.cz>
16361 * opts.c (finish_options): Add quotes.
16362 (common_handle_option): Likewise.
16364 2017-07-14 Martin Liska <mliska@suse.cz>
16366 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16367 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16368 Remove N_SO_PASCAL.
16369 * dwarf2out.c (lower_bound_default): Do not handle
16371 (gen_compile_unit_die): Likewise.
16372 * gcc.c: Remove default extension binding for GNU Pascal.
16373 * stmt.c: Remove Pascal language from a comment.
16374 * xcoffout.c: Likewise.
16376 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16379 * diagnostic-show-locus.c (fixit_cmp): New function.
16380 (layout::layout): Sort m_fixit_hints.
16381 (column_range::column_range): Assert that the values are valid.
16382 (struct char_span): New struct.
16383 (correction::overwrite): New method.
16384 (struct source_line): New struct.
16385 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16386 calls in terms of classes source_line and char_span, and
16387 correction::overwrite.
16388 (selftest::test_overlapped_fixit_printing_2): New function.
16389 (selftest::diagnostic_show_locus_c_tests): Call it.
16391 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16393 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16394 early if there is no lhs.
16396 2017-07-13 Martin Liska <mliska@suse.cz>
16398 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16399 (gen_reference_type_die): Likewise.
16400 * stor-layout.c: Remove Pascal-related comment.
16402 2017-07-13 Martin Liska <mliska@suse.cz>
16404 * opts.c (finish_options): Add quotes to error messages.
16405 (parse_sanitizer_options): Likewise.
16407 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16409 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16411 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16413 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16415 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16417 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16419 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16421 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16424 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16425 provides the hardware capability bits, define the macro
16426 __BUILTIN_CPU_SUPPORTS__.
16427 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16428 if GLIBC does not provide the hardware capability bits. Add a
16429 gcc_unreachable call if the built-in cpu function is neither
16430 __builtin_cpu_is nor __builtin_cpu_supports.
16431 (rs6000_get_function_versions_dispatcher): Change the warning
16432 that an old GLIBC is used which does not export the capability
16433 bits to be an error.
16434 * doc/extend.texi (target_clones attribute): Document the
16435 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16436 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16437 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16438 the macros defined by GCC if the newer GLIBC is available.
16440 2017-07-12 Jeff Law <law@redhat.com>
16442 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16443 remaining includes slightly.
16444 * config/riscv/riscv-builtins.c: Include profile-count.h.
16446 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16449 * config/avr/avr.c (avr_set_current_function): In diagnostic
16450 messages: Quote keywords and (parts of) identifiers.
16451 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16454 2017-07-12 Carl Love <cel@us.ibm.com>
16456 * config/rs6000/rs6000-c.c: Add support for built-in functions
16457 vector bool char vec_revb (vector bool char);
16458 vector bool short vec_revb (vector short char);
16459 vector bool int vec_revb (vector bool int);
16460 vector bool long long vec_revb (vector bool long long);
16461 * doc/extend.texi: Update the built-in documentation file for the
16462 new built-in functions.
16464 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16466 * config/s390/s390.md: Remove movcc splitter.
16468 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16470 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16471 load/store on condition.
16473 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16476 * config/avr/avr.c (avr_encode_section_info)
16477 [progmem && !TREE_READONLY]: Error if progmem object needs
16480 2017-07-11 Michael Collison <michael.collison@arm.com>
16482 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16485 2017-07-11 Carl Love <cel@us.ibm.com>
16487 * config/rs6000/rs6000-c.c: Add support for builtins
16488 vector unsigned int vec_parity_lsbb (vector signed int);
16489 vector unsigned int vec_parity_lsbb (vector unsigned int);
16490 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16491 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16492 vector unsigned long long vec_parity_lsbb (vector signed long long);
16493 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16494 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16495 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16496 * doc/extend.texi: Update the built-in documentation file for the
16497 new built-in functions.
16499 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16501 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16502 (layout::m_primary_loc): New field.
16503 (layout::layout): Initialize new field. Move location filtering
16504 logic from here to...
16505 (layout::maybe_add_location_range): ...this new method. Add
16506 support for filtering to just the lines already specified by other
16508 (layout::will_show_line_p): New method.
16509 (gcc_rich_location::add_location_if_nearby): New method.
16510 (selftest::test_add_location_if_nearby): New test function.
16511 (selftest::diagnostic_show_locus_c_tests): Call it.
16512 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16515 2017-07-11 Tom de Vries <tom@codesourcery.com>
16517 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16518 (bb_first_real_insn): New function.
16519 (nvptx_single): Add extra initialization of broadcasted condition
16522 2017-07-11 Nathan Sidwell <nathan@acm.org>
16524 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16526 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16528 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16529 Binutils doc as TEXI will mess them up.
16530 * doc/invoke.texi (AVR Options): Same here.
16532 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16534 * config/sparc/sparc.opt (mfix-ut700): New option.
16535 (mfix-gr712rc): Likewise.
16536 (sparc_fix_b2bst): New variable.
16537 * doc/invoke.texi (SPARC options): Document them.
16538 (ARM options): Fix warnings.
16539 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16540 instructions to prevent sequences that can trigger the store-store
16541 errata for certain LEON3FT processors.
16542 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16543 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16544 * config/sparc/sparc.md (fix_b2bst): New attribute.
16545 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16547 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16550 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16552 (*rsqrtsf2_sse): Ditto.
16554 (div<mode>3): Macroize insn from divdf3 and divsf3
16555 using MODEF mode iterator.
16557 2017-07-10 Martin Sebor <msebor@redhat.com>
16559 PR tree-optimization/80397
16560 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16561 instead of testing for equality to INTEGER_TYPE.
16563 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16565 * config.gcc: Remove uclibc from arc target spec.
16567 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16569 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16571 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16574 * lto-wrapper.c (remove_option): New function.
16575 (merge_and_complain): Merge PIC/PIE options more realistically.
16577 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16579 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16583 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16584 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16585 * config.in: Regenerate.
16586 * configure: Regenerate.
16587 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16588 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16589 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16590 (TARGET_GASISR_PROLOGUES): ...target mask.
16591 * common/config/avr/avr-common.c
16592 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16593 Set -mgas-isr-prologues.
16594 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16595 INSERT_PASS_BEFORE for it.
16596 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16597 * config/avr/avr.c (avr_option_override)
16598 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16599 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16600 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16601 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16602 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16604 (make_avr_pass_pre_proep): New function.
16605 (emit_push_sfr) <treg>: Add argument to function and use it
16606 instead of TMP_REG.
16607 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16608 and set machine->gasisr.yes.
16609 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16610 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16611 __gcc_isr.n_pushed to .L__stack_usage.
16612 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16613 (avr_asm_final_postscan_insn): ...this new static function.
16614 * config/avr/avr.h (machine_function)
16615 <is_no_gccisr, use_L__stack_usage>: New fields.
16616 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16617 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16618 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16619 (gasisr, *gasisr): New expander and insn.
16620 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16621 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16622 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16624 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16626 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16629 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16631 Move jump-tables out of .text again.
16634 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16635 (ASM_OUTPUT_ADDR_VEC): New function.
16636 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16637 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16638 INSN_ADDRESSes as asm comment.
16639 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16640 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16641 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16642 * config/avr/avr.md (*tablejump): Adjust comment.
16643 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16644 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16646 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16647 (avr_output_addr_vec): New proto.
16648 (avr_log_t) <insn_addresses>: New field.
16650 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16653 * config/i386/i386.c (ix86_function_arg_advance): Set
16654 outgoing_args_on_stack to true if there are outgoing arguments
16656 (ix86_function_arg): Likewise.
16657 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16658 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16659 * config/i386/i386.h (machine_function): Add
16660 outgoing_args_on_stack.
16662 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16664 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16665 supporting pthreds.
16666 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16668 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16670 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16671 (REAL_H): Remove $(MACHMODE_H).
16672 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
16674 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
16675 $(MACHMODE_H) and double-int.h.
16676 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
16678 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
16681 2017-07-07 Andrew Pinski <apinski@cavium.com>
16683 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
16684 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
16686 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16688 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16689 Add warning if GCC was not configured to link against a GLIBC that
16690 exports the hardware capability bits.
16691 (make_resolver_func): Make resolver function private and not a
16692 COMDAT function. Create the name with clone_function_name instead
16693 of make_unique_name.
16696 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16697 correct operand in doing the split.
16699 2017-07-07 Carl Love <cel@us.ibm.com>
16701 * config/rs6000/rs6000-c: Add support for built-in function
16702 vector unsigned short vec_pack_to_short_fp32 (vector float,
16704 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16705 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16706 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16707 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16708 (convert_4f32_8i16): Add define_expand.
16709 * doc/extend.texi: Update the built-in documentation file for the
16710 new built-in function.
16712 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16714 * config/sparc/m8.md: New file.
16715 * config/sparc/sparc.md: Include m8.md.
16717 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16719 * config/sparc/sparc.opt: New option -mvis4b.
16720 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16721 (sparc_option_override): Handle VIS4B.
16722 (enum sparc_builtins): Define
16723 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16724 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16725 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16726 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16727 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16728 (check_constant_argument): New function.
16729 (sparc_vis_init_builtins): Define builtins
16730 __builtin_vis_dictunpack{8,16,32},
16731 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16732 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16733 __builtin_vis_fpcmpde{8,16,32}shl and
16734 __builtin_vis_fpcmpur{8,16,32}shl.
16735 (sparc_expand_builtin): Check that the constant operands to
16736 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16737 constant and in range.
16738 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16740 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16741 (SPARC_IMM5_P): Likewise.
16742 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16743 (enabled): Handle vis4b.
16744 (UNSPEC_DICTUNPACK): New unspec.
16745 (UNSPEC_FPCMPSHL): Likewise.
16746 (UNSPEC_FPUCMPSHL): Likewise.
16747 (UNSPEC_FPCMPDESHL): Likewise.
16748 (UNSPEC_FPCMPURSHL): Likewise.
16749 (cpu_feature): New CPU feature `vis4b'.
16750 (dictunpack{8,16,32}): New insns.
16751 (FPCSMODE): New mode iterator.
16752 (fpcscond): New code iterator.
16753 (fpcsucond): Likewise.
16754 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16755 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16756 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16757 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16758 * config/sparc/constraints.md: Define constraints `q' for unsigned
16759 2-bit integer constants and `t' for unsigned 5-bit integer
16761 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16763 (imm5_operand_dictunpack16): Likewise.
16764 (imm5_operand_dictunpack32): Likewise.
16765 (imm2_operand): Likewise.
16766 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16767 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16768 ditunpack* and fpcmp*shl builtins.
16770 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16772 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16773 * config.in: Add HAVE_AS_SPARC6 define.
16774 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16776 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16778 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16779 (CPP_CPU_SPEC): Handle m8.
16780 (ASM_CPU_SPEC): Likewise.
16781 * config/sparc/sparc-opts.h (enum processor_type): Add
16783 * config/sparc/sparc.c (m8_costs): New struct.
16784 (sparc_option_override): Handle TARGET_CPU_m8.
16785 (sparc32_initialize_trampoline): Likewise.
16786 (sparc64_initialize_trampoline): Likewise.
16787 (sparc_issue_rate): Likewise.
16788 (sparc_register_move_cost): Likewise.
16789 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16790 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16791 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16792 (CPP_CPU_SPEC): Handle M8.
16793 (ASM_CPU_SPEC): Likewise.
16794 (AS_M8_FLAG): Define.
16795 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16796 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16797 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16799 * configure: Regenerate.
16800 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16803 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16805 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16807 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16808 ("*movdi_insn_sp32"): Do not set v3pipe.
16809 ("*movsi_insn"): Likewise.
16810 ("*movdi_insn_sp64"): Likewise.
16811 ("*movsf_insn"): Likewise.
16812 ("*movdf_insn_sp32"): Likewise.
16813 ("*movdf_insn_sp64"): Likewise.
16814 ("*zero_extendsidi2_insn_sp64"): Likewise.
16815 ("*sign_extendsidi2_insn"): Likewise.
16816 ("*mov<VM32:mode>_insn"): Likewise.
16817 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16818 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16819 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16820 ("<vlop:code><VL:mode>3"): Likewise.
16821 ("*not_<vlop:code><VL:mode>3"): Likewise.
16822 ("*nand<VL:mode>_vis"): Likewise.
16823 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16824 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16825 ("one_cmpl<VL:mode>2"): Likewise.
16826 ("faligndata<VM64:mode>_vis"): Likewise.
16827 ("alignaddrsi_vis"): Likewise.
16828 ("alignaddrdi_vis"): Likweise.
16829 ("alignaddrlsi_vis"): Likewise.
16830 ("alignaddrldi_vis"): Likewise.
16831 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16832 ("bmaskdi_vis"): Likewise.
16833 ("bmasksi_vis"): Likewise.
16834 ("bshuffle<VM64:mode>_vis"): Likewise.
16835 ("cmask8<P:mode>_vis"): Likewise.
16836 ("cmask16<P:mode>_vis"): Likewise.
16837 ("cmask32<P:mode>_vis"): Likewise.
16838 ("pdistn<P:mode>_vis"): Likewise.
16839 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16841 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16843 * config/sparc/sparc.md ("subtype"): New insn attribute.
16844 ("*wrgsr_sp64"): Set insn subtype.
16845 ("*rdgsr_sp64"): Likewise.
16846 ("alignaddrsi_vis"): Likewise.
16847 ("alignaddrdi_vis"): Likewise.
16848 ("alignaddrlsi_vis"): Likewise.
16849 ("alignaddrldi_vis"): Likewise.
16850 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16851 ("fexpand_vis"): Likewise.
16852 ("fpmerge_vis"): Likewise.
16853 ("faligndata<VM64:mode>_vis"): Likewise.
16854 ("bshuffle<VM64:mode>_vis"): Likewise.
16855 ("cmask8<P:mode>_vis"): Likewise.
16856 ("cmask16<P:mode>_vis"): Likewise.
16857 ("cmask32<P:mode>_vis"): Likewise.
16858 ("fchksm16_vis"): Likewise.
16859 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16860 ("fmean16_vis"): Likewise.
16861 ("fp<plusminus_insn>64_vis"): Likewise.
16862 ("<plusminus_insn>v8qi3"): Likewise.
16863 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16864 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16865 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16866 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16867 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16868 ("*movqi_insn"): Likewise.
16869 ("*movhi_insn"): Likewise.
16870 ("*movsi_insn"): Likewise.
16871 ("movsi_pic_gotdata_op"): Likewise.
16872 ("*movdi_insn_sp32"): Likewise.
16873 ("*movdi_insn_sp64"): Likewise.
16874 ("movdi_pic_gotdata_op"): Likewise.
16875 ("*movsf_insn"): Likewise.
16876 ("*movdf_insn_sp32"): Likewise.
16877 ("*movdf_insn_sp64"): Likewise.
16878 ("*zero_extendhisi2_insn"): Likewise.
16879 ("*zero_extendqihi2_insn"): Likewise.
16880 ("*zero_extendqisi2_insn"): Likewise.
16881 ("*zero_extendqidi2_insn"): Likewise.
16882 ("*zero_extendhidi2_insn"): Likewise.
16883 ("*zero_extendsidi2_insn_sp64"): Likewise.
16884 ("ldfsr"): Likewise.
16885 ("prefetch_64"): Likewise.
16886 ("prefetch_32"): Likewise.
16887 ("tie_ld32"): Likewise.
16888 ("tie_ld64"): Likewise.
16889 ("*tldo_ldub_sp32"): Likewise.
16890 ("*tldo_ldub1_sp32"): Likewise.
16891 ("*tldo_ldub2_sp32"): Likewise.
16892 ("*tldo_ldub_sp64"): Likewise.
16893 ("*tldo_ldub1_sp64"): Likewise.
16894 ("*tldo_ldub2_sp64"): Likewise.
16895 ("*tldo_ldub3_sp64"): Likewise.
16896 ("*tldo_lduh_sp32"): Likewise.
16897 ("*tldo_lduh1_sp32"): Likewise.
16898 ("*tldo_lduh_sp64"): Likewise.
16899 ("*tldo_lduh1_sp64"): Likewise.
16900 ("*tldo_lduh2_sp64"): Likewise.
16901 ("*tldo_lduw_sp32"): Likewise.
16902 ("*tldo_lduw_sp64"): Likewise.
16903 ("*tldo_lduw1_sp64"): Likewise.
16904 ("*tldo_ldx_sp64"): Likewise.
16905 ("*mov<VM32:mode>_insn"): Likewise.
16906 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16907 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16909 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16911 * config/sparc/sparc.md ("type"): New insn type viscmp.
16912 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16914 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16915 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16916 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16917 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16919 ("n7_vis_logical_11cycle"): Likewise.
16920 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16921 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16922 * config/sparc/niagara.md ("niag_vis"): Likewise.
16923 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16924 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16926 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16928 * config/sparc/sparc.md: New instruction type `bmask'.
16929 (bmaskdi_vis): Use the `bmask' type.
16930 (bmasksi_vis): Likewise.
16931 * config/sparc/ultra3.md (us3_array): Likewise.
16932 * config/sparc/niagara7.md (n7_array): Likewise.
16933 * config/sparc/niagara4.md (n4_array): Likewise.
16934 * config/sparc/niagara2.md (niag2_vis): Likewise.
16935 (niag3_vis): Likewise.
16936 * config/sparc/niagara.md (niag_vis): Likewise.
16938 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16940 * ipa-comdats.c: Remove optimize check from gate.
16941 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16942 for functions not optimized.
16943 (ipa_fn_summary_read): Skip optimize check.
16944 (ipa_fn_summary_write): Likewise.
16945 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16947 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16949 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16950 (check_callers): Give up on caller which is not optimized.
16951 (inline_small_functions): Likewise.
16952 (ipa_inline): Do not give up when not optimizing.
16953 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16954 away unoptimizes cdtors.
16955 (whole_program_function_and_variable_visibility): Do
16956 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16957 * ipa.c (process_references): Do not check optimize.
16958 (symbol_table::remove_unreachable_nodes): Update optimize check.
16959 (set_writeonly_bit): Update optimize check.
16960 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16961 (pass_ipa_single_use::gate): Remove.
16963 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16965 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16966 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16967 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16968 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16969 permute_load, permute_store, adjust_extract, adjust_splat,
16970 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16971 replace_swap_with_copy, dump_swap_insn_table,
16972 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16973 recombine_lvx_pattern, recombine_stvx_pattern,
16974 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16975 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16976 to file rs6000-p8swap.c.
16977 * config/rs6000/rs6000-p8swap.c: New file.
16978 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16979 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16980 and rs6000*-*-* targets.
16982 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16984 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
16986 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16988 * lto-wrapper.c (merge_and_complain): Do not merge
16989 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
16990 fsigned_zeros, ftrapping_math, fwrapv.
16991 (append_compiler_options): Do not track these options.
16992 (append_linker_options): Likewie
16994 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16996 * cgraphunit.c (cgraph_node::finalize_function): When
16997 !flag_toplevel_reorde set no_reorder flag.
16998 (varpool_node::finalize_decl): Likewise.
16999 (symbol_table::compile): Drop no toplevel reorder path.
17001 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17003 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17004 edges; zero probability is not better than uninitialized.
17006 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17008 * asan.h (asan_sanitize_allocas_p): Declare.
17009 * asan.c (asan_sanitize_allocas_p): New function.
17010 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17011 (handle_builtin_alloca): Likewise.
17012 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17013 if !asan_sanitize_allocas_p.
17014 * params.def (asan-instrument-allocas): Add new option.
17015 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17016 * opts.c (common_handle_option): Disable allocas sanitization for
17019 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17021 * asan.c: Include gimple-fold.h.
17022 (get_last_alloca_addr): New function.
17023 (handle_builtin_stackrestore): Likewise.
17024 (handle_builtin_alloca): Likewise.
17025 (asan_emit_allocas_unpoison): Likewise.
17026 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17027 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17028 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17029 (instrument_builtin_call): Pass gimple iterator to
17030 get_mem_refs_of_builtin_call.
17031 (last_alloca_addr): New global.
17032 * asan.h (asan_emit_allocas_unpoison): Declare.
17033 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17034 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17035 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17036 if function calls alloca.
17037 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17038 * gimple-fold.h (replace_call_with_value): Declare.
17039 * internal-fn.c: Include asan.h.
17040 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17041 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17043 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17045 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17046 (C_SELFTEST_FLAGS): New.
17047 (CPP_SELFTEST_FLAGS): New.
17048 (SELFTEST_DEPS): New, from deps of s-selftest.
17049 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17050 (CPP_SELFTEST_DEPS): New.
17051 (selftest): Add dependency on s-selftest-c++.
17052 (s-selftest): Rename to...
17053 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17054 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17055 than SELFTEST_FLAGS.
17056 (selftest-gdb): Rename to...
17057 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17059 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17060 (selftest-valgrind): Rename to...
17061 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17063 (selftest-valgrind): Reintroduce as an alias for
17064 selftest-c-valgrind.
17065 (s-selftest-c++): New.
17066 (selftest-c++-gdb): New.
17067 (selftest-c++-valgrind): New.
17069 2017-07-06 Olivier Hainque <hainque@adacore.com>
17071 * gcc.c (process_command): When deciding if undefined variables
17072 should be ignored when processing specs, accept "gcc -v" as well.
17074 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17076 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17077 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17079 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17081 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17082 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17083 * config/arm/arm-tables.opt: Regenerate.
17084 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17086 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17088 2017-07-06 Carl Love <cel@us.ibm.com>
17090 * ChangeLog: Clean up from mid air collision
17092 2017-07-06 Carl Love <cel@us.ibm.com>
17094 * config/rs6000/rs6000-c.c: Add support for built-in functions
17095 vector signed int vec_subc (vector signed int, vector signed int);
17096 vector signed __int128 vec_subc (vector signed __int128,
17097 vector signed __int128);
17098 vector unsigned __int128 vec_subc (vector unsigned __int128,
17099 vector unsigned __int128);
17100 vector signed int vec_sube (vector signed int, vector signed int,
17101 vector signed int);
17102 vector unsigned int vec_sube (vector unsigned int,
17103 vector unsigned int,
17104 vector unsigned int);
17105 vector signed __int128 vec_sube (vector signed __int128,
17106 vector signed __int128,
17107 vector signed__int128);
17108 vector unsigned __int128 vec_sube (vector unsigned __int128,
17109 vector unsigned __int128,
17110 vector unsigned __int128);
17111 vector signed int vec_subec (vector signed int, vector signed int,
17112 vector signed int);
17113 vector unsigned int vec_subec (vector unsigned int,
17114 vector unsigned int,
17115 vector unsigned int);
17116 vector signed __int128 vec_subec (vector signed __int128,
17117 vector signed __int128,
17118 vector signed__int128);
17119 vector unsigned __int128 vec_subec (vector unsigned __int128,
17120 vector unsigned __int128,
17121 vector unsigned __int128);
17122 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17123 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17124 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17125 BU_ALTIVEC_OVERLOAD_X definitions.
17126 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17127 * doc/extend.texi: Update the built-in documentation file for the new
17128 built-in functions.
17130 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17133 * diagnostic-show-locus.c (layout::layout): Use start and finish
17134 spelling location for the start and finish of each range.
17135 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17136 Add unused aspect param.
17137 * input.c (expand_location_1): Add "aspect" param, and use it
17138 to access the correct part of the location.
17139 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17141 (expand_location_to_spelling_point): Likewise.
17142 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17143 param, and pass it to expand_location_1.
17145 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17147 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17148 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17149 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17150 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17151 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17152 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17153 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17154 _mm_maskz_getmant_ss): New intrinsics.
17155 (__builtin_ia32_getexpss128_mask): Changed to ...
17156 __builtin_ia32_getexpss128_round ... this.
17157 (__builtin_ia32_getexpsd128_mask): Changed to ...
17158 __builtin_ia32_getexpsd128_round ... this.
17159 * config/i386/i386-builtin-types.def
17160 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17161 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17162 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17163 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17164 __builtin_ia32_getmantss_mask_round): New builtins.
17165 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17166 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17167 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17168 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17169 * config/i386/sse.md
17170 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17171 avx512f_sgetexp<mode><mask_scalar_name>
17172 <round_saeonly_scalar_name> ... this.
17173 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
17174 %0, %1, %2<round_saeonly_op3>}): Changed to ...
17175 vgetexp<ssescalarmodesuffix>
17176 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17177 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
17178 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
17179 avx512f_vgetmant<mode><mask_scalar_name>
17180 <round_saeonly_scalar_name> ... this.
17181 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17182 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
17183 vgetmant<ssescalarmodesuffix>
17184 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
17185 %0<mask_scalar_operand4>, %1, %2
17186 <round_saeonly_scalar_mask_op4>, %3} ... this.
17187 * config/i386/subst.md (mask_scalar_operand4,
17188 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
17189 round_saeonly_scalar_nimm_predicate): New subst attributes.
17191 2017-07-06 Julia Koval <julia.koval@intel.com>
17193 * config/i386/i386.c (ix86_erase_embedded_rounding):
17194 Remove code for old rounding pattern.
17196 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
17198 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
17200 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
17202 * doc/sourcebuild.texi (Test Directives, Variants of
17203 dg-require-support): Add documentation for dg-require-stack-check.
17205 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
17207 * config/i386/subst.md (mask_scalar, round_scalar,
17208 round_saeonly_scalar): New meta-templates.
17209 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
17210 round_scalar_mask_operand3, round_scalar_mask_op3,
17211 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
17212 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
17213 round_saeonly_scalar_constraint,
17214 round_saeonly_scalar_prefix): New subst attribute.
17215 * config/i386/sse.md
17216 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
17217 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
17218 <round_scalar_name> ... this.
17219 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
17220 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
17221 <round_scalar_name> ... this.
17222 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
17223 <sse>_vm<code><mode>3<mask_scalar_name>
17224 <round_saeonly_scalar_name> ... this.
17225 (v<plusminus_mnemonic><ssescalarmodesuffix>
17226 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17227 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17228 v<plusminus_mnemonic><ssescalarmodesuffix>
17229 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17230 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17231 (v<multdiv_mnemonic><ssescalarmodesuffix>
17232 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17233 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17234 v<multdiv_mnemonic><ssescalarmodesuffix>
17235 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17236 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17237 (v<maxmin_float><ssescalarmodesuffix>
17238 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17239 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17240 v<maxmin_float><ssescalarmodesuffix>
17241 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17242 %0<mask_scalar_operand3>, %1, %<iptr>2
17243 <round_saeonly_scalar_mask_op3>} ... this.
17245 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17247 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17248 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17250 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17251 Alan Hayward <alan.hayward@arm.com>
17252 David Sherwood <david.sherwood@arm.com>
17254 * combine.c (simplify_if_then_else): Remove "enum" before
17256 * compare-elim.c (can_eliminate_compare): Likewise.
17257 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17259 (aarch64_lookup_simd_builtin_type): Likewise.
17260 (aarch64_simd_builtin_type): Likewise.
17261 (aarch64_init_simd_builtin_types): Likewise.
17262 (aarch64_simd_expand_args): Likewise.
17263 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17265 (aarch64_reverse_mask): Likewise.
17266 (aarch64_simd_emit_reg_reg_move): Likewise.
17267 (aarch64_gen_adjusted_ldpstp): Likewise.
17268 (aarch64_ccmp_mode_to_code): Likewise.
17269 (aarch64_operands_ok_for_ldpstp): Likewise.
17270 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17271 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17273 (aarch64_min_divisions_for_recip_mul): Likewise.
17274 (aarch64_reassociation_width): Likewise.
17275 (aarch64_get_condition_code_1): Likewise.
17276 (aarch64_simd_emit_reg_reg_move): Likewise.
17277 (aarch64_simd_attr_length_rglist): Likewise.
17278 (aarch64_reverse_mask): Likewise.
17279 (aarch64_operands_ok_for_ldpstp): Likewise.
17280 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17281 (aarch64_gen_adjusted_ldpstp): Likewise.
17282 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17284 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17285 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17286 (arm_lookup_simd_builtin_type): Likewise.
17287 (arm_simd_builtin_type): Likewise.
17288 (arm_init_simd_builtin_types): Likewise.
17289 (arm_expand_builtin_args): Likewise.
17290 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17291 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17292 (ft32_setup_incoming_varargs): Likewise.
17293 (ft32_function_arg): Likewise.
17294 (ft32_function_arg_advance): Likewise.
17295 (ft32_pass_by_reference): Likewise.
17296 (ft32_arg_partial_bytes): Likewise.
17297 (ft32_valid_pointer_mode): Likewise.
17298 (ft32_addr_space_pointer_mode): Likewise.
17299 (ft32_addr_space_legitimate_address_p): Likewise.
17300 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17302 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17303 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17304 (iamcu_alignment): Likewise.
17305 (canonicalize_vector_int_perm): Likewise.
17306 (ix86_noce_conversion_profitable_p): Likewise.
17307 (ix86_mpx_bound_mode): Likewise.
17308 (ix86_operands_ok_for_move_multiple): Likewise.
17309 * config/microblaze/microblaze-protos.h
17310 (microblaze_expand_conditional_branch_reg): Likewise.
17311 * config/microblaze/microblaze.c
17312 (microblaze_expand_conditional_branch_reg): Likewise.
17313 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17315 (rs6000_reassociation_width): Likewise.
17316 (rs6000_invalid_binary_op): Likewise.
17317 (fusion_p9_p): Likewise.
17318 (emit_fusion_p9_load): Likewise.
17319 (emit_fusion_p9_store): Likewise.
17320 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17322 (riscv_hard_regno_mode_ok_p): Likewise.
17323 (riscv_address_insns): Likewise.
17324 (riscv_split_symbol): Likewise.
17325 (riscv_legitimize_move): Likewise.
17326 (riscv_function_value): Likewise.
17327 (riscv_hard_regno_nregs): Likewise.
17328 (riscv_expand_builtin): Likewise.
17329 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17330 (riscv_build_integer): Likewise.
17331 (riscv_split_integer): Likewise.
17332 (riscv_legitimate_constant_p): Likewise.
17333 (riscv_cannot_force_const_mem): Likewise.
17334 (riscv_regno_mode_ok_for_base_p): Likewise.
17335 (riscv_valid_base_register_p): Likewise.
17336 (riscv_valid_offset_p): Likewise.
17337 (riscv_valid_lo_sum_p): Likewise.
17338 (riscv_classify_address): Likewise.
17339 (riscv_legitimate_address_p): Likewise.
17340 (riscv_address_insns): Likewise.
17341 (riscv_load_store_insns): Likewise.
17342 (riscv_force_binary): Likewise.
17343 (riscv_split_symbol): Likewise.
17344 (riscv_force_address): Likewise.
17345 (riscv_legitimize_address): Likewise.
17346 (riscv_move_integer): Likewise.
17347 (riscv_legitimize_const_move): Likewise.
17348 (riscv_legitimize_move): Likewise.
17349 (riscv_address_cost): Likewise.
17350 (riscv_subword): Likewise.
17351 (riscv_output_move): Likewise.
17352 (riscv_canonicalize_int_order_test): Likewise.
17353 (riscv_emit_int_order_test): Likewise.
17354 (riscv_function_arg_boundary): Likewise.
17355 (riscv_pass_mode_in_fpr_p): Likewise.
17356 (riscv_pass_fpr_single): Likewise.
17357 (riscv_pass_fpr_pair): Likewise.
17358 (riscv_get_arg_info): Likewise.
17359 (riscv_function_arg): Likewise.
17360 (riscv_function_arg_advance): Likewise.
17361 (riscv_arg_partial_bytes): Likewise.
17362 (riscv_function_value): Likewise.
17363 (riscv_pass_by_reference): Likewise.
17364 (riscv_setup_incoming_varargs): Likewise.
17365 (riscv_print_operand): Likewise.
17366 (riscv_elf_select_rtx_section): Likewise.
17367 (riscv_save_restore_reg): Likewise.
17368 (riscv_for_each_saved_reg): Likewise.
17369 (riscv_register_move_cost): Likewise.
17370 (riscv_hard_regno_mode_ok_p): Likewise.
17371 (riscv_hard_regno_nregs): Likewise.
17372 (riscv_class_max_nregs): Likewise.
17373 (riscv_memory_move_cost): Likewise.
17374 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17375 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17376 (rl78_addr_space_address_mode): Likewise.
17377 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17379 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17380 (rs6000_reassociation_width): Likewise.
17381 (rs6000_invalid_binary_op): Likewise.
17382 (fusion_p9_p): Likewise.
17383 (emit_fusion_p9_load): Likewise.
17384 (emit_fusion_p9_store): Likewise.
17385 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17386 (ok_for_simple_move_operands): Likewise.
17387 (ok_for_simple_move_strict_operands): Likewise.
17388 (ok_for_simple_arith_logic_operands): Likewise.
17389 (visium_legitimize_reload_address): Likewise.
17390 (visium_select_cc_mode): Likewise.
17391 (output_cbranch): Likewise.
17392 (visium_split_double_move): Likewise.
17393 (visium_expand_copysign): Likewise.
17394 (visium_expand_int_cstore): Likewise.
17395 (visium_expand_fp_cstore): Likewise.
17396 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17397 (visium_function_arg): Likewise.
17398 (visium_function_arg_advance): Likewise.
17399 (visium_libcall_value): Likewise.
17400 (visium_setup_incoming_varargs): Likewise.
17401 (visium_legitimate_constant_p): Likewise.
17402 (visium_legitimate_address_p): Likewise.
17403 (visium_legitimize_address): Likewise.
17404 (visium_secondary_reload): Likewise.
17405 (visium_register_move_cost): Likewise.
17406 (visium_memory_move_cost): Likewise.
17407 (prepare_move_operands): Likewise.
17408 (ok_for_simple_move_operands): Likewise.
17409 (ok_for_simple_move_strict_operands): Likewise.
17410 (ok_for_simple_arith_logic_operands): Likewise.
17411 (visium_function_value_1): Likewise.
17412 (rtx_ok_for_offset_p): Likewise.
17413 (visium_legitimize_reload_address): Likewise.
17414 (visium_split_double_move): Likewise.
17415 (visium_expand_copysign): Likewise.
17416 (visium_expand_int_cstore): Likewise.
17417 (visium_expand_fp_cstore): Likewise.
17418 (visium_split_cstore): Likewise.
17419 (visium_select_cc_mode): Likewise.
17420 (visium_split_cbranch): Likewise.
17421 (output_cbranch): Likewise.
17422 (visium_print_operand_address): Likewise.
17423 * expmed.c (flip_storage_order): Likewise.
17424 * expmed.h (emit_cstore): Likewise.
17425 (flip_storage_order): Likewise.
17426 * genrecog.c (validate_pattern): Likewise.
17427 * hsa-gen.c (gen_hsa_addr): Likewise.
17428 * internal-fn.c (expand_arith_overflow): Likewise.
17429 * ira-color.c (allocno_copy_cost_saving): Likewise.
17430 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17431 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17432 (process_invariant_for_inheritance): Likewise.
17433 * lra-eliminations.c (move_plus_up): Likewise.
17434 * omp-low.c (lower_oacc_reductions): Likewise.
17435 * simplify-rtx.c (simplify_subreg): Likewise.
17436 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17437 (TARGET_CHKP_BOUND_MODE): Likewise..
17438 * targhooks.c (default_chkp_bound_mode): Likewise.
17439 (default_setup_incoming_vararg_bounds): Likewise.
17440 * targhooks.h (default_chkp_bound_mode): Likewise.
17441 (default_setup_incoming_vararg_bounds): Likewise.
17442 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17443 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17444 (have_whole_vector_shift): Likewise.
17445 * tree-vect-stmts.c (vectorizable_load): Likewise.
17446 * doc/tm.texi: Regenerate.
17448 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17450 Graceful degrade if Binutils PR21472 is not available.
17453 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17454 .rodata in flash test fails.
17455 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17456 * confgure: Regenerate.
17457 * config.in: Regenerate.
17458 * config/avr/avr.c (avr_asm_named_section)
17459 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17460 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17461 (avr_asm_init_sections): Same.
17463 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17465 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17466 (fma<VH:mode>4_intrinsic): Likewise.
17467 (*fmsub<VCVTF:mode>4): Likewise.
17468 (*fmsub<VH:mode>4_intrinsic): Likewise.
17470 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17473 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17474 Don't depend on "optimize > 0".
17475 (out_movhi_r_mr, out_movqi_mr_r): Same.
17476 (out_movhi_mr_r, out_movqi_r_mr): Same.
17477 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17478 io_address_operand on "optimize > 0".
17480 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17482 * tree-loop-distribution.c: Add general explanantion on the pass.
17483 (generate_loops_for_partition): Mark distributed loop.
17484 (pg_add_dependence_edges): New parameter. Handle alias data
17485 dependence specially and record it in the parameter if asked.
17486 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17487 (init_partition_graph_vertices, add_partition_graph_edge): New.
17488 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17489 (free_partition_graph_vdata, build_partition_graph): New.
17490 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17491 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17492 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17493 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17494 (version_for_distribution_p, finalize_partitions): New.
17495 (distribute_loop): Handle alias data dependence specially. Factor
17496 out loop fusion code as functions and call these functions.
17498 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17500 * tree-loop-distribution.c (classify_partition): New parameter and
17501 better handle reduction statement.
17502 (rdg_build_partitions): Revise comment.
17503 (distribute_loop): Compute statements in all partitions and pass it
17504 to classify_partition.
17506 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17508 * tree-loop-distribution.c (enum partition_type): New.
17509 (struct partition): New field type.
17510 (partition_merge_into): Add parameter. Update partition type.
17511 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17512 (build_rdg_partition_for_vertex): Compute partition type.
17513 (rdg_build_partitions): Dump partition type.
17514 (distribute_loop): Update calls to partition_merge_into.
17516 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17518 * tree-loop-distribution.c (struct ddr_hasher): New.
17519 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17521 (classify_partition): Call get_data_dependence.
17522 (pg_add_dependence_edges): Ditto.
17523 (distribute_loop): Release data dependence hash table.
17525 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17527 * tree-loop-distribution.c (ref_base_address): Delete.
17528 (similar_memory_accesses): Rename ...
17529 (share_memory_accesses): ... to this. Check if partitions access
17530 the same memory reference.
17531 (distribute_loop): Call share_memory_accesses.
17533 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17535 * tree-loop-distribution.c (struct partition): New field recording
17536 its data reference.
17537 (partition_alloc, partition_free): Init and release data refs.
17538 (partition_merge_into): Merge data refs.
17539 (build_rdg_partition_for_vertex): Collect data refs for partition.
17540 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17542 (distribute_loop): Remve data refs from vertice data of partition
17545 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17547 * tree-loop-distribution.c (params.h): Include header file.
17548 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17549 (datarefs_vec): New global var.
17550 (create_rdg_vertices): Use datarefs_vec directly.
17551 (free_rdg): Don't free data references.
17552 (build_rdg): Update use. Don't free data references.
17553 (distribute_loop): Compute global variable for data references.
17554 Bail out if there are too many data references.
17556 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17558 * tree-loop-distribution.c (loop_nest): New global var.
17559 (build_rdg): Use loop directly, rather than loop nest.
17560 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17562 (distribute_loop): Compute global variable loop nest. Update use.
17564 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17566 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17567 (partition_merge_into): New parameter. Dump reason for fusion.
17568 (distribute_loop): Update use of partition_merge_into.
17570 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17572 * tree-loop-distribution.c (bb_top_order_index): New.
17573 (bb_top_order_index_size, bb_top_order_cmp): New.
17574 (stmts_from_loop): Use topological order.
17575 (pass_loop_distribution::execute): Compute and release topological
17576 order for basic blocks.
17578 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17580 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17583 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17585 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17586 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17587 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17588 * internal-fn.def (LOOP_DIST_ALIAS): New.
17589 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17590 (fold_loop_internal_call): ... this.
17591 (vect_loop_dist_alias_call): New function.
17592 (set_uid_loop_bbs): Call fold_loop_internal_call.
17593 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17596 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17599 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17600 Require dead FLAGS_REG at the beginning of a peephole.
17602 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17605 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17606 arguments in the call to __builtin_ia32_sbb_u32.
17607 (_subborrow_u64): Swap _X and _Y arguments in the call to
17608 __builtin_ia32_sbb_u64.
17610 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17613 * tree-vrp.c (compare_assert_loc): Turn into a function template
17614 with stable template parameter. Only test if a->e is NULL,
17615 !a->e == !b->e has been verified already. Use e == NULL or
17616 e != NULL instead of e or ! e tests. If stable is true, don't use
17617 iterative_hash_expr, on the other side allow a or b or both NULL
17618 and sort the NULLs last.
17619 (process_assert_insertions): Sort using compare_assert_loc<false>
17620 instead of compare_assert_loc, later sort using
17621 compare_assert_loc<true> before calling process_assert_insertions_for
17622 in a loop. Use break instead of continue once seen NULL pointer.
17624 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17626 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17627 Cortex-R7 and Cortex-R8 processors.
17629 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17631 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17632 uninitialized while src is not.
17634 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17636 * common/config/arm/arm-common.c: Adjust include path for
17638 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17639 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17640 (arm-cpu-data.h): Likewise.
17641 (arm-cpu-cdata.h): Likewise.
17642 * config/arm/arm-cpu.h: Delete.
17643 * config/arm/arm-cpu-cdata.h: Delete.
17644 * config/arm/arm-cpu-data.h: Delete.
17646 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17648 * config/arm/arm-cpus.in (cortex-a55): New.
17649 (cortex-a75): Likewise.
17650 (cortex-a75.cortex-a55): Likewise.
17651 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17653 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17654 * config/arm/arm-cpu-cdata.h: Regenerate.
17655 * config/arm/arm-cpu-data.h: Regenerate.
17656 * config/arm/arm-cpu.h: Regenerate.
17657 * config/arm/arm-tables.opt: Regenerate.
17658 * config/arm/arm-tune.md: Regenerate.
17660 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17662 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17664 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17666 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17669 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17671 PR tree-optimization/81292
17672 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
17673 full_string_p, also call adjust_related_strinfos if the adjustment
17674 is simple, otherwise invalidate related strinfos.
17676 2017-07-04 Martin Liska <mliska@suse.cz>
17679 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
17680 newly created variable as DECL_IGNORED_P.
17682 2017-07-04 Martin Liska <mliska@suse.cz>
17685 * ipa-inline.c (inline_small_functions):
17686 Use xstrdup_for_dump.
17688 2017-07-04 Tom de Vries <tom@codesourcery.com>
17690 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17692 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17695 * config/darwin.c (darwin_function_switched_text_sections):
17698 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17700 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17702 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17704 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17706 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17708 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17709 min_profitable_iters, and th as inclusive lower bounds.
17710 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17711 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17712 for min_profitable_iters and min_profitable_estimate.
17713 (vect_transform_loop): Treat th as an inclusive lower bound.
17714 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17716 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17719 * config/darwin.c (darwin_function_switched_text_sections):
17720 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17721 in two pieces, and suppress the use of buf.
17723 2017-07-03 Nathan Sidwell <nathan@acm.org>
17725 * hash-table.h (hash_table_mod1): Fix indentation.
17727 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17729 PR middle-end/81290
17730 * predict.c (force_edge_cold): Be more careful about propagation
17732 * profile-count.h (profile_probability::guessed,
17733 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17735 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17737 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17739 * doc/invoke.texi (rcpc architecture extension): Document it.
17741 2017-07-03 Richard Biener <rguenther@suse.de>
17743 PR tree-optimization/60510
17744 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17745 the scalar reduction PHI and use it.
17746 (vectorizable_reduction): Properly guard the single_defuse_cycle
17747 path for non-SLP reduction chains where we cannot use it.
17748 Rework reduc_def/index and vector type deduction. Rework
17749 vector operand gathering during reduction op code-gen.
17750 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17751 chains dissolve the chain and leave it to non-SLP reduction
17754 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17756 * tree-data-ref.h (dr_alignment): Declare.
17757 * tree-data-ref.c (dr_alignment): New function.
17758 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17759 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17761 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17763 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17765 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17766 and base_misalignment fields.
17767 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17768 * tree-data-ref.c: Include builtins.h.
17769 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17770 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17771 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17772 * tree-vect-data-refs.c: Include tree-cfg.h.
17773 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17774 fields instead of calculating an alignment here.
17775 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17776 innermost_loop_behavior fields.
17778 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17780 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17782 (DR_STEP_ALIGNMENT): New macro.
17783 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17784 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17785 (create_data_ref): Print it.
17786 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17787 to tell whether the step preserves vector (mis)alignment.
17788 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17789 Move the check for an integer step and generalise to all INTEGER_CST.
17790 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17791 Print the outer step alignment.
17793 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17795 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17796 with offset_alignment.
17797 (DR_ALIGNED_TO): Delete.
17798 (DR_OFFSET_ALIGNMENT): New macro.
17799 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17800 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17801 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17802 (create_data_ref): Likewise.
17803 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17804 (vect_analyze_data_refs): Likewise.
17805 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17806 creating dummy innermost behavior.
17808 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17810 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17811 with a "innermost_loop_behavior *" and refeence tree.
17812 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17813 (create_data_ref): Update call accordingly.
17814 * tree-predcom.c (find_looparound_phi): Likewise.
17816 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17818 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17819 fields with dr_wrt_vec_loop.
17820 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17821 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17822 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17823 (vect_dr_behavior): New function.
17824 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17825 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17826 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17827 track whether the step preserves the misalignment.
17828 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17829 Use vect_dr_behavior.
17830 (vect_setup_realignment): Update call accordingly.
17831 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17832 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17833 call to vect_create_addr_base_for_vector_ref.
17834 (vect_create_cond_for_align_checks): Likewise.
17835 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17836 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17837 (vect_recog_mask_conversion_pattern): Likewise.
17838 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17839 (new_stmt_vec_info): Remove redundant zeroing.
17841 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17843 * common/config/arm/arm-common.c (arm_be8_option): New function.
17844 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17845 (ISA_ARMv6): Add isa_bit_be8.
17846 * config/arm/arm.h (arm_be8_option): Add prototype.
17847 (BE8_SPEC_FUNCTION): New define.
17848 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17849 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17850 (mlittle-endian): Similarly.
17851 (mbe8, mbe32): New options.
17852 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17853 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17855 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17857 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17859 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17861 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17862 (cleanup_tree_cfg_bb): Use it.
17863 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17865 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17867 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17870 * loop-doloop.c (add_test): Update profile.
17872 2017-07-03 Martin Liska <mliska@suse.cz>
17875 * sanopt.c (rewrite_usage_of_param): New function.
17876 (sanitize_rewrite_addressable_params): Likewise.
17877 (pass_sanopt::execute): Call rewrite_usage_of_param.
17879 2017-07-03 Richard Biener <rguenther@suse.de>
17881 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17882 back to using VIEW_CONVERT_EXPR.
17884 2017-07-03 Martin Liska <mliska@suse.cz>
17887 * doc/extend.texi: Document when a resolver function is
17888 generated for target_clones.
17890 2017-07-03 Martin Liska <mliska@suse.cz>
17892 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17893 and shadow memory of auto variables which are subject of
17894 use-after-scope sanitization.
17895 (asan_expand_mark_ifn): Add do set only when is_poison.
17897 2016-07-03 Richard Biener <rguenther@suse.de>
17899 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17901 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17902 (vectorizable_reduction): Perform reduction PHI creation when
17903 visiting a reduction PHI and adjust and simplify code generation
17904 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17905 (vect_transform_loop): Visit reduction PHIs.
17906 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17907 defs into the SLP tree.
17908 (vect_build_slp_tree): Reduction defs terminate the recursion.
17909 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17911 (vect_get_vec_defs_for_stmt_copy): Export.
17912 (vect_get_vec_defs): Likewise.
17913 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17915 (vect_get_vec_defs_for_stmt_copy): Declare.
17916 (vect_get_vec_defs): Likewise.
17918 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17920 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17921 parameter with a "loop" parameter and use it instead of the
17922 loop containing DR_STMT. Don't check simple_iv when doing
17923 BB analysis. Describe the two analysis modes in the comment.
17925 2017-07-03 Tom de Vries <tom@codesourcery.com>
17927 PR tree-optimization/69468
17928 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17929 (find_same_succ_bb): Handle ignore_edge_flags.
17931 2017-07-03 Tom de Vries <tom@codesourcery.com>
17933 PR tree-optimization/81192
17934 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17936 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17938 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17940 2017-07-03 Tom de Vries <tom@codesourcery.com>
17942 PR tree-optimization/81192
17943 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17944 BB_SAME_SUCC (bb) == NULL.
17946 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17948 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17951 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17953 * dumpfile.c: Include profile-count.h
17954 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17956 (insert_cond_bb): Update profile.
17957 * tree-cfg.h (insert_cond_bb): Update prototype.
17958 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17959 * tree-dump.c: Do not include tree-cfg.
17961 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17963 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17965 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17967 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17970 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17972 * tree-complex.c (expand_complex_div_wide): update profile.
17974 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17975 Alan Hayward <alan.hayward@arm.com>
17976 David Sherwood <david.sherwood@arm.com>
17978 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17979 (CORETYPES_H): New define.
17980 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17981 (insn-modes-inline.h, s-modes-inline-h): New rules.
17982 (generated_files): Add insn-modes-inline.h.
17983 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17984 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
17985 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
17986 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
17987 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
17988 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
17989 (build/gencodes.o, build/genconditions.o): Likewise.
17990 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
17991 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
17992 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
17993 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
17994 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
17995 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
17996 * coretypes.h: Include everything up to real.h for generators.
17997 Include insn-modes.h first. Include wide-int-print.h after
17998 wide-int.h. Include insn-modes-inline.h and then machmode.h.
17999 * machmode.h: Don't include insn-modes.h here.
18000 * function-tests.c: Remove includes of signop.h, machmode.h,
18001 double-int.h and wide-int.h.
18003 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18005 * optc-save-gen.awk: Likewise.
18006 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18007 * godump.c: Remove include of wide-int-print.h.
18008 * pretty-print.h: Likewise.
18009 * wide-int-print.cc: Likewise.
18010 * wide-int.cc: Likewise.
18011 * hash-map-tests.c: Remove include of signop.h.
18012 * hash-set-tests.c: Likewise.
18013 * rtl-tests.c: Likewise.
18014 * mkconfig.sh: Remove include of machmode.h.
18015 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18017 (emit_insn_modes_inline_h): ...this new function. Emit the code
18018 into an insn-modes-inline.h header file, adding appropriate
18019 include guards and end comments.
18020 (emit_insn_modes_c_header): Remove include of machmode.h.
18021 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18023 (main): Handle -i flag and call emit_insn_modes_inline_h when
18026 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18028 * tree-ssa-strlen.c (strinfo): Rename the length field to
18029 nonzero_chars. Add a full_string_p field.
18030 (compare_nonzero_chars, zero_length_string_p): New functions.
18031 (get_addr_stridx): Add an offset_out parameter.
18032 Use compare_nonzero_chars.
18033 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18034 (new_strinfo): Update after above changes to strinfo.
18035 (set_endptr_and_length): Set full_string_p.
18036 (get_string_length): Update after above changes to strinfo.
18037 (unshare_strinfo): Update call to new_strinfo.
18038 (maybe_invalidate): Likewise.
18039 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18040 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18041 as a uhwi instead of an shwi. Update after above changes to
18042 strinfo and new_strinfo.
18043 (zero_length_string): Assert that chainsi contains full strings.
18044 Use zero_length_string_p. Update call to new_strinfo.
18045 (adjust_related_strinfos): Update after above changes to strinfo.
18046 Copy full_string_p from origsi.
18047 (adjust_last_stmt): Use zero_length_string_p.
18048 (handle_builtin_strlen): Update after above changes to strinfo and
18049 new_strinfo. Install the lhs as the string length if the previous
18050 entry didn't describe a full string.
18051 (handle_builtin_strchr): Update after above changes to strinfo
18053 (handle_builtin_strcpy): Likewise.
18054 (handle_builtin_strcat): Likewise.
18055 (handle_builtin_malloc): Likewise.
18056 (handle_pointer_plus): Likewise.
18057 (handle_builtin_memcpy): Likewise. Track nonzero characters
18058 that aren't necessarily followed by a nul terminator.
18059 (handle_char_store): Likewise.
18061 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18063 PR tree-optimization/80769
18064 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18065 for malloc and calloc. Document the new invariant that all related
18066 strinfos have delayed lengths or none do.
18067 (verify_related_strinfos): Move earlier in file.
18068 (set_endptr_and_length): New function, split out from...
18069 (get_string_length): ...here. Also set the lengths of related
18071 (zero_length_string): Assert that chainsi has known (rather than
18073 (adjust_related_strinfos): Likewise.
18075 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18077 PR tree-optimization/81136
18078 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18079 assert that two references with the same misalignment have the same
18080 compile-time misalignment if those compile-time misalignments
18083 2017-07-01 Andi Kleen <ak@linux.intel.com>
18085 * print-tree.c (print_node): Print all attributes.
18087 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18089 * cfg.c (scale_bbs_frequencies): New function.
18090 * cfg.h (scale_bbs_frequencies): Declare it.
18091 * cfgloopanal.c (single_likely_exit): Cleanup.
18092 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18094 (scale_loop_profile): Likewise.
18095 (loop_version): Likewise.
18096 (create_empty_loop_on_edge): Update.
18097 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18098 scale_loop_frequencies, scale_loop_profile, loopify,
18099 loop_version): Update prototypes.
18100 * modulo-sched.c (sms_schedule): Update.
18101 * predict.c (unlikely_executed_edge_p): Also check probability.
18102 (probably_never_executed_edge_p): Fix typo.
18103 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18104 * tree-parloops.c (gen_parallel_loop): Update.
18105 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18106 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18107 * tree-ssa-loop-split.c (split_loop): Update.
18108 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18109 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18110 (vect_loop_versioning): Update.
18111 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18113 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18115 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18117 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18119 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18120 to keep profile consistent.
18122 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18124 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18125 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18126 * profile-count.h (max_safe_multiplier): Make unsigned.
18127 (profile_count::guessed_zero): New.
18129 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18131 * bb-reorder.c (fix_up_crossing_landing_pad,
18132 fix_crossing_conditional_branches): Use make_single_succ_edge
18133 to keep profile consistent.
18135 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18137 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18140 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18143 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18144 the right scopes, make sure cond_jump isn't preserved between multiple
18145 iterations. Search for fallthru edge whenever there are 3+ edges and
18146 use find_fallthru_edge for it.
18148 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18150 Patch by Alexander Monakov <amonakov@ispras.ru>
18151 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18152 probabilities consistently.
18154 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18156 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18157 emit_cmp_and_jump_insns.
18159 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18162 * tree-inline.c (expand_call_inline): Combine profile statuses.
18164 2017-06-30 Andrew Pinski <apinski@cavium.com>
18166 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18167 fold_stmt returned true.
18169 2017-06-30 Nathan Sidwell <nathan@acm.org>
18171 * ggc.h (empty_string): Delete.
18172 * cfgexpand.c (expand_asm_stmt): Use plain "".
18173 * optabs.c (expand_asm_memory_barrier): Likewise.
18174 * stringpool.c (empty_string): Delete.
18175 (digit_vector, digit_string): Delete.
18176 (ggc_alloc_string): Use plain "", don't optimize single digit
18177 strings. Use ggc_alloc_atomic.
18179 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
18181 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
18182 comparison set and one other set, use the cost of the non-comparison
18185 2017-06-30 Nathan Sidwell <nathan@acm.org>
18187 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
18190 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
18192 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
18193 loops. Remove now unneeded calls to gimple_switch_set_label() that
18194 just set removed labels to NULL_TREE.
18196 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
18198 * tree-ssanames.c (set_range_info_raw): Abstract from ...
18199 (set_range_info): ...here. Only call set_range_info_raw if domain
18201 (set_nonzero_bits): Call set_range_info_raw.
18202 * tree-ssanames.h (set_range_info_raw): New.
18204 2017-06-30 Jakub Jelinek <jakub@redhat.com>
18207 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
18208 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
18209 of nonimmediate_operand and <store_mask_constraint> instead of m
18210 for the input operand. For V8FI iterator, always split if input
18211 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
18212 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
18213 <store_mask_predicate> instead of register_operand and
18214 <store_mask_constraint> instead of v for the input operand. Make
18215 sure both operands aren't MEMs for if not <mask_applied>.
18217 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
18219 * lto-wrapper.c (copy_file) Close both file descriptors before
18222 2017-06-30 Martin Liska <mliska@suse.cz>
18225 * multiple_target.c (create_dispatcher_calls): Make ifunc
18226 also for function that don't have calls or are not referenced.
18228 2017-06-30 Richard Biener <rguenther@suse.de>
18230 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
18231 analyze the first scalar stmt. Move vector type computation
18232 for the BB case here from ...
18233 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
18234 live operation processing in the SLP case properly.
18236 2017-06-30 Richard Biener <rguenther@suse.de>
18238 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18240 2017-06-30 Martin Liska <mliska@suse.cz>
18243 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18244 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18246 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18248 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18251 2017-06-30 Richard Biener <rguenther@suse.de>
18253 PR tree-optimization/81249
18254 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18255 condition reduction result to original scalar type.
18257 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18259 * profile-count.h (enum profile_quality): Fix typos and whitespace
18262 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18264 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18265 type for branch probabilities.
18267 2017-06-29 Julian Brown <julian@codesourcery.com>
18268 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18270 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18271 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18272 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18273 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18275 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18277 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18278 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18279 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18280 CC usage from generic code to here.
18281 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18282 CC usage into the target macros.
18284 2017-06-29 Maya Rashish <coypu@sdf.org>
18286 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18289 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18291 * arm/arm-builtins.c: Include profile-count.h
18292 * except.c (sjlj_emit_function_enter): Use
18293 profile_probability::unlikely.
18295 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18297 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18298 and tocrel_offset be pointer args rather than implicitly using
18300 (legitimate_constant_pool_address_p, rs6000_emit_move,
18301 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18302 tocrel_offset and use in toc_relative_expr_p call.
18303 (print_operand, print_operand_address): Use static tocrel_base_oac
18304 and tocrel_offset_oac.
18305 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18308 2017-06-29 Maya Rashish <coypu@sdf.org>
18310 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18312 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18314 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18315 objects, take into account only the alignment of 'op0' and 'mode1' if
18318 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18320 * ccmp.c (ccmp_tree_comparison_p): New function.
18321 (ccmp_candidate_p): Update to use above function.
18322 (get_compare_parts): New function.
18323 (expand_ccmp_next): Update to use new functions.
18324 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18326 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18327 take mode as argument.
18328 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18329 * expr.c (expand_expr_real_1): Pass mode as argument.
18331 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18333 * combine.c (combine_instructions): Print insns to dump_file, together
18336 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18338 * asan.c (asan_emit_stack_protection): Update.
18339 (create_cond_insert_point): Update.
18340 * auto-profile.c (afdo_propagate_circuit): Update.
18341 * basic-block.h (struct edge_def): Turn probability to
18342 profile_probability.
18343 (EDGE_FREQUENCY): Update.
18344 * bb-reorder.c (find_traces_1_round): Update.
18345 (better_edge_p): Update.
18346 (sanitize_hot_paths): Update.
18347 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18348 (make_single_succ_edge): Update.
18349 (check_bb_profile): Update.
18350 (dump_edge_info): Update.
18351 (update_bb_profile_for_threading): Update.
18352 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18353 probabilitycount to 0.
18354 * cfgbuild.c (compute_outgoing_frequencies): Update.
18355 * cfgcleanup.c (try_forward_edges): Update.
18356 (outgoing_edges_match): Update.
18357 (try_crossjump_to_edge): Update.
18358 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18359 (expand_gimple_tailcall): Update.
18360 (construct_init_block): Use make_single_succ_edge.
18361 (construct_exit_block): Use make_single_succ_edge.
18362 * cfghooks.c (verify_flow_info): Update.
18363 (redirect_edge_succ_nodup): Update.
18364 (split_edge): Update.
18365 (account_profile_record): Update.
18366 * cfgloopanal.c (single_likely_exit): Update.
18367 * cfgloopmanip.c (scale_loop_profile): Update.
18368 (set_zero_probability): Remove.
18369 (duplicate_loop_to_header_edge): Update.
18370 * cfgloopmanip.h (loop_version): Update prototype.
18371 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18372 (force_nonfallthru_and_redirect): Update.
18373 (update_br_prob_note): Update.
18374 (rtl_verify_edges): Update.
18375 (purge_dead_edges): Update.
18376 (rtl_lv_add_condition_to_bb): Update.
18377 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18378 * cgraphunit.c (init_lowered_empty_function): Update.
18379 (cgraph_node::expand_thunk): Update.
18380 * cilk-common.c: Include profile-count.h
18381 * dojump.c (inv): Remove.
18382 (jumpifnot): Update.
18383 (jumpifnot_1): Update.
18384 (do_jump_1): Update.
18386 (do_jump_by_parts_greater_rtx): Update.
18387 (do_compare_rtx_and_jump): Update.
18388 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18389 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18390 * dwarf2cfi.c: Include profile-count.h
18391 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18392 (sjlj_emit_dispatch_table): Likewise.
18393 * explow.c: Include profile-count.h
18394 * expmed.c (emit_store_flag_force): Update.
18395 (do_cmp_and_jump): Update.
18396 * expr.c (compare_by_pieces_d::generate): Update.
18397 (compare_by_pieces_d::finish_mode): Update.
18398 (emit_block_move_via_loop): Update.
18399 (store_expr_with_bounds): Update.
18400 (store_constructor): Update.
18401 (expand_expr_real_2): Update.
18402 (expand_expr_real_1): Update.
18403 * expr.h (try_casesi, try_tablejump): Update prototypes.
18404 * gimple-pretty-print.c (dump_probability): Update.
18405 (dump_profile): New.
18406 (dump_gimple_label): Update.
18407 (dump_gimple_bb_header): Update.
18408 * graph.c (draw_cfg_node_succ_edges): Update.
18409 * hsa-gen.c (convert_switch_statements): Update.
18410 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18411 (find_if_case_1): Update.
18412 (find_if_case_2): Update.
18413 * internal-fn.c (expand_arith_overflow_result_store): Update.
18414 (expand_addsub_overflow): Update.
18415 (expand_neg_overflow): Update.
18416 (expand_mul_overflow): Update.
18417 (expand_vector_ubsan_overflow): Update.
18418 * ipa-cp.c (good_cloning_opportunity_p): Update.
18419 * ipa-split.c (split_function): Use make_single_succ_edge.
18420 * ipa-utils.c (ipa_merge_profiles): Update.
18421 * loop-doloop.c (add_test): Update.
18422 (doloop_modify): Update.
18423 * loop-unroll.c (compare_and_jump_seq): Update.
18424 (unroll_loop_runtime_iterations): Update.
18425 * lra-constraints.c (lra_inheritance): Update.
18426 * lto-streamer-in.c (input_cfg): Update.
18427 * lto-streamer-out.c (output_cfg): Update.
18428 * mcf.c (adjust_cfg_counts): Update.
18429 * modulo-sched.c (sms_schedule): Update.
18430 * omp-expand.c (expand_omp_for_init_counts): Update.
18431 (extract_omp_for_update_vars): Update.
18432 (expand_omp_ordered_sink): Update.
18433 (expand_omp_for_ordered_loops): Update.
18434 (expand_omp_for_generic): Update.
18435 (expand_omp_for_static_nochunk): Update.
18436 (expand_omp_for_static_chunk): Update.
18437 (expand_cilk_for): Update.
18438 (expand_omp_simd): Update.
18439 (expand_omp_taskloop_for_outer): Update.
18440 (expand_omp_taskloop_for_inner): Update.
18441 * omp-simd-clone.c (simd_clone_adjust): Update.
18442 * optabs.c (expand_doubleword_shift): Update.
18443 (expand_abs): Update.
18444 (emit_cmp_and_jump_insn_1): Update.
18445 (expand_compare_and_swap_loop): Update.
18446 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18447 * predict.c (predictable_edge_p): Update.
18448 (edge_probability_reliable_p): Update.
18449 (set_even_probabilities): Update.
18450 (combine_predictions_for_insn): Update.
18451 (combine_predictions_for_bb): Update.
18452 (propagate_freq): Update.
18453 (estimate_bb_frequencies): Update.
18454 (force_edge_cold): Update.
18455 * profile-count.c (profile_count::dump): Add missing space into dump.
18456 (profile_count::debug): Add newline.
18457 (profile_count::differs_from_p): Explicitly convert to unsigned.
18458 (profile_count::stream_in): Update.
18459 (profile_probability::dump): New member function.
18460 (profile_probability::debug): New member function.
18461 (profile_probability::differs_from_p): New member function.
18462 (profile_probability::differs_lot_from_p): New member function.
18463 (profile_probability::stream_in): New member function.
18464 (profile_probability::stream_out): New member function.
18465 * profile-count.h (profile_count_quality): Rename to ...
18466 (profile_quality): ... this one.
18467 (profile_probability): New.
18468 (profile_count): Update.
18469 * profile.c (compute_branch_probabilities): Update.
18470 * recog.c (peep2_attempt): Update.
18471 * sched-ebb.c (schedule_ebbs): Update.
18472 * sched-rgn.c (find_single_block_region): Update.
18473 (compute_dom_prob_ps): Update.
18474 (schedule_region): Update.
18475 * sel-sched-ir.c (compute_succs_info): Update.
18476 * stmt.c (struct case_node): Update.
18477 (do_jump_if_equal): Update.
18478 (get_outgoing_edge_probs): Update.
18479 (conditional_probability): Update.
18480 (emit_case_dispatch_table): Update.
18481 (expand_case): Update.
18482 (expand_sjlj_dispatch_table): Update.
18483 (emit_case_nodes): Update.
18484 * targhooks.c: Update.
18485 * tracer.c (better_p): Update.
18486 (find_best_successor): Update.
18487 * trans-mem.c (expand_transaction): Update.
18488 * tree-call-cdce.c: Update.
18489 * tree-cfg.c (gimple_split_edge): Upate.
18490 (move_sese_region_to_fn): Upate.
18491 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18492 * tree-eh.c (lower_resx): Upate.
18493 (cleanup_empty_eh_move_lp): Upate.
18494 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18495 * tree-inline.c (copy_edges_for_bb): Update.
18496 (copy_cfg_body): Update.
18497 * tree-parloops.c (gen_parallel_loop): Update.
18498 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18499 (gimple_gen_time_profiler): Update.
18500 * tree-ssa-dce.c (remove_dead_stmt): Update.
18501 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18502 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18503 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18504 (unloop_loops): Update.
18505 (try_peel_loop): Update.
18506 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18507 * tree-ssa-loop-split.c (connect_loops): Update.
18508 (split_loop): Update.
18509 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18510 (hoist_guard): Update.
18511 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18512 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18513 (value_replacement): Update.
18514 * tree-ssa-reassoc.c (branch_fixup): Update.
18515 * tree-ssa-tail-merge.c (replace_block_by): Update.
18516 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18517 (create_edge_and_update_destination_phis): Update.
18518 (compute_path_counts): Update.
18519 (recompute_probabilities): Update.
18520 (update_joiner_offpath_counts): Update.
18521 (freqs_to_counts_path): Update.
18522 (duplicate_thread_path): Update.
18523 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18524 (struct switch_conv_info): Update.
18525 (gen_inbound_check): Update.
18526 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18527 (vect_do_peeling): Update.
18528 (vect_loop_versioning): Update.
18529 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18530 (optimize_mask_stores): Update.
18531 * ubsan.c (ubsan_expand_null_ifn): Update.
18532 * value-prof.c (gimple_divmod_fixed_value): Update.
18533 (gimple_divmod_fixed_value_transform): Update.
18534 (gimple_mod_pow2): Update.
18535 (gimple_mod_pow2_value_transform): Update.
18536 (gimple_mod_subtract): Update.
18537 (gimple_mod_subtract_transform): Update.
18538 (gimple_ic): Update.
18539 (gimple_stringop_fixed_value): Update.
18540 (gimple_stringops_transform): Update.
18541 * value-prof.h: Update.
18543 2017-06-29 Carl Love <cel@us.ibm.com>
18545 * config/rs6000/rs6000-c.c: Add support for built-in functions
18546 vector signed int vec_signed (vector float);
18547 vector signed long long vec_signed (vector double);
18548 vector signed int vec_signed2 (vector double, vector double);
18549 vector signed int vec_signede (vector double);
18550 vector signed int vec_signedo (vector double);
18551 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18552 instruction generator.
18553 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18554 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18555 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18557 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18558 vunsignede_v2df): Add define_expands.
18559 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18560 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18561 VEC_UNSIGNEDO): Add definitions.
18562 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18563 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18564 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18565 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18566 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18567 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18568 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18569 vec_unsignede, vec_unsignedo): Add builtin defines.
18570 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18572 * doc/extend.texi: Update the built-in documentation file for the
18573 new built-in functions.
18575 2017-06-29 Richard Biener <rguenther@suse.de>
18577 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18578 reduction chains to LOOP_VINFO_REDUCTIONS.
18579 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18580 SLP reductions after processing reduction chains.
18582 2017-06-29 Nathan Sidwell <nathan@acm.org>
18584 * builtins.c (fold_builtin_FUNCTION): Use
18585 lang_hooks.decl_printable_name.
18587 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18589 PR middle-end/81194
18590 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18591 with only one label.
18592 * stmt.c (expand_case): Assert NCASES is greater than one.
18594 2017-06-29 Richard Biener <rguenther@suse.de>
18596 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18598 (group_case_labels): Likewise.
18599 (find_taken_edge): Push sanity checking on val to workers...
18600 (find_taken_edge_cond_expr): ... here
18601 (find_taken_edge_switch_expr): ... and here, handle cases
18602 with just a default label.
18603 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18604 (group_case_labels): Likewise.
18605 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18606 group_case_labels does anything cleanup the CFG again.
18608 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18610 PR tree-optimization/81196
18611 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18612 exit condition comparing two IVs.
18614 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18616 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18617 profile to the dummy entry at the end of the list of architectures.
18618 * config/arm/arm-cpu-cdata.h: Regenerated.
18620 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18621 Michael Collison <michael.collison@arm.com>
18624 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18626 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18627 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18628 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18629 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18630 of shift when the shift amount is masked with constant equal to
18631 the size of the mode.
18632 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18635 2017-06-29 Martin Liska <mliska@suse.cz>
18637 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18639 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18642 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18645 (edge_growth_cache_entry::edge_growth_cache_entry): New
18647 (reset_edge_growth_cache): Update to use constructor.
18649 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18651 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18652 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18653 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18655 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18657 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18658 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18660 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18662 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18663 (*-linux-uclibc*): Add t-uclibc tmake_file.
18664 * config/t-musl: New.
18665 * config/t-uclibc: New.
18667 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18669 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18671 (gen_comm_data): Emit architectural setting of arch_prof.
18672 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
18674 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
18675 (armv8-m.base, armv8-m.main): Likewise.
18676 * arm-protos.h (arm_build_target): Add profile field.
18677 (arch_option): Likewise.
18678 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
18680 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
18681 arm_active_target.profile.
18683 2017-06-28 Richard Biener <rguenther@suse.de>
18685 PR middle-end/81227
18686 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
18687 TYPE_OVERFLOW_WRAPS.
18688 * match.pd (negate_expr_p): Likewise.
18689 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18690 fold_build2, not fold_binary.
18692 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18694 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18695 Convert memory address to Pmode.
18696 (aarch64_print_operand): Assert MEM operands are always Pmode.
18698 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18701 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18702 Remove redundant if.
18703 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18704 * config/arm/aarch-common-protos.h
18705 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18706 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18708 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18711 * multiple_target.c (create_dispatcher_calls): Set the default
18712 clone to be static, not public.
18714 2017-06-28 Richard Biener <rguenther@suse.de>
18716 * tree-vect-loop.c (vectorizable_reduction): Move special
18717 cond reduction IV var creation ...
18718 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18719 parameter. Use STMT_VINFO_VECTYPE.
18720 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18723 2017-06-28 Martin Liska <mliska@suse.cz>
18726 * ipa-visibility.c (non_local_p): Handle visibility.
18728 2017-06-28 Martin Liska <mliska@suse.cz>
18731 * common.opt: Add IntegerRange to various options.
18732 * opt-functions.awk (integer_range_info): New function.
18733 * optc-gen.awk: Add integer_range_info to cl_options struct.
18734 * opts-common.c (decode_cmdline_option): Handle
18735 CL_ERR_INT_RANGE_ARG.
18736 (cmdline_handle_error): Likewise.
18737 * opts.c (print_filtered_help): Show valid interval in
18738 when --help is provided.
18739 * opts.h (struct cl_option): Add range_min and range_max fields.
18740 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18742 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18744 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18745 (x * C EQ/NE y * C): New transformation.
18747 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18749 * genmultilib (combination_space): Accept '+' in option names.
18751 2017-06-28 Martin Liska <mliska@suse.cz>
18754 * asan.c (instrument_derefs): Bail out inner references
18755 that are hard register variables.
18757 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18760 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18761 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18763 2017-06-28 Richard Biener <rguenther@suse.de>
18765 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18766 (vect_get_slp_defs): Adjust.
18767 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18769 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18771 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18772 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18773 (vectorizable_reduction): Adjust.
18774 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18776 (vect_get_slp_defs): Likewise.
18777 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18778 (vectorizable_bswap): Adjust.
18779 (vectorizable_call): Likewise.
18780 (vectorizable_conversion): Likewise.
18781 (vectorizable_assignment): Likewise.
18782 (vectorizable_shift): Likewise.
18783 (vectorizable_operation): Likewise.
18784 (vectorizable_store): Likewise.
18785 (vectorizable_condition): Likewise.
18786 (vectorizable_comparison): Likewise.
18788 2017-06-28 Michael Collison <michael.collison@arm.com>
18791 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18793 (arm_gen_movmemqi): Removed unused variable 'i'.
18794 Convert 'for' loop into 'while' loop.
18795 (arm_expand_prologue): Remove last unnecessary set of insn.
18796 (thumb_pop): Remove unused variable 'pushed_words'.
18797 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18799 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18801 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18802 * config/s390/s390-protos.h: Add prototype of
18803 s390_rel_address_ok_p.
18804 * config/s390/s390.c (s390_got_symbol): New function.
18805 (s390_rel_address_ok_p): New function.
18806 (legitimize_pic_address): Use s390_rel_address_ok_p.
18807 (s390_load_got): Use s390_got_symbol.
18808 (s390_option_override): Issue error if
18809 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18810 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18812 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18814 2017-06-27 Andrew Pinski <apinski@cavium.com>
18816 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18817 (X * copysign (1.0, X)): New pattern.
18818 (X * copysign (1.0, -X)): New pattern.
18819 (copysign (-1.0, CST)): New pattern.
18821 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18823 * genmultilib (combination_space): Remove variable.
18824 Validate reuse rules against regular expression for any sequence
18825 of multilib options in any order.
18827 2017-06-27 Michael Collison <michael.collison@arm.com>
18829 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18830 call aarch64_split_simd_combine.
18831 * (aarch64_combine_internal<mode>): Delete pattern.
18832 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18833 Allow register and subreg operands.
18835 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18837 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18838 specific need, just fallback on defaults.
18839 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18841 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18842 Olivier Hainque <hainque@adacore.com>
18844 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18846 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18847 targets. Pick a default if no particular attempt applied.
18848 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18851 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18853 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18854 (x86_64-wrs-vxworks7): Likewise.
18856 2017-06-27 Marek Polacek <polacek@redhat.com>
18859 * ubsan.c (instrument_null): Check get_base_address's result for null.
18861 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18863 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18865 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18867 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18868 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18869 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18870 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18871 New function types.
18872 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18873 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18874 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18875 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18876 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18877 BUILT_IN_FEUPDATEENV): New builtins.
18878 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18879 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18880 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18881 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18883 (builtin_structptr_types): Adjust size.
18884 * tree.c (builtin_structptr_types): Add four entries.
18886 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18887 Olivier Hainque <hainque@adacore.com>
18889 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18890 (TLS_SYM): New local macro, forcing reference to __tls__ on
18891 link command lines for VxWorks 7 RTPs, triggering initialization
18893 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18894 OS features TLS support, true for RTPs on VxWorks 7.
18895 * config/vxworks.c (vxworks_override_options): Setup emutls
18898 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18900 * predict.c (test_prediction_value_range): Use -1U instead of -1
18901 to avoid narrowing conversion warning.
18902 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18903 to avoid narrowing conversion warning.
18904 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18906 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18908 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18910 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18911 64bit configurations.
18912 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18913 (SIZE_TYPE): Likewise.
18914 * config/vxworks.c (vxworks_emutls_var_fields): Use
18915 long_unsigned_type_node instead of unsigned_type_node as the offset
18916 field type, which is "pointer" mode in emutls.c.
18918 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18921 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18923 PR middle-end/81207
18924 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18925 gimple_vuse copying separately from gimple_vdef copying.
18927 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18929 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18932 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18933 Olivier Hainque <hainque@adacore.com>
18935 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18936 for all vxworks7 targets.
18937 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18938 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18939 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18940 variations for VX6/VX7 and 32/64bits later on in ...
18941 (VXWORKS_LIB_SPEC): Leverage new macros.
18942 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18943 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18945 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18947 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18948 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18950 2017-06-26 Carl Love <cel@us.ibm.com>
18952 * config/rs6000/rs6000-c.c: Add support for built-in functions
18953 vector bool char vec_reve (vector bool char);
18954 vector signed char vec_reve (vector signed char);
18955 vector unsigned char vec_reve (vector unsigned char);
18956 vector bool int vec_reve (vector bool int);
18957 vector signed int vec_reve (vector signed int);
18958 vector unsigned int vec_reve (vector unsigned int);
18959 vector bool long long vec_reve (vector bool long long);
18960 vector signed long long vec_reve (vector signed long long);
18961 vector unsigned long long vec_reve (vector unsigned long long);
18962 vector bool short vec_reve (vector bool short);
18963 vector signed short vec_reve (vector signed short);
18964 vector double vec_reve (vector double);
18965 vector float vec_reve (vector float);
18966 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18967 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18968 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18969 (altivec_vreve): New pattern.
18970 * config/rs6000/altivec.h (vec_reve): New define.
18971 * doc/extend.texi (vec_rev): Update the built-in documentation file
18972 for the new built-in functions.
18974 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18976 PR tree-optimization/71815
18977 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18979 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18981 (slsr_process_phi): Likewise.
18982 (replace_uncond_cands_and_profitable_phis): Don't replace a
18983 multiply candidate with a stride of 1 (copy or cast).
18984 (phi_incr_cost): Call uses_consumed_by_stmt rather than
18986 (lowest_cost_path): Likewise.
18987 (total_savings): Likewise.
18989 2017-06-26 Richard Biener <rguenther@suse.de>
18992 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
18993 Use def_builtin_pure for all gather builtins.
18995 2017-06-26 Richard Biener <rguenther@suse.de>
18997 PR tree-optimization/81203
18998 * tree-tailcall.c (find_tail_calls): Do not move stmts into
18999 non-dominating BBs.
19001 2017-06-26 Marek Polacek <polacek@redhat.com>
19004 * doc/invoke.texi: Document -Wmultistatement-macros.
19006 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19008 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19009 arm_neon_ok_no_float_abi effective target.
19011 2017-06-26 Richard Biener <rguenther@suse.de>
19013 PR tree-optimization/80928
19014 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19015 (copy_bbs): Set BB_DUPLICATED flag early.
19016 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19018 (execute_on_shrinking_pred): Likewise.
19019 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19020 BB_DUPLICATED blocks.
19021 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19022 iterate over all PHIs considering removal of *gsi.
19024 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19026 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19029 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19031 * config/rs6000/rs6000-string.c: (expand_block_clear,
19032 do_load_for_compare, select_block_compare_mode,
19033 compute_current_alignment, expand_block_compare,
19034 expand_strncmp_align_check, expand_strn_compare,
19035 expand_block_move, rs6000_output_load_multiple)
19036 Move functions related to string/block move/compare
19037 to a separate file.
19038 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19039 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19040 for this function which is now used in two files.
19041 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19042 * config.gcc: Add rs6000-string.o to extra_objs for
19043 targets powerpc*-*-* and rs6000*-*-*.
19045 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19048 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19049 32-bit, since indexed is not valid for DImode.
19050 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19051 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19052 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19053 (define_peephole2 for Altivec d-form store): Likewise.
19056 * multiple_target.c (create_dispatcher_calls): Only create the
19057 dispatcher call if the function is the default clone of a
19058 versioned function.
19060 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19062 PR middle-end/80902
19063 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19064 a call, force the call to not be a tail call.
19066 2017-06-23 Jeff Law <law@redhat.com>
19068 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19071 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19073 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19074 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19075 handling for early expansion of vector shifts (sl,sr,sra,rl).
19076 (builtin_function_type): Add vector shift right instructions
19077 to the unsigned argument list.
19079 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19081 rtl-optimizatoin/79286
19082 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19083 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19084 trap. PIC register plus a const unspec without offset can never trap.
19086 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19088 * tree.h (builtin_structptr_type): New type.
19089 (builtin_structptr_types): Declare new array.
19090 * tree.c (builtin_structptr_types): New array.
19091 (free_lang_data, build_common_tree_nodes): Use it.
19093 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19096 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19099 2017-06-22 Matt Turner <mattst88@gmail.com>
19101 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19102 Lake models to skylake case. Assume skylake for unknown
19103 models with clflushopt.
19105 2017-06-22 Jeff Law <law@redhat.com>
19107 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19108 frame sizes that do not satisfy aarch64_uimm12_shift.
19110 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19112 * profile-count.h (apply_probability,
19113 apply_scale, probability_in): Fix checks for zero.
19115 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19117 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19118 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19120 2016-06-22 Richard Biener <rguenther@suse.de>
19122 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19123 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19124 REDUC_MAX_EXPR support.
19125 (vectorizable_reduction): Likewise.
19126 (vect_create_epilog_for_reduction): Likewise.
19128 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19130 * match.pd (A / (1 << B) -> A >> B): New.
19131 * generic-match-head.c: Include optabs-tree.h.
19132 * gimple-match-head.c: Likewise.
19133 * optabs-tree.h (target_supports_op_p): New.
19134 * optabs-tree.c (target_supports_op_p): New.
19136 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19138 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19139 $gcc_cv_ld --help output.
19140 (gcc_cv_ld_demangle): Likewise.
19141 (gcc_cv_ld_eh_frame_hdr): Likewise.
19142 (gcc_cv_ld_pie): Likewise.
19143 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19144 (gcc_cv_ld_buildid): Likewise.
19145 (gcc_cv_ld_sysroot): Likewise.
19146 (ld_bndplt_support): Likewise.
19147 (ld_pushpopstate_support): Likewise.
19148 * configure: Regenerate.
19149 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19151 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19154 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19155 operands indexes to avoid gap between operands and match_dups.
19157 2017-06-21 Andrew Pinski <apinski@cavium.com>
19159 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19160 Increment Arith_shift and Arith_shift_reg by 1.
19161 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19163 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19164 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19165 (aarch64_strip_extend): Add new argument and test for it.
19166 (aarch64_cheap_mult_shift_p): New function.
19167 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19168 add a cost if it is true.
19169 Update calls to aarch64_strip_extend.
19170 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19172 2017-06-21 Andrew Pinski <apinski@cavium.com>
19174 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
19176 (thunderxt88): Likewise.
19177 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
19178 (thunderx_prefetch_tune): New variable.
19179 (thunderx2t99_prefetch_tune): Update for the correct values.
19180 (thunderxt88_tunings): New variable.
19181 (thunderx_tunings): Use thunderx_prefetch_tune instead of
19182 generic_prefetch_tune.
19183 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
19185 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19187 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
19188 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
19189 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
19190 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
19191 (aarch64_atomic_cas<mode>, GPI): Likewise.
19193 2017-06-21 Martin Liska <mliska@suse.cz>
19195 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
19196 statements on cold and hot labels.
19197 * predict.c (tree_estimate_probability_bb): Remove the
19198 prediction from this place.
19200 2017-06-21 Martin Liska <mliska@suse.cz>
19202 PR tree-optimization/79489
19203 * gimplify.c (maybe_add_early_return_predict_stmt): New
19205 (gimplify_return_expr): Call the function.
19206 * predict.c (tree_estimate_probability_bb): Remove handling
19208 * predict.def: Update comment about early return predictor.
19209 * gimple-predict.h (is_gimple_predict): New function.
19210 * predict.def: Change default value of early return to 66.
19211 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
19213 * passes.def: Put pass_strip_predict_hints to the beginning of
19216 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
19218 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
19219 FUNCTION_DECL declarations.
19220 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
19222 (dwaf2out_decl): Likewise.
19223 * godump.c (go_early_global_decl): Skip call to the real debug hook
19224 for FUNCTION_DECL declarations.
19225 * passes.c (rest_of_decl_compilation): Skip call to the
19226 early_global_decl debug hook for FUNCTION_DECL declarations, unless
19227 -fdump-go-spec is passed.
19229 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19231 * config/i386/i386.c (struct builtin_isa): New field pure_p.
19232 Reorder for compactness.
19233 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
19234 (def_builtin_pure, def_builtin_pure2): New functions.
19235 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19237 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19239 * match.pd (nop_convert): New predicate.
19240 ((A +- CST1) +- CST2): Allow some NOP conversions.
19242 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19245 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19246 with ctors/dtors if GOVD_SHARED is set.
19248 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19250 * config/aarch64/aarch64.md (movti_aarch64):
19251 Emit mov rather than orr.
19252 (movtf_aarch64): Likewise.
19253 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19254 Emit mov rather than orr.
19256 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19258 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19259 Swap alternatives, make integer dup more expensive.
19261 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19263 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19264 Return true for non-tls symbols.
19266 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19268 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19269 (cortex-a75): Likewise.
19270 (cortex-a75.cortex-a55): Likewise.
19271 * config/aarch64/aarch64-tune.md: Regenerate.
19272 * doc/invoke.texi (-mtune): Document new values for -mtune.
19274 2017-06-21 Tom de Vries <tom@codesourcery.com>
19276 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19277 stack_size feature.
19278 (Effective-Target Keywords, Other attributes): Suggest using
19279 dg-add-options stack_size feature to get stack limit in stack_size
19280 effective target documentation.
19282 2017-06-21 Julian Brown <julian@codesourcery.com>
19283 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19285 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19286 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19287 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19289 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19290 attribute type list for neon_multiply.
19291 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19292 attribute type list for neon_multiply.
19293 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19294 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19295 attribute type list for neon_multiply.
19296 * config/arm/types.md (crypto_pmull): Add.
19297 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19298 attribute type list.
19300 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19302 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19305 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19307 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19308 to make sure not to dereference a NULL cost_classes_ptr pointer.
19310 2017-06-20 Carl Love <cel@us.ibm.com>
19312 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19313 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19314 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19315 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19316 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19317 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19318 VMULOSW): New enum "unspec" values.
19319 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19320 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19321 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19322 altivec_vmulosw): New patterns.
19323 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19324 VMULOSW): Add definitions.
19326 2017-06-20 Julia Koval <julia.koval@intel.com>
19328 * config/i386/i386.c: Fix rounding expand for new pattern.
19329 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19331 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19333 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19334 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19336 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19338 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19341 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19343 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19344 architecture, then by implementer ID.
19345 * config/aarch64/aarch64-tune.md: Regenerate.
19347 2017-06-20 Richard Biener <rguenther@suse.de>
19349 PR middle-end/81097
19350 * fold-const.c (split_tree): Fold to type before negating.
19352 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19354 * diagnostic-show-locus.c
19355 (selftest::test_fixit_deletion_affecting_newline): New function.
19356 (selftest::diagnostic_show_locus_c_tests): Call it.
19358 2017-06-20 Andreas Schwab <schwab@suse.de>
19361 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19364 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19366 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19367 __ARM_FEATURE_COPROC according to support.
19369 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19371 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19372 Rewritten to avoid overflow for > 32-bit pointers.
19375 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19376 by removing enum keyword.
19377 (ubsan_type_descriptor): Likewise. Formatting fix.
19380 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19381 splitter): Require TARGET_SSE2 in the condition.
19383 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19386 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19387 for doing vector set of SFmode on ISA 3.0.
19388 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19389 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19391 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19392 SFmode value into a V4SF variable that was extracted from another
19393 V4SF variable without converting the element to double precision
19394 and back to single precision vector format.
19395 (vsx_insert_extract_v4sf_p9_2): Likewise.
19397 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19399 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19400 in UWHI to avoid undefined overflow.
19403 * ubsan.h (enum ubsan_encode_value_phase): New.
19404 (ubsan_encode_value): Change second argument to
19405 enum ubsan_encode_value_phase with default value of
19406 UBSAN_ENCODE_VALUE_GENERIC.
19407 * ubsan.c (ubsan_encode_value): Change second argument to
19408 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19409 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19410 create_tmp_var_raw instead of create_tmp_var and use a
19412 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19413 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19414 ubsan_encode_value callers.
19417 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19418 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19419 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19421 2017-06-19 Richard Biener <rguenther@suse.de>
19423 PR middle-end/81118
19424 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19425 estimates if we changed anything.
19427 2017-06-19 Richard Biener <rguenther@suse.de>
19429 PR tree-optimization/80887
19430 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19431 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19432 simplified lookups, then reset mprts_hook.
19433 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19435 (try_to_simplify): Likewise.
19437 2017-06-19 Martin Liska <mliska@suse.cz>
19440 * gimplify.c (gimplify_switch_expr):
19441 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19443 2017-06-19 Martin Liska <mliska@suse.cz>
19445 * doc/install.texi: Document that PGO runs in 4 stages.
19447 2017-06-19 Martin Liska <mliska@suse.cz>
19450 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19451 to dispatcher function name.
19452 * multiple_target.c (replace_function_decl): New function.
19453 (create_dispatcher_calls): Redirect both edges and references.
19455 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19457 * profile-count.c (profile_count::dump): Dump quality.
19458 (profile_count::differs_from_p): Update for unsigned val.
19459 * profile-count.h (profile_count_quality): New enum.
19460 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19462 2017-06-19 Richard Biener <rguenther@suse.de>
19464 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19465 struct function as arg.
19466 (estimate_numbers_of_iterations): Export overload with loop arg.
19467 (free_numbers_of_iterations_estimates_loop): Use an overload of
19468 free_numbers_of_iterations_estimates instead.
19469 * tree-cfg.c (remove_bb): Adjust.
19470 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19471 * tree-parloops.c (gen_parallel_loop): Likewise.
19472 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19474 (tree_unroll_loops_completely): Likewise.
19475 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19476 Use an overload instead and export.
19477 (estimated_loop_iterations): Adjust.
19478 (max_loop_iterations): Likewise.
19479 (likely_max_loop_iterations): Likewise.
19480 (estimate_numbers_of_iterations): Take struct function as arg
19482 (loop_exits_before_overflow): Adjust.
19483 (free_numbers_of_iterations_estimates_loop): Use an overload.
19484 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19485 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19487 2017-06-19 Richard Biener <rguenther@suse.de>
19490 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19491 RANGE_EXPR conservatively.
19493 2017-06-16 Carl Love <cel@us.ibm.com>
19495 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19496 definitions for vec_float, vec_float2, vec_floato,
19497 vec_floate built-ins.
19498 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19499 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19501 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19502 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19503 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19504 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19505 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19506 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19507 vec_floato): Add builtin defines.
19508 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19509 Update the built-in documentation file for the new built-in
19512 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19514 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19515 (mthumb): Mark as the negative of -marm.
19517 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19519 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19521 (ARM Options, -mtune): Document that this accepts the same
19522 extension options as -mcpu.
19523 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19525 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19527 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19528 permitted extensions.
19530 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19532 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19533 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19534 (armv8-m.main): Add option +nodsp.
19535 * config/arm/arm-cpu-cdata.h: Regenerated.
19537 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19539 * config/arm/t-fuchsia: New file.
19540 * config.gcc (arm*-*-fuchsia*): Use it.
19542 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19544 * config/arm/t-symbian: Rewrite for new option infrastructure.
19546 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19548 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19549 (MULTILIB_REQUIRED): Likewise.
19551 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19553 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19554 (MULTILIB_RESUE): Likewise.
19555 (MULTILIB_MATCHES): Likewise.
19556 (MULTLIB_REQUIRED): Likewise.
19558 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19560 * config/arm/t-rtems: Rewrite for new option framework.
19562 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19564 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19565 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19566 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19567 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19568 * config/arm/t-multilib: ... here.
19569 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19570 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19571 armv7-a and armv8*-a when A-profile libraries have not been built.
19572 * config/arm/t-rmprofile: Rewrite.
19574 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19576 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19577 with a backslash. Remove the backslash after substituting unescaped
19579 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19581 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19583 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19584 the driver to pass through the default setting of -mfloat-abi.
19585 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19587 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19588 (all_feat_combs): New rule.
19589 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19591 * config/arm/t-aprofile: Rewrite.
19593 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19595 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19596 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19597 fpu is not specified by the user/command-line.
19598 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19599 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19600 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19601 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19602 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19603 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19605 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19607 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19608 * config/arm/t-arm-elf: Rewritten.
19610 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19612 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19613 have some floating-point instructions.
19614 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19615 (TARGET_MAYBE_HARD_FLOAT): New macro.
19616 * config/arm/arm-builtins.c (arm_init_builtins): Use
19617 TARGET_MAYBE_HARD_FLOAT.
19618 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19620 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19622 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19623 (configargs.h): Include it.
19624 (arm_print_hint_for_fpu_option): New function.
19625 (arm_parse_fpu_option): New function.
19626 (candidate_extension): New class.
19627 (arm_canon_for_multilib): New function.
19628 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19629 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19630 (ARCH_CANONICAL_SPECS): New macro.
19631 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19633 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19635 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19636 are set after handling multilib fragments. Set target_cpu_default2
19639 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19641 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19643 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19644 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19646 (check_arch): Likewise.
19647 * config/arm/arm.c (arm_configure_build_target): Handle
19648 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19649 options in the default.
19651 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19653 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19654 when an option is an alias of another.
19655 * config/arm/parsecpu.awk (optalias): New parser token.
19656 (gen_comm_data): Mark non-alias options as such. Emit entries
19657 for extension aliases.
19658 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19659 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19660 (armv6kz, armv6zk, armv6t2): Likewise.
19661 (armv7): Make vfpv3-d16 an alias.
19662 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19664 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19665 Sort in canonical order.
19666 (armv8-a): Sort in canonical order.
19667 (armv8.1-a, armv8.2-a): Likewise.
19668 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19670 (cortex-a9): Sort in canonical order.
19671 * config/arm/arm.c (selftests.h): Include it.
19672 (arm_test_cpu_arch_data): New function.
19673 (arm_run_self_tests): New function.
19674 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
19675 (targetm): Move declaration to the end of the file.
19676 * arm-cpu-cdata.h: Regenerated.
19678 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19680 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
19681 call to target_mode_check describing the type of option passed.
19682 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
19683 (arm_target_thumb_only): Use arm_parse_arch_option_name or
19684 arm_parse_cpu_option_name to match parameters against list of
19686 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
19687 arm_arch_core_flags data structure.
19688 * config/arm/arm-cpu_cdata.h: Regenerated.
19690 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19692 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19694 (arm_print_hint_for_cpu_option): Likewise.
19695 (arm_print_hint_for_arch_option): Likewise.
19696 (arm_parse_cpu_option_name): Likewise.
19697 (arm_parse_arch_option_name): Likewise.
19698 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19699 of entries in the all_fpus list.
19700 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19701 (arm_parse_cpu_option_name): Declare.
19702 (arm_parse_arch_option_name): Declare.
19703 (arm_parse_option_features): Declare.
19704 (arm_intialize_isa): Declare.
19705 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19707 (gen_comm_data): ... here. Make definitions non-static.
19708 * config/arm/arm-cpu-data.h: Regenerated.
19709 * config/arm/arm-cpu-cdata.h: Regenerated.
19711 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19713 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19714 (cpu_arch_extension): New structure.
19715 (cpu_arch_option, arch_option, cpu_option): New structures.
19716 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19717 architecture types.
19718 (gen_data): Generate new format data tables.
19719 * config/arm/arm.c (cpu_tune): New structure.
19720 (cpu_option, processors): Delete.
19721 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19722 (arm_print_hint_for_cpu_option): ... this and ...
19723 (arm_print_hint_for_arch_option): ... this.
19724 (arm_parse_arch_cpu_name): Delete. Replace with ...
19725 (arm_parse_cpu_option_name): ... this and ...
19726 (arm_parse_arch_option_name): ... this.
19727 (arm_unrecognized_feature): Change type of target parameter to
19729 (arm_parse_arch_cpu_features): Delete. Replace with ...
19730 (arm_parse_option_features): ... this.
19731 (arm_configure_build_target): Rework to use new configuration data
19733 (arm_print_tune_info): Rework for new configuration data tables.
19734 * config/arm/arm-cpu-data.h: Regenerated.
19735 * config/arm/arm-cpu.h: Regenerated.
19737 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19739 * Makefile.in (OBJS): Move sbitmap.o from here ...
19740 (OBJS-libcommon): ... to here.
19742 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19744 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19745 (ISA_ALL_CRYPTO): New macro.
19746 (ISA_ALL_SIMD): New macro
19747 (ISA_ALL_FP): New macro.
19748 * config/arm/arm.c (fpu_bitlist): Update initializer.
19749 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19751 (arm9e): Add fpu. Add option for nofp
19752 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19753 (arm926ej-s, arm1026ej-s): Likewise.
19754 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19755 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19756 neon-fp16, neon-vfpv4, nofp and nosimd.
19757 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19758 (cortex-a8): Add fpu. Add option for nofp.
19759 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19760 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19761 (cortex-r4f): Add fpu.
19762 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19763 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19765 (cortex-r8): Likewise.
19766 (cortex-m4): Add fpu. Add option for nofp.
19767 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19768 (cortex-a17.cortex-a7): Likewise.
19769 (cortex-a32): Add fpu. Add options for crypto and nofp.
19770 (cortex-a35, cortex-a53): Likewise.
19771 (cortex-a57): Add fpu. Add option for crypto.
19772 (cortex-a72, cortex-a73): Likewise.
19773 (exynos-m1): Likewise.
19774 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19775 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19776 (cortex-m33): Add fpu. Add option for nofp.
19777 * config/arm/arm-cpu-cdata.h: Regenerated
19778 * config/arm/arm-cpu-data.h: Regenerated.
19780 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19782 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19783 (armv5te, armv5tej): Likewise.
19784 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19785 (armv7): Add options fp and vfpv3-d16.
19786 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19787 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19789 (armv7ve): Likewise.
19790 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19791 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19792 (armv8-a): Add nocrypto option.
19793 (armv8.1-a, armv8.2-a): Likewise.
19794 (armv8-m.main): add options fp, fp.dp and nofp.
19796 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19798 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19800 (armv8-a+crc): Delete.
19801 (armv8.1-a): Add options simd, crypto and nofp.
19802 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19803 (armv8.2-a+fp16): Delete.
19804 (armv8-m.main): Add option dsp.
19805 (armv8-m.main+dsp): Delete.
19806 (cortex-a8): Add fpu. Add nofp option.
19807 (cortex-a9): Add fpu. Add nofp and nosimd options.
19808 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19809 link to main cpu and architecture data structures.
19810 (gen_comm_data): Only put isa attributes from the main architecture
19812 (option): New statement for architecture and CPU entries.
19813 * arm.c (struct cpu_option): New structure.
19814 (struct processors): Add entry for options.
19815 (arm_unrecognized_feature): New function.
19816 (arm_parse_arch_cpu_name): Ignore any characters after the first
19818 (arm_parse_arch_cpu_feature): New function.
19819 (arm_configure_build_target): Separate out any CPU and architecture
19820 features and parse separately. Don't error out if -mfpu=auto is
19821 used with only an architecture string.
19822 (arm_print_asm_arch_directives): New function.
19823 (arm_file_start): Call it.
19824 * config/arm/arm-cpu-cdata.h: Regenerated.
19825 * config/arm/arm-cpu-data.h: Likewise.
19826 * config/arm/arm-tables.opt: Likewise.
19828 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19830 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19831 assembler when it is not -mfpu=auto.
19833 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19835 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19836 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19837 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19838 (ASM_CPU_SPEC): Rewrite.
19839 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19840 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19841 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19842 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19843 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19844 copied string is NUL-terminated. Also strip any characters prefixed
19846 (arm_rewrite_selected_arch): New function.
19847 (arm_rewrite_march): New function.
19849 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19851 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19852 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19853 (march, mcpu, mtune): Convert to string-based options.
19854 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19855 (arm_parse_arch_cpu_name): New function.
19856 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19857 identify selected architecture or CPU.
19858 (arm_option_save): New function.
19859 (TARGET_OPTION_SAVE): Redefine.
19860 (arm_option_restore): Restore string options.
19861 (arm_option_print): Print string options.
19863 2017-06-16 Martin Sebor <msebor@redhat.com>
19865 PR tree-optimization/80933
19866 PR tree-optimization/80934
19867 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19868 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19869 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19870 (gimple_fold_builtin): Call them.
19872 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19874 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19875 as unlikely; update profile.
19877 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19879 * predict.c (force_edge_cold): Handle declaring edges impossible
19882 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19884 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19886 (try_unroll_loop_completely): Fix reporting.
19888 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19890 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19892 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19895 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19896 if given a non-constant argument for an intrinsic which requires a
19899 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19901 * profile.c (compare_freqs): New function.
19902 (branch_prob): Sort edge list.
19903 (find_spanning_tree): Assume that the list is priority sorted.
19905 2017-06-16 Richard Biener <rguenther@suse.de>
19907 PR tree-optimization/81090
19908 * passes.def (pass_record_bounds): Remove.
19909 * tree-pass.h (make_pass_record_bounds): Likewise.
19910 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19911 make_pass_record_bounds): Likewise.
19912 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19913 not free niter estimates at the beginning but at the end.
19914 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19916 2017-06-16 Richard Biener <rguenther@suse.de>
19918 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19919 initializer to workaround ICE in host GCC 4.8.
19921 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19923 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19925 (clone_inlined_nodes): Update.
19927 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19929 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19930 prefetch settings, and enable prefetching by default at -O3.
19932 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19934 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19935 Set flag_prefetch_loop_arrays according to tuning data.
19937 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19939 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19940 New tune structure.
19941 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19942 [Unrelated to main purpose of the patch] Place the pointer field last
19943 to enable type checking errors when tune structure are wrongly merged.
19944 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19945 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19946 (thunderx2t99_prefetch_tune): New tune constants.
19947 (tune_params *_tunings): Update all tunings (no functional change).
19948 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19949 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19950 from tunings structures.
19952 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19955 * ubsan.c (instrument_null): Add T argument, use it instead
19956 of computing it based on IS_LHS.
19957 (instrument_object_size): Likewise.
19958 (pass_ubsan::execute): Adjust instrument_null and
19959 instrument_object_size callers to pass gimple_get_lhs or
19960 gimple_assign_rhs1 result to it. Use instrument_null instead of
19961 calling get_base_address and instrument_mem_ref. Handle
19962 aggregate call arguments for object-size sanitization.
19964 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19966 PR tree-optimization/81089
19967 * tree-vrp.c (is_masked_range_test): Validate operands of
19970 2017-06-15 Martin Sebor <msebor@redhat.com>
19973 * dumpfile.c (dump_register): Avoid calling memset to initialize
19974 a class with a default ctor.
19975 * gcc.c (struct compiler): Remove const qualification.
19976 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19977 * hash-table.h: Ditto.
19978 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19980 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19981 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19983 * params.h (struct param_info): Make struct members non-const.
19984 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
19985 with default initialization.
19986 * vec.h (vec_copy_construct, vec_default_construct): New helper
19988 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
19989 with vec_copy_construct.
19990 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
19991 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
19992 * doc/invoke.texi (-Wclass-memaccess): Document.
19994 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19996 * emit-rtl.h (is_leaf): Update comment about local
19997 register allocator.
19999 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20002 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20003 for a variable to have a section before checking if the section has a
20005 Set section to.persistent if persistent attribute is set.
20006 Warn if .persistent attribute is used on an automatic variable.
20008 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20010 PR rtl-optimization/80474
20011 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20013 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20015 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20018 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20020 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20021 (casesi): Emit bounds checking as RTL.
20022 (casesi_internal_mips16_<mode>): Remove bounds checking.
20024 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20026 * config/xtensa/xtensa.c (xtensa_option_override): Append
20027 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20028 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20029 xtensa_doloop_hooks): Define unconditionally.
20030 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20032 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20033 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20034 for it in xtensa_option_override.
20035 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20036 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20038 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20040 * doc/cppopts.texi: Document '-' special value to -MF.
20042 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20044 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20045 (cortex_a53_fconst): Likewise.
20046 (cortex_a53_fpmul): Likewise.
20047 (cortex_a53_f_load_64): Likewise.
20048 (cortex_a53_f_load_many): Likewise.
20049 (cortex_a53_advsimd_alu): Likewise.
20050 (cortex_a53_advsimd_alu_q): Likewise.
20051 (cortex_a53_advsimd_mul): Likewise.
20052 (cortex_a53_advsimd_mul_q): Likewise.
20053 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20054 Add missing fmul, r2f_cvt and fconst cases.
20056 2017-06-14 Richard Biener <rguenther@suse.de>
20058 PR middle-end/81088
20059 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20061 (fold_binary_loc): When associating do not treat pre-existing
20062 TREE_OVERFLOW on literal constants as a reason to allow
20063 TREE_OVERFLOW on associated literal constants.
20065 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20067 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20068 (MASK_FEATURES): New macro.
20069 * config/sparc/sparc.c (sparc_option_override): Remove the special
20070 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20072 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20074 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20075 a division of 0 if non-call exceptions are enabled.
20077 2017-06-14 Andrew Pinski <apinski@cavium.com>
20078 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20081 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20082 Improve vector initialization code gen for only variable case.
20084 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20086 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20088 2017-06-14 Richard Biener <rguenther@suse.de>
20090 PR tree-optimization/81083
20091 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20094 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20096 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20097 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20098 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20099 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20100 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20101 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20103 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20105 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20106 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20108 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20110 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20112 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20114 * config/rs6000/t-rtems: Don't handle SPE.
20116 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20118 * config/rs6000/t-linux: Don't handle SPE.
20120 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20122 * config/rs6000/eabispe.h: Delete file.
20124 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20126 * config/rs6000/t-spe: Delete file.
20128 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20130 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20131 (rs6000_legitimate_offset_address_p): Return false for anything in
20132 V2SImode or V2SFmode.
20134 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20136 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20137 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20138 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20140 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20141 (rs6000_legitimate_offset_address_p): Ditto.
20142 (rs6000_emit_move): Ditto.
20143 (rs6000_init_builtins): Remove V4HI_type_node.
20145 2017-06-13 Martin Liska <mliska@suse.cz>
20148 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20149 (gate_asan): Likewise.
20150 * asan.h (asan_no_sanitize_address_p): Remove the function.
20151 (sanitize_flags_p): New function.
20152 * builtins.def: Fix coding style.
20153 * common.opt: Use renamed enum value.
20154 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20155 * doc/extend.texi: Document no_sanitize attribute.
20156 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20157 to SANITIZE_UNDEFINED_NONDEFAULT.
20158 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20159 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20160 Use sanitize_flags_p.
20161 * gimplify.c (gimplify_function_tree): Likewise.
20162 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20163 * opts.c (parse_no_sanitize_attribute): New function.
20164 (common_handle_option): Use renamed enum value.
20165 * opts.h (parse_no_sanitize_attribute): Declare.
20166 * tree.c (sanitize_flags_p): New function.
20167 * tree.h: Declared here.
20168 * tsan.c: Use sanitize_flags_p.
20169 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20170 (instrument_mem_ref): Likewise.
20171 (instrument_bool_enum_load): Likewise.
20172 (do_ubsan_in_current_function): Remove the function.
20173 (pass_ubsan::execute): Use sanitize_flags_p.
20174 * ubsan.h: Remove do_ubsan_in_current_function
20175 * tree-cfg.c (print_no_sanitize_attr_value): New function.
20176 (dump_function_to_file): Use it here.
20178 2017-06-13 Martin Jambor <mjambor@suse.cz>
20180 PR tree-optimization/80803
20181 PR tree-optimization/81063
20182 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
20183 (propagate_subaccesses_across_link): Enqueue subtree whenever
20184 necessary instead of relying on the caller.
20186 2017-06-13 Martin Jambor <mjambor@suse.cz>
20188 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
20189 that have a first_link.
20190 (sort_and_splice_var_accesses): Do not check first_link before
20192 (subtree_mark_written_and_enqueue): Likewise.
20193 (propagate_all_subaccesses): Likewise and do not stop at first
20194 parent with a first_link.
20196 2017-06-13 Martin Jambor <mjambor@suse.cz>
20198 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
20201 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20203 * match.pd: New pattern.
20205 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20207 * tree-vrp.c (is_masked_range_test): New function.
20208 (register_edge_assert_for): Determine ranges for
20211 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20213 PR tree-optimization/67328
20214 * fold-const.c (maskable_range_p): New function.
20215 (build_range_check): Generate bittests if possible.
20217 2017-06-13 Martin Liska <mliska@suse.cz>
20219 * gimple-pretty-print.c (dump_probability): Add new argument.
20220 (dump_edge_probability): Dump both probability and count.
20221 (dump_gimple_label): Likewise.
20222 (dump_gimple_bb_header): Likewise.
20224 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
20227 * config/avr/avr-devices.c: Fix indentation.
20228 * config/avr/gen-avr-mmcu-specs.c: Dito.
20230 2017-06-13 Richard Biener <rguenther@suse.de>
20232 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
20233 instead get vector type from stmt_info.
20234 (vectorizable_reduction): Adjust. Remove dead code.
20236 2017-06-13 Richard Biener <rguenther@suse.de>
20238 PR middle-end/81065
20239 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20240 case of C * (x * C2 + C3).
20241 (fold_addr_of_array_ref_difference): Properly fold index difference.
20243 2017-06-12 David S. Miller <davem@davemloft.net>
20246 * config/sparc/sparc.md (return expander): Emit frame blockage if
20247 function uses alloca.
20249 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20251 * combine.c (make_field_assignment): Check len rather than the mode
20252 precision when calling force_to_mode.
20254 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20256 Support multilibs and devices that see flash in RAM address range.
20259 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20260 (avr_mcu_t) <flash_pm_offset>: New field.
20261 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20262 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20263 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20264 (AVR_TINY_PM_OFFSET): Remove macro.
20265 * config/avr/avr.opt (-mshort-calls): New option.
20266 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20267 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20268 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20269 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20270 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20271 instead of avr_arch->have_jmp_call.
20272 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20273 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20274 avr_arch->flash_pm_offset to define.
20275 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20276 new field flash_pm_offset. Add entry for avrxmega3.
20277 (avr_texinfo): Add entry for avrxmega3.
20278 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20279 attiny212, attiny214,
20280 attiny412, attiny414, attiny416, attiny417,
20281 attiny814, attiny816, attiny817,
20282 attiny1614, attiny1616, attiny1617,
20283 attiny3214, attiny3216, attiny3217.
20284 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20285 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20286 (avr_print_operand_address) [AVR_TINY]: Same.
20287 (avr_asm_init_sections) <readonly_data_section>: Only patch
20288 callback if avr_arch->flash_pm_offset = 0.
20289 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20290 for rodata if avr_arch->flash_pm_offset != 0.
20291 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20292 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20293 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20294 (m_options): Append opt_rcall.
20295 (m_dirnames): Append dir_rcall.
20296 * config/avr/t-multilib: Regenerate.
20298 * configure.ac [target=avr]: Check whether avrxmega3 default
20299 linker description file works as needed.
20300 * configure: Regenerate.
20301 * doc/avr-mmcu.texi: Regenerate.
20302 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20303 <__AVR_ARCH__>: Document avrxmega3 and 103.
20304 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20305 <__AVR_SHORT_CALLS__>: Document it.
20306 <__AVR_PM_BASE_ADDRESS__>: Document it.
20307 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20308 (AVR Variable Attributes) <progmem>: Document this is
20309 not needed for avrxmega3.
20310 (AVR Named Address Spaces) <__flash>: Dito.
20312 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20314 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20316 2017-06-12 Doug Rupp <rupp@adacore.com>
20318 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20319 Append vxworks-stdint.h to the tm_file list.
20320 * config/vxworks-stdint.h: New file.
20322 2017-06-12 Martin Liska <mliska@suse.cz>
20324 PR tree-optimization/81041
20325 * tree-profile.c (gimple_gen_ic_func_profiler):
20326 Create an extra BB in profile-generate
20327 (gimple_gen_time_profiler): Likewise.
20329 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20331 PR tree-optimization/81003
20332 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20333 (update_range_test): Use it instead of force_gimple_operand_gsi.
20335 2017-06-12 Richard Biener <rguenther@suse.de>
20337 PR tree-optimization/81053
20338 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20339 with backedge value not defined in loop. Simplify def stmt
20342 2017-06-11 Tom de Vries <tom@codesourcery.com>
20345 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20347 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20348 nvptx_cannot_force_const_mem.
20350 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20352 * opts.c (finish_options): Move test for flag_split_stack after
20353 it has been initialized.
20355 2017-06-11 Jason Merrill <jason@redhat.com>
20357 * tree.h (id_equal): New.
20358 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20359 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20360 instead of strcmp of IDENTIFIER_POINTER.
20362 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20364 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20365 (mark_all_inlined_calls_cdtor): Fix formating.
20366 (inline_transform): Rescale profile before inlining.
20368 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20370 * cgraph.h (cgraph_edge::clone): Update prototype.
20371 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20372 (cgraph_node::create_clone): Update.
20373 (cgraph_node::create_version_clone): Update.
20374 * tree-inline.c (copy_bb): Update.
20375 (expand_call_inline): Update.
20377 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20379 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20380 factored out from ...
20381 (rs6000_emit_prologue): ... here.
20383 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20385 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20386 factored out from ...
20387 (rs6000_emit_prologue): ... here.
20389 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20391 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20393 (handle_missing_profiles): Fix computation of tp_first_run.
20394 (counts_to_freqs): Do not touch freqs when count is 0.
20396 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20398 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20401 2017-06-10 Tom de Vries <tom@codesourcery.com>
20403 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20404 attributes): Document signal effective target.
20406 2017-06-10 Tom de Vries <tom@codesourcery.com>
20408 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20409 Document effective target stack_size.
20411 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20413 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20414 to the edit_context if they can be auto-applied.
20416 2017-06-9 Ian Lance Taylor <iant@golang.org>
20418 * opts.c (finish_options): If -fsplit-stack, disable implicit
20419 -forder-blocks-and-partition.
20420 * doc/invoke.texi (Optimize Options): Document that when using
20421 -fsplit-stack -forder-blocks-and-partition is not implicitly
20424 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20426 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20427 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20428 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20429 * builtins.def (abort, trap, unreachable): Declare cold.
20430 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20431 * tree-core.h (ECF_COLD): New.
20432 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20433 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20435 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20437 * predict.c (unlikely_executed_stmt_p): Cleanup.
20439 2017-06-09 Richard Biener <rguenther@suse.de>
20441 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20442 model if the ref is always written to.
20444 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20446 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20448 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20450 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20453 2017-06-09 Tom de Vries <tom@codesourcery.com>
20456 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20457 "target cannot support label values" when encountering LABEL_REF.
20459 2017-06-09 Martin Liska <mliska@suse.cz>
20461 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20462 (gimple_gen_ic_func_profiler): Emit direct comparison
20463 of __gcov_indirect_call_callee with NULL.
20464 (gimple_gen_time_profiler): Change probability from
20465 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20467 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20469 * profile.c (edge_gcov_counts): Turn to pointer.
20470 (compute_branch_probabilities, compute_branch_probabilities): Update.
20471 (branch_prob): Do not clear edge_gcov_count.
20472 * profile.h (edge_gcov_counts): Turn to pointer.
20473 (edge_gcov_count): Update.
20475 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20477 * gimple.h (gimple_check_failed): Mark cold.
20479 2017-06-09 Richard Biener <rguenther@suse.de>
20481 PR tree-optimization/66623
20482 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20483 refactor check_reduction into two parts, properly computing
20484 whether we have to check reduction validity for outer loop
20487 2017-06-09 Richard Biener <rguenther@suse.de>
20489 PR tree-optimization/79483
20490 * graphite-scop-detection.c (order): New global.
20491 (get_order): Compute bb to order mapping that satisfies code
20492 generation constraints.
20493 (cmp_pbbs): New helper.
20494 (build_scops): Start domwalk at entry block, sort generated
20497 2017-06-09 Richard Biener <rguenther@suse.de>
20499 PR middle-end/81007
20500 * ipa-polymorphic-call.c
20501 (ipa_polymorphic_call_context::restrict_to_inner_class):
20502 Skip FIELD_DECLs with error_mark_node type.
20503 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20506 2017-06-09 Martin Liska <mliska@suse.cz>
20508 * predict.c (struct branch_predictor): New struct.
20509 (test_prediction_value_range): New test.
20510 (predict_c_tests): New function.
20511 * selftest-run-tests.c (selftest::run_tests): Run the function.
20512 * selftest.h: Declare new tests.
20514 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20517 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20518 gen_add3_insn did not fail.
20519 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20520 r0, construct that number in a temporary reg and add that reg to r0.
20521 If asked to put the result in r0 as well, fail.
20523 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20525 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20526 for early expansion of vec_eqv.
20528 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20530 PR middle-end/81005
20531 * ubsan.c (instrument_null): Avoid pointless code temporary.
20532 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20534 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20538 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20541 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20544 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20546 * predict.c (unlikely_executed_edge_p): Move ahead.
20547 (probably_never_executed_edge_p): Use it.
20549 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20551 PR middle-end/79988
20552 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20553 gimple_call_builtin_p call.
20555 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20557 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20558 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20559 rtl_check_failed_type2, rtl_check_failed_code1,
20560 rtl_check_failed_code2, rtl_check_failed_code_mode,
20561 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20562 rtvec_check_failed_bounds, rtl_check_failed_flag,
20563 _fatal_insn_not_found, _fatal_insn): Likewise.
20564 * tree.h (tree_contains_struct_check_failed,
20565 tree_check_failed, tree_not_check_failed,
20566 tree_class_check_failed, tree_range_check_failed,
20567 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20568 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20569 tree_operand_check_failed, omp_clause_check_failed,
20570 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20573 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20575 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20576 flag_branch_probabilities.
20577 * ipa-inline.c (edge_badness): Likewise.
20578 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20579 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20580 * predict.c (maybe_hot_frequency_p): Likewise.
20581 (probably_never_executed): Likewise.
20582 * sched-ebb.c (schedule_ebbs): Likewise.
20583 * sched-rgn.c (find_single_block_region): Likewise.
20584 * tracer.c (tail_duplicate): Likewise.
20586 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20588 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20589 longer requires x_flag_profile_use.
20591 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20593 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20594 instead of flag_reorder_blocks_and_partition.
20595 * dbxout.c (dbxout_function_end): Likewise.
20596 * dwarf2out.c (gen_subprogram_die): Likewise.
20597 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20598 * hw-doloop.c (reorg_loops): Likewise.
20599 * varasm.c (assemble_start_function,
20600 assemble_end_function): Likewise.
20601 (decide_function_section): Do not check for
20602 flag_reorder_blocks_and_partition.
20604 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20606 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20608 (chkp_get_hard_register_fake_addr_expr): Ditto.
20609 (chkp_build_addr_expr): Add check for hard reg case.
20610 (chkp_parse_array_and_component_ref): Ditto.
20611 (chkp_find_bounds_1): Ditto.
20612 (chkp_process_stmt): Don't generate bounds store for
20615 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20617 * predict.c (maybe_hot_bb_p): Do not check profile status.
20618 (maybe_hot_edge_p): Likewise.
20619 (probably_never_executed): Check for zero counts even if profile
20621 (unlikely_executed_edge_p): New function.
20622 (unlikely_executed_stmt_p): New function.
20623 (unlikely_executed_bb_p): New function.
20624 (set_even_probabilities): Use unlikely predicates.
20625 (combine_predictions_for_bb): Likewise.
20626 (predict_paths_for_bb): Likewise.
20627 (predict_paths_leading_to_edge): Likewise.
20628 (determine_unlikely_bbs): New function.
20629 (estimate_bb_frequencies): Use it.
20630 (compute_function_frequency): Use zero counts even if profile is
20632 * profile-count.h: Fix typo.
20634 2017-08-08 Julia Koval <julia.koval@intel.com>
20636 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20637 _mm512_mask_cvtsepi16_storeu_epi8,
20638 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20639 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20640 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20641 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20642 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20643 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20644 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20645 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20646 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20647 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20648 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20649 __builtin_ia32_pmovuswb256mem_mask,
20650 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20651 __builtin_ia32_pmovwb512mem_mask): New builtins.
20653 2017-08-08 Julia Koval <julia.koval@intel.com>
20655 PR target/73350,80862
20656 * config/i386/subst.md (round): Fix round pattern.
20657 * config/i386/i386.c (ix86_erase_embedded_rounding):
20658 Fix erasing rounding for the fixed pattern.
20660 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20662 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20664 2017-06-08 Martin Liska <mliska@suse.cz>
20666 PR gcov-profile/80911
20667 * gcov.c (block_info::block_info): New constructor.
20669 2017-06-07 Carl Love <cel@us.ibm.com>
20671 * config/rs6000/rs6000-c: The return type of the following
20672 built-in functions was implemented as int not long long. Fix sign
20673 of return value for the unsigned version of vec_mulo and vec_mule.
20674 vector unsigned long long vec_bperm (vector unsigned long long,
20675 vector unsigned char)
20676 vector signed long long vec_mule (vector signed int,
20678 vector unsigned long long vec_mule (vector unsigned int,
20679 vector unsigned int)
20680 vector signed long long vec_mulo (vector signed int,
20682 vector unsigned long long vec_mulo (vector unsigned int,
20683 vector unsigned int)
20684 * doc/extend.texi: Fix the documentation for the built-in
20687 2017-06-07 Carl Love <cel@us.ibm.com>
20690 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20693 2017-06-07 Carl Love <cel@us.ibm.com>
20695 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20696 support, Generate doublehv for signed int/float for BE case only.
20698 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20700 * doc/invoke.texi (mcx16): Rewrite.
20702 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20704 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20705 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20706 *mov<mode>_softfloat, and an anonymous splitter): Use
20707 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20709 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20711 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20713 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20714 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20715 (rs6000_debug_reg_global): Adjust.
20716 (rs6000_init_hard_regno_mode_ok): Adjust.
20717 (rs6000_dbx_register_number): Adjust.
20718 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20719 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20720 Remove SPE_ACC and SPEFSCR.
20721 (REG_ALLOC_ORDER): Ditto.
20722 (FRAME_POINTER_REGNUM): Change to 111.
20723 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20724 (REG_CLASS_NAMES): Ditto.
20725 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20726 (REGISTER_NAMES): Ditto.
20727 (ADDITIONAL_REG_NAMES): Ditto.
20728 (rs6000_reg_names): Ditto.
20729 * config/rs6000/rs6000.md: Renumber some register number
20732 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20734 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20736 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20737 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20739 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20740 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20741 Delete the SPE high registers.
20742 (REG_ALLOC_ORDER): Ditto.
20743 (enum reg_class): Remove SPE_HIGH_REGS.
20744 (REG_CLASS_NAMES): Ditto.
20745 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20746 (REGISTER_NAMES): Ditto.
20747 (rs6000_reg_names): Ditto.
20748 * doc/tm.texi.in: Remove SPE as example.
20749 * doc/tm.texi: Regenerate.
20751 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20753 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20754 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20755 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20756 * config/rs6000/rs6000.md (type): Remove "brinc".
20758 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20760 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20761 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20762 * config/rs6000/linuxspe.h: Delete file.
20763 * config/rs6000/rs6000.md: Don't include spe.md.
20764 * config/rs6000/spe.h: Delete file.
20765 * config/rs6000/spe.md: Delete file.
20766 * config/rs6000/t-rs6000: Remove spe.md.
20768 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20770 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20771 (reg_or_none500mem_operand): Delete.
20772 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20773 instead of reg_or_none500mem_operand.
20775 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20777 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20778 handling of SPE flags.
20779 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20781 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20783 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20785 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20786 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20787 paired_divv2sf3): Similar.
20788 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20789 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20790 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20792 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20793 Rename the paired_* instruction patterns.
20794 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20796 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20797 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20798 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20799 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20800 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20801 PAIRED_VECTOR_MODE.
20802 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20803 (spe_func_has_64bit_regs_p): Delete.
20804 (spe_expand_predicate_builtin): Delete.
20805 (spe_expand_evsel_builtin): Delete.
20806 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20807 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20808 (invalid_e500_subreg): Delete.
20809 (rs6000_legitimize_address): Always force_reg op2 as well, for
20810 paired single memory accesses.
20811 (rs6000_member_type_forces_blk): Delete.
20812 (rs6000_spe_function_arg): Delete.
20813 (rs6000_expand_unop_builtin): Delete SPE handling.
20814 (rs6000_expand_binop_builtin): Ditto.
20815 (spe_expand_stv_builtin): Delete.
20816 (bdesc_2arg_spe): Delete.
20817 (spe_expand_builtin): Delete.
20818 (spe_expand_predicate_builtin): Delete.
20819 (spe_expand_evsel_builtin): Delete.
20820 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20821 (spe_init_builtins): Delete.
20822 (spe_func_has_64bit_regs_p): Delete.
20823 (savres_routine_name): Delete "info" parameter. Adjust callers.
20824 (rs6000_emit_stack_reset): Ditto.
20825 (rs6000_dwarf_register_span): Delete.
20826 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20827 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20828 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20830 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20832 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20833 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20834 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20835 mulv2sf3, divv2sf3): Delete expanders.
20837 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20839 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20841 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20843 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20844 * config/rs6000/rs6000.c: Ditto.
20846 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20848 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20849 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20850 comparison_operator.
20852 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20854 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20855 * config/rs6000/rs6000.opt: Ditto.
20856 * config/rs6000/t-rtems: Ditto.
20858 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20860 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20861 TARGET_E500_SINGLE by 0, simplify.
20862 * config/rs6000/rs6000.c: Ditto.
20863 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20864 (spe_build_register_parallel): Delete.
20865 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20866 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20867 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20868 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20869 (E500_CONVERT): Delete.
20870 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20872 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20874 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20875 * config/rs6000/dfp.md: Ditto.
20876 (negdd2, *negdd2_fpr): Merge.
20877 (absdd2, *absdd2_fpr): Merge.
20878 (negtd2, *negtd2_fpr): Merge.
20879 (abstd2, *abstd2_fpr): Merge.
20880 * config/rs6000/e500.h: Delete file.
20881 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20882 TARGET_FPRS by 1 and simplify.
20883 * config/rs6000/rs6000-c.c: Ditto.
20884 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20885 TARGET_DF_SPE by 0.
20886 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20888 * config/rs6000/rs6000.md: Ditto.
20889 (floatdidf2, *floatdidf2_fpr): Merge.
20890 (move_from_CR_gt_bit): Delete.
20891 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20892 (E500_CR_IOR_COMPARE): Delete.
20893 (All patterns that require !TARGET_FPRS): Delete.
20894 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20896 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20898 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20900 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20902 * graphds.c (add_edge): Intitialize edge's attached data.
20903 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20904 pointer parameter. Call pointed function on each edge during
20905 graph traversing. Skip traversing the edge when the function
20907 (graphds_dfs, graphds_scc): Ditto.
20908 (for_each_edge): New parameter. Pass the new parameter to callback
20910 * graphds.h (skip_edge_callback): New function pointer type.
20911 (graphds_dfs, graphds_scc): New function pointer parameter.
20912 (graphds_edge_callback, for_each_edge): New parameter.
20914 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20916 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20917 out code checking if runtime alias check is possible to below ...
20918 Call the new function.
20919 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20920 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20922 2017-06-07 Marek Polacek <polacek@redhat.com>
20925 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20926 TYPE_OVERFLOW_WRAPS check.
20928 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20930 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20931 if versioning is required.
20932 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20933 peeling with the check for versioning.
20935 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20937 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20938 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20939 Set true to new parameter if new ssa variable is defined.
20940 (vect_gen_vector_loop_niters): Refactor. Set range information
20941 for the new vector loop bound variable.
20942 (vect_do_peeling): Ditto.
20944 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20946 * tree-affine.c (ssa.h): Include header file.
20947 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20948 has wrapping overflow behavior.
20950 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20952 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20954 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20956 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20957 (tree_to_aff_combination): ... here.
20959 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20961 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20962 reg_pressure model function.
20963 (ivopts_global_cost_for_size): Delete.
20964 (determine_set_costs, iv_ca_recount_cost): Call new model function
20965 ivopts_estimate_reg_pressure.
20967 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20969 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20970 expensive than udiv. Remove floating point cases from mod.
20972 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20974 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20975 Increase idiv cost.
20977 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20979 * config/aarch64/aarch64.md
20980 (copysignsf3): Fix mask generation.
20982 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20984 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
20986 (class dump_manager): Add register_dumps method.
20987 * dumpfile.c: Include langhooks.h.
20988 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
20989 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
20990 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
20991 (dump_manager::dump_register): Start with 512 entries instead of 32.
20992 (dump_manager::register_dumps): New method.
20993 * toplev.c (general_init): Instead of invoking register_dumps
20994 langhook, invoke register_dumps method on the dump manager.
20995 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
20998 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21000 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21001 Say that # requires an associated define_split to exist, and that
21002 the define_split must be suitable for use after register allocation.
21004 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21006 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21007 (compute_outgoing_frequencies): Also initialize zero counts.
21008 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21009 around loops; preserve more of profile when nothing changes.
21011 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21013 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21015 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21016 * config/arm/arm-cpu-cdata.h: Regenerate.
21017 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21018 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21019 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21020 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21021 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21023 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21025 * config/arm/t-rmprofile: Likewise.
21026 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21028 2017-06-06 David S. Miller <davem@davemloft.net>
21031 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21032 blockage if function uses alloca.
21034 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21036 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21037 New "uid" fields to hold pretty-print IDs of group and ref.
21038 Memory references are now identified as <group_id>:<ref_id>
21039 instead of using [random] addresses.
21040 (dump_mem_details): Simplify, no functional change.
21041 (dump_mem_ref): Simplify and make output more concise.
21042 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21043 (find_or_create_group): Initialize group uid.
21044 (record_ref): Initialize ref uid. Improve debug output.
21045 (prune_group_by_reuse, should_issue_prefetch_p,)
21046 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21047 (mark_nontemporal_store, determine_loop_nest_reuse):
21048 Improve debug output.
21050 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21052 * dbgcnt.def (prefetch): New debug counter.
21053 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21054 (schedule_prefetches): Stop issueing prefetches if debug counter
21057 2017-06-06 Tom de Vries <tom@codesourcery.com>
21059 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21060 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21062 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21064 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21065 Use aarch64_reg_or_zero predicate for operand 4.
21066 (aarch64_compare_and_swap<mode> define_insn_and_split):
21067 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21068 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21070 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21072 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21073 (arm_compute_save_core_reg_mask): This.
21074 (thumb1_compute_save_reg_mask): Rename into ...
21075 (thumb1_compute_save_core_reg_mask): This.
21076 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21077 (arm_compute_frame_layout): Likewise.
21079 2017-06-06 Richard Biener <rguenther@suse.de>
21081 PR tree-optimization/80974
21082 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21083 keep or clear leaders SSA info.
21085 2017-06-06 Tom de Vries <tom@codesourcery.com>
21087 * config/nvptx/nvptx.c (split_mode_p): New function.
21088 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21090 2017-06-06 Tom de Vries <tom@codesourcery.com>
21092 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21094 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21097 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21100 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21102 * shrink-wrap.c (handle_simple_exit): Update profile.
21103 (try_shrink_wrapping): Upate profile.
21105 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21107 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21108 (tree_guess_outgoing_edge_probabilities): New.
21109 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21110 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21112 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21114 * ipa-split.c (split_function): Initialize return bb profile.
21116 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21118 * profile.c (compute_branch_probabilities): Also initialize
21119 EXIT_BLOCK profile.
21121 2017-06-06 Richard Biener <rguenther@suse.de>
21123 PR tree-optimization/80928
21124 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21125 (vect_analyze_loop_operations): Properly guard analysis for
21127 (vect_transform_loop): Likewise.
21128 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21129 (vect_model_induction_cost): Do not cost for pure SLP.
21130 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21131 of induction in inner loop vectorization.
21132 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21133 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21134 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21136 (vect_analyze_slp_cost_1): Cost induction.
21137 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21138 (vect_get_slp_vect_defs): Likewise.
21139 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21140 (vect_transform_stmt): Handle SLP reductions.
21141 * tree-vectorizer.h (vectorizable_induction): Adjust.
21143 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21145 * config/rs6000/rs6000.c (make_resolver_func): Update
21146 init_lowered_empty_function call.
21148 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21150 * doc/invoke.texi: Document the -fprofile-abs-path option.
21151 * common.opt (fprofile-abs-path): New option.
21152 * gcov-io.h (gcov_write_filename): Declare.
21153 * gcov-io.c (gcov_write_filename): New function.
21154 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21155 * profile.c (output_location): Likewise.
21157 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21159 * shring-wrap.c: Revert accidental commit.
21161 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21163 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21165 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21167 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21169 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21170 profile in callgraph edge.
21171 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21172 (apply_scale): Likewise.
21173 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
21174 Also scale profile when inlining function with zero profile.
21175 (initialize_cfun): Update exit block profile even when it is zero.
21176 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
21177 when profile is read.
21179 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21181 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
21182 (CLONE_*): New constants to define the processors we can generate
21183 code for with the target_clone attribute.
21184 (rs6000_clone_map): New array to identify which clone processors
21185 the current program is running on.
21186 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
21187 target_clone attribute.
21188 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
21189 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
21190 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
21191 (cpu_expand_builtin): Add support for target_clone attribute.
21192 (rs6000_valid_attribute_p): Allow "default" attribute.
21193 (get_decl_name): New debug function to simplify printing the
21194 current function name in debugging statements.
21195 (rs6000_clone_priority): New functions to support the target_clone
21196 attribute, and be able to generate code to switch between ISA 2.05
21197 through ISA 3.0 (power6 through power9).
21198 (rs6000_compare_version_priority): Likewise.
21199 (rs6000_get_function_versions_dispatcher): Likewise.
21200 (make_resolver_func): Likewise.
21201 (add_condition_to_bb): Likewise.
21202 (dispatch_function_versions): Likewise.
21203 (rs6000_generate_version_dispatcher_body): Likewise.
21204 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
21205 (fusion_gpr_load_p): Fix a spacing issue.
21206 * doc/extend.texi (Common Function Attributes): Document that the
21207 PowerPC supports the target_clone attribute.
21209 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21211 * config/arm/arm.h: explain F symbol found in description of ARM
21212 register allocation in its legend.
21214 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21216 * config/mips/frame-header-opt.c: Include profile-count.h.
21217 * config/riscv/riscv.c: Include profile-count.h
21219 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21221 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
21223 (sm_set_flag_if_changed): Add bbs field.
21224 (execute_sm_if_changed_flag_set): Pass BBS.
21225 (execute_sm): Update.
21227 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21229 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
21232 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21234 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
21235 (peephole2): New peephole2 to emit the above.
21236 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21238 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21240 * config/aarch64/aarch64.c (define_peephole2 above
21241 *sub_<shift>_<mode>): New peephole.
21243 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21245 * config/i386/i386.c (make_resolver_func): Update.
21246 * Makefile.in: Add profile-count.h and profile-count.o
21247 * auto-profile.c (afdo_indirect_call): Update to new API.
21248 (afdo_set_bb_count): Update.
21249 (afdo_propagate_edge): Update.
21250 (afdo_propagate_circuit): Update.
21251 (afdo_calculate_branch_prob): Update.
21252 (afdo_annotate_cfg): Update.
21253 * basic-block.h: Include profile-count.h
21254 (struct edge_def): Turn count to profile_count.
21255 (struct basic_block_def): Likewie.
21256 (REG_BR_PROB_BASE): Move to profile-count.h
21257 (RDIV): Move to profile-count.h
21258 * bb-reorder.c (max_entry_count): Turn to profile_count.
21259 (find_traces): Update.
21260 (rotate_loop):Update.
21261 (connect_traces):Update.
21262 (sanitize_hot_paths):Update.
21263 * bt-load.c (migrate_btr_defs): Update.
21264 * cfg.c (RDIV): Remove.
21265 (init_flow): Use alloc_block.
21266 (alloc_block): Uninitialize count.
21267 (unchecked_make_edge): Uninitialize count.
21268 (check_bb_profile): Update.
21269 (dump_edge_info): Update.
21270 (dump_bb_info): Update.
21271 (update_bb_profile_for_threading): Update.
21272 (scale_bbs_frequencies_int): Update.
21273 (scale_bbs_frequencies_gcov_type): Update.
21274 (scale_bbs_frequencies_profile_count): New.
21275 * cfg.h (update_bb_profile_for_threading): Update.
21276 (scale_bbs_frequencies_profile_count): Declare.
21277 * cfgbuild.c (compute_outgoing_frequencies): Update.
21278 (find_many_sub_basic_blocks): Update.
21279 * cfgcleanup.c (try_forward_edges): Update.
21280 (try_crossjump_to_edge): Update.
21281 * cfgexpand.c (expand_gimple_tailcall): Update.
21282 (construct_exit_block): Update.
21283 * cfghooks.c (verify_flow_info): Update.
21284 (dump_bb_for_graph): Update.
21285 (split_edge): Update.
21286 (make_forwarder_block): Update.
21287 (duplicate_block): Update.
21288 (account_profile_record): Update.
21289 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21290 (get_estimated_loop_iterations): Update.
21291 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21292 (single_likely_exit): Update.
21293 * cfgloopmanip.c (scale_loop_profile): Update.
21295 (set_zero_probability): Update.
21296 (lv_adjust_loop_entry_edge): Update.
21297 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21298 (purge_dead_edges): Update.
21299 (rtl_account_profile_record): Update.
21300 * cgraph.c (cgraph_node::create): Uninitialize count.
21301 (symbol_table::create_edge): Uninitialize count.
21302 (cgraph_update_edges_for_call_stmt_node): Update.
21303 (cgraph_edge::dump_edge_flags): Update.
21304 (cgraph_node::dump): Update.
21305 (cgraph_edge::maybe_hot_p): Update.
21306 * cgraph.h: Include profile-count.h
21307 (create_clone), create_edge, create_indirect_edge): Update.
21308 (cgraph_node): Turn count to profile_count.
21309 (cgraph_edge0: Likewise.
21310 (make_speculative, clone): Update.
21311 (create_edge): Update.
21312 (init_lowered_empty_function): Update.
21313 * cgraphclones.c (cgraph_edge::clone): Update.
21314 (duplicate_thunk_for_node): Update.
21315 (cgraph_node::create_clone): Update.
21316 * cgraphunit.c (cgraph_node::analyze): Update.
21317 (cgraph_node::expand_thunk): Update.
21318 * final.c (dump_basic_block_info): Update.
21319 * gimple-streamer-in.c (input_bb): Update.
21320 * gimple-streamer-out.c (output_bb): Update.
21321 * graphite.c (print_global_statistics): Update.
21322 (print_graphite_scop_statistics): Update.
21323 * hsa-brig.c: Include basic-block.h.
21324 * hsa-dump.c: Include basic-block.h.
21325 * hsa-gen.c (T sum_slice): Update.
21326 (convert_switch_statements):Update.
21327 * hsa-regalloc.c: Include basic-block.h.
21328 * ipa-chkp.c (chkp_produce_thunks): Update.
21329 * ipa-cp.c (struct caller_statistics): Update.
21330 (init_caller_stats): Update.
21331 (gather_caller_stats): Update.
21332 (ipcp_cloning_candidate_p): Update.
21333 (good_cloning_opportunity_p): Update.
21334 (get_info_about_necessary_edges): Update.
21335 (dump_profile_updates): Update.
21336 (update_profiling_info): Update.
21337 (update_specialized_profile): Update.
21338 (perhaps_add_new_callers): Update.
21339 (decide_about_value): Update.
21340 (ipa_cp_c_finalize): Update.
21341 * ipa-devirt.c (struct odr_type_warn_count): Update.
21342 (struct decl_warn_count): Update.
21343 (struct final_warning_record): Update.
21344 (possible_polymorphic_call_targets): Update.
21345 (ipa_devirt): Update.
21346 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21347 * ipa-icf.c (sem_function::merge): Update.
21348 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21349 * ipa-inline.c (compute_uninlined_call_time): Update.
21350 (compute_inlined_call_time): Update.
21351 (want_inline_small_function_p): Update.
21352 (want_inline_self_recursive_call_p): Update.
21353 (edge_badness): Update.
21354 (lookup_recursive_calls): Update.
21355 (recursive_inlining): Update.
21356 (inline_small_functions): Update.
21357 (dump_overall_stats): Update.
21358 (dump_inline_stats): Update.
21359 * ipa-profile.c (ipa_profile_generate_summary): Update.
21360 (ipa_propagate_frequency): Update.
21361 (ipa_profile): Update.
21362 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21363 * ipa-utils.c (ipa_merge_profiles): Update.
21364 * loop-doloop.c (doloop_modify): Update.
21365 * loop-unroll.c (report_unroll): Update.
21366 (unroll_loop_runtime_iterations): Update.
21367 * lto-cgraph.c (lto_output_edge): Update.
21368 (lto_output_node): Update.
21369 (input_node): Update.
21370 (input_edge): Update.
21371 (merge_profile_summaries): Update.
21372 * lto-streamer-in.c (input_cfg): Update.
21373 * lto-streamer-out.c (output_cfg): Update.
21374 * mcf.c (create_fixup_graph): Update.
21375 (adjust_cfg_counts): Update.
21376 (sum_edge_counts): Update.
21377 * modulo-sched.c (sms_schedule): Update.
21378 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21379 * predict.c (maybe_hot_count_p): Update.
21380 (probably_never_executed): Update.
21381 (dump_prediction): Update.
21382 (combine_predictions_for_bb): Update.
21383 (propagate_freq): Update.
21384 (handle_missing_profiles): Update.
21385 (counts_to_freqs): Update.
21386 (rebuild_frequencies): Update.
21387 (force_edge_cold): Update.
21388 * predict.h: Include profile-count.h
21389 (maybe_hot_count_p, counts_to_freqs): UPdate.
21390 * print-rtl-function.c: Do not include cfg.h
21391 * print-rtl.c: Include basic-block.h
21392 * profile-count.c: New file.
21393 * profile-count.h: New file.
21394 * profile.c (is_edge_inconsistent): Update.
21395 (correct_negative_edge_counts): Update.
21396 (is_inconsistent): Update.
21397 (set_bb_counts): Update.
21398 (read_profile_edge_counts): Update.
21399 (compute_frequency_overlap): Update.
21400 (compute_branch_probabilities): Update; Initialize and deinitialize
21402 (branch_prob): Update.
21403 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21404 (edge_gcov_count): New.
21405 (bb_gcov_count): New.
21406 * shrink-wrap.c (try_shrink_wrapping): Update.
21407 * tracer.c (better_p): Update.
21408 * trans-mem.c (expand_transaction): Update.
21409 (ipa_tm_insert_irr_call): Update.
21410 (ipa_tm_insert_gettmclone_call): Update.
21411 * tree-call-cdce.c: Update.
21412 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21413 (gimple_duplicate_sese_tail): Update.
21414 (gimple_account_profile_record): Update.
21415 (execute_fixup_cfg): Update.
21416 * tree-inline.c (copy_bb): Update.
21417 (copy_edges_for_bb): Update.
21418 (initialize_cfun): Update.
21419 (freqs_to_counts): Update.
21420 (copy_cfg_body): Update.
21421 (expand_call_inline): Update.
21422 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21423 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21424 (try_unroll_loop_completely): Update.
21425 (try_peel_loop): Update.
21426 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21427 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21428 * tree-ssa-loop-split.c (connect_loops): Update.
21429 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21430 * tree-ssa-reassoc.c (branch_fixup): Update.
21431 * tree-ssa-tail-merge.c (replace_block_by): Update.
21432 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21433 (compute_path_counts): Update.
21434 (update_profile): Update.
21435 (recompute_probabilities): Update.
21436 (update_joiner_offpath_counts): Update.
21437 (estimated_freqs_path): Update.
21438 (freqs_to_counts_path): Update.
21439 (clear_counts_path): Update.
21440 (ssa_fix_duplicate_block_edges): Update.
21441 (duplicate_thread_path): Update.
21442 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21443 (struct switch_conv_info): Update.
21444 * tree-tailcall.c (decrease_profile): Update.
21445 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21446 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21447 * value-prof.c (check_counter): Update.
21448 (gimple_divmod_fixed_value): Update.
21449 (gimple_mod_pow2): Update.
21450 (gimple_mod_subtract): Update.
21451 (gimple_ic_transform): Update.
21452 (gimple_stringop_fixed_value): Update.
21453 * value-prof.h (gimple_ic): Update.
21455 2017-06-02 Carl Love <cel@us.ibm.com>
21457 * config/rs6000/rs6000-c: Add support for built-in functions
21458 vector double vec_doublee (vector signed int);
21459 vector double vec_doublee (vector unsigned int);
21460 vector double vec_doublee (vector float);
21461 vector double vec_doubleh (vector signed int);
21462 vector double vec_doubleh (vector unsigned int);
21463 vector double vec_doubleh (vector float);
21464 vector double vec_doublel (vector signed int);
21465 vector double vec_doublel (vector unsigned int);
21466 vector double vec_doublel (vector float);
21467 vector double vec_doubleo (vector signed int);
21468 vector double vec_doubleo (vector unsigned int);
21469 vector double vec_doubleo (vector float);.
21470 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21471 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21473 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21474 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21475 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21477 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21478 vec_doublel, vec_doubleh.
21479 * doc/extend.texi: Update the built-in documentation file for the
21480 new built-in functions.
21482 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21485 * ipa-inline-analysis.c (free_growth_caches): Set
21486 edge_removal_hook_holder to NULL after removing it.
21488 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21490 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21491 comparision with zero.
21493 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21494 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21495 for early expansion of vec_min and vec_max builtins.
21496 (builtin_function_type): Add min/max unsigned variants to those
21497 identified as having unsigned arguments.
21499 2017-06-02 Olivier Hainque <hainque@adacore.com>
21501 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21503 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21505 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21506 Use VALL_F16 iterator rather than VALL.
21508 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21510 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21511 Emit CBNZ inside loop when doing a strong exchange and comparing
21512 against zero. Generate the CC flags after the loop.
21514 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21516 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21517 (dl_section_ref): New.
21518 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21519 On AIX, append an expression to subtract the size of the
21520 section length to dl_section_ref.
21522 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21524 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21525 for early expansion of vector absolute builtins.
21527 2017-06-02 Richard Biener <rguenther@suse.de>
21529 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21530 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21532 2017-06-02 Richard Biener <rguenther@suse.de>
21534 PR tree-optimization/80948
21535 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21536 stmt order as well.
21538 2017-06-02 Richard Biener <rguenther@suse.de>
21540 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21542 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21543 for inductions as relevant.
21545 2017-06-02 Richard Biener <rguenther@suse.de>
21547 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21548 (vectorizable_induction): ... this. Remove dead code.
21550 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21552 * builtins. (expand_builtin_alloca): Remove second parameter and
21553 infer its value from the first parameter instead.
21554 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21556 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21558 PR rtl-optimization/80903
21559 * loop-doloop.c (add_test): Unshare sequence.
21561 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21563 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21565 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21567 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21569 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21570 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21571 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21572 (xlogue_layout::get_stub_rtx): Make static.
21573 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21574 (xlogue_layout::compute_stub_managed_regs): Rename to...
21575 (xlogue_layout::count_stub_managed_regs): ...this.
21576 (xlogue_layout::is_stub_managed_reg): New function.
21577 (xlogue_layout::m_stub_names): Rename to...
21578 (xlogue_layout::s_stub_names): ...this, make static.
21579 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21580 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21581 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21582 xlogue_layout::s_stub_names): Instantiate statics.
21583 (stub_managed_regs): Remove.
21584 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21585 (disable_call_ms2sysv_xlogues): Rename to...
21586 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21587 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21589 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21590 change after reload_completed.
21591 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21593 (ix86_expand_prologue): Likewise.
21594 (ix86_expand_epilogue): Likewise.
21595 (ix86_expand_split_stack_prologue): Likewise.
21596 (ix86_compute_frame_layout): Remove frame parameter ...
21597 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21598 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21600 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21601 (ix86_frame): Move from here ...
21602 * config/i386/i386.h (ix86_frame): ... to here.
21603 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21604 complete ix86_frame data structure instead. Remove some_ld_name.
21606 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21608 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21609 symbols that hold a DECL_VALUE_EXPR.
21611 2017-06-01 Martin Jambor <mjambor@suse.cz>
21613 PR tree-optimization/80898
21614 * tree-sra.c (process_subtree_disqualification): Removed.
21615 (disqualify_candidate): Do not acll
21616 process_subtree_disqualification.
21617 (subtree_mark_written_and_enqueue): New function.
21618 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21619 RHS has been disqualified and re-queue LHS if necessary. Apart
21620 from that, ignore disqualified RHS.
21622 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21624 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21625 address fetch for z10 or later.
21627 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21629 * config/arc/arc.md (tst_movb): Add guard when splitting.
21631 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21633 * config/arc/arc.c (arc_can_eliminate): Test against
21634 arc_frame_pointer_needed.
21636 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21638 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21639 to prevent store reordering.
21640 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21641 (type): Add block type.
21642 (stack_tie): Define special instruction to be used in
21645 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21647 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21648 constraint. It is not valid for the pattern.
21649 (noncommutative_binary_comparison): Likewise.
21651 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21653 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21656 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21658 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21659 be used by the reg-alloc.
21661 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21663 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21664 reg-alloc when having mul64 or mul32x16 instructions.
21665 (mulsidi3): Likewise.
21666 (umulsidi3): Likewise.
21667 (mulsi32x16): New pattern.
21668 (mulsi64): Likewise.
21669 (mulsidi64): Likewise.
21670 (umulsidi64): Likewise.
21671 (MUL32x16_REG): Define.
21672 (mul64_600): Use MUL32x16_REG.
21673 (mac64_600): Likewise.
21674 (umul64_600): Likewise.
21675 (umac64_600): Likewise.
21677 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21679 * config/arc/arc.md (mulsi3_700): Make it commutative.
21681 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
21683 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
21685 (*sign_extendsidi2_insn): Likewise for movstosw.
21687 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21689 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21690 the type of the input discriminant value. Convert the
21691 discriminant value of signedness vary.
21693 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21695 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21696 Add to -Wall section.
21698 2017-06-01 Richard Biener <rguenther@suse.de>
21700 PR middle-end/66313
21701 * fold-const.c (fold_plusminus_mult_expr): If the factored
21702 factor may be zero use a wrapping type for the inner operation.
21703 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21704 and handle moved defs.
21705 (process_assignment): Properly guard the unary op case. Return a
21706 tri-state indicating that moving the stmt before the call may allow
21707 to continue. Pass through to_move.
21708 (find_tail_calls): Handle moving unrelated defs before
21711 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21714 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21715 splitter result in the canonical way.
21717 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21719 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21720 also for 32bit target. Update insn attributes.
21721 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21723 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21725 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21726 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21727 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21728 (_mm_maskz_min_ss): New intrinsics.
21730 2017-05-31 Martin Liska <mliska@suse.cz>
21732 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21733 Change comment style to one we normally use.
21734 (vectorizable_reduction): Likewise.
21735 (vectorizable_induction): Likewise.
21736 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21737 (vectorizable_call): Likewise.
21738 (vectorizable_simd_clone_call): Likewise.
21739 (vectorizable_conversion): Likewise.
21740 (vectorizable_assignment): Likewise.
21741 (vectorizable_shift): Likewise.
21742 (vectorizable_operation): Likewise.
21743 (vectorizable_store): Likewise.
21744 (vectorizable_load): Likewise.
21745 * tree-vectorizer.h: Likewise.
21747 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21749 * passes.c (emergency_dump_function): New.
21750 * tree-pass.h (emergency_dump_function): Declare.
21751 * plugin.c (plugins_internal_error_function): Remove.
21752 * plugin.h (plugins_internal_error_function): Remove declaration.
21753 * toplev.c (internal_error_function): New static function. Use it...
21754 (general_init): ...here.
21756 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21758 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21759 (arc_rtx_costs): Add costs for new patterns.
21760 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21761 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21763 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21765 * tree-ssa-strlen.c (get_next_strinfo): New function.
21766 (get_stridx_plus_constant): Use it.
21767 (zero_length_string): Likewise.
21768 (adjust_related_strinfos): Likewise.
21769 (adjust_last_stmt): Likewise.
21771 2017-05-31 Richard Biener <rguenther@suse.de>
21774 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21775 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21777 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21779 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21780 loop_vinfo argument and use of dependence distance vectors.
21781 Check instead whether the two references differ only in their
21782 initial value and assume that they have the same alignment if the
21783 difference is a multiple of the vector alignment.
21784 (vect_analyze_data_refs_alignment): Update call accordingly.
21786 2017-05-31 Martin Liska <mliska@suse.cz>
21789 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21791 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21793 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21794 (create_intersect_range_checks): Move from ...
21795 * tree-data-ref.c (create_intersect_range_checks_index)
21796 (create_intersect_range_checks): ... to here.
21797 (create_runtime_alias_checks): New function factored from ...
21798 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21799 here. Call above function.
21800 * tree-data-ref.h (create_runtime_alias_checks): New function.
21802 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21804 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21805 segment length for dr_b and compute it in wide_int.
21807 2017-05-31 Richard Biener <rguenther@suse.de>
21809 PR tree-optimization/80906
21810 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21811 and pass through iv_map.
21812 (copy_bb_and_scalar_dependences): Adjust.
21813 (translate_pending_phi_nodes): Likewise.
21814 (copy_loop_close_phi_args): Handle code-generating IVs instead
21817 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21819 * diagnostic-color.c (color_dict): Add "type-diff".
21820 (parse_gcc_colors): Update comment.
21821 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21822 -fdiagnostics-show-template-tree and -fno-elide-type.
21823 (GCC_COLORS): Add type-diff to example.
21825 (-fdiagnostics-show-template-tree): New.
21826 (-fno-elide-type): New.
21827 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21828 the pp_format_decoder callback. Call any m_format_postprocessor's
21830 (pretty_printer::pretty_printer): Initialize
21831 m_format_postprocessor.
21832 (pretty_printer::~pretty_printer): Delete any
21833 m_format_postprocessor.
21834 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21835 (class format_postprocessor): New class.
21836 (struct pretty_printer::format_decoder): Document the new parameters.
21837 (struct pretty_printer::m_format_postprocessor): New field.
21838 * tree-diagnostic.c (default_tree_printer): Update for new
21839 bool and const char ** params.
21840 * tree-diagnostic.h (default_tree_printer): Likewise.
21842 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21844 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21845 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21846 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21847 rs6000_gen_cell_microcode code.
21848 (rs6000_final_prescan_insn): Delete.
21849 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21850 "warn-cell-microcode" entries.
21851 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21852 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21853 throughout. Change cc_reg_not_micro_cr0_operand to
21854 cc_reg_not_cr0_operand throughout.
21855 (*extendhi<mode>2_noload): Delete.
21856 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21857 (mwarn-cell-microcode): Delete.
21858 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21859 -mgen-cell-microcode and -mwarn-cell-microcode.
21861 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21864 * config/i386/constraints.md (Yd): New constraint.
21866 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21867 and (?Yd, r) alternatives. Update insn attributes.
21868 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21869 and (?*Yd, r) alternatives. Update insn attributes.
21870 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21872 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21874 * gimplify.c (gimplify_modify_expr): Don't create a
21875 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21878 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21880 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21882 2017-05-30 Richard Biener <rguenther@suse.de>
21884 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21885 and reduc_def fields.
21886 (STMT_VINFO_REDUC_TYPE): New define.
21887 (STMT_VINFO_REDUC_DEF): Likewise.
21888 (vect_force_simple_reduction): Adjust prototype.
21889 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21890 (vect_is_simple_reduction): Remove check_reduction argument.
21891 (vect_force_simple_reduction): Adjust and set
21892 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21893 (vectorizable_reduction): Do not re-do reduction analysis
21894 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21895 * tree-parloops.c (gather_scalar_reductions): Adjust.
21897 2017-05-30 Richard Biener <rguenther@suse.de>
21899 PR middle-end/80901
21900 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21903 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21905 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21906 Introduce unknown_misalignment parameter and remove vf.
21907 (vect_peeling_hash_get_lowest_cost):
21908 Pass unknown_misalignment parameter.
21909 (vect_enhance_data_refs_alignment):
21910 Fix unsupportable data ref treatment.
21912 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21914 * tree-vect-data-refs.c (vect_get_data_access_cost):
21915 Workaround for SLP handling.
21916 (vect_enhance_data_refs_alignment):
21917 Compute costs for doing no peeling at all, compare to the best
21918 peeling costs so far and avoid peeling if cheaper.
21920 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21922 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21923 Return peeling info and set costs to zero for unlimited cost
21925 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21926 with unknown misalignment. Compute and costs for unknown
21927 misalignment, compare them to the costs for known misalignment
21928 and choose the cheapest for peeling.
21930 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21932 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21933 (vect_get_peeling_costs_all_drs): Create function.
21934 (vect_peeling_hash_get_lowest_cost):
21935 Use vect_get_peeling_costs_all_drs.
21936 (vect_peeling_supportable): Create function.
21937 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21939 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21941 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21942 DR_HAS_NEGATIVE_STEP.
21943 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21944 (vect_enhance_data_refs_alignment): Use.
21945 (vect_duplicate_ssa_name_ptr_info): Use.
21946 * tree-vectorizer.h (dr_misalignment): Use.
21947 (known_alignment_for_access_p): Use.
21949 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21952 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21954 (has_section_name): New function.
21956 2017-05-30 Martin Liska <mliska@suse.cz>
21959 * auto-profile.c (get_function_decl_from_block): Fix
21962 2017-05-30 Richard Biener <rguenther@suse.de>
21964 PR middle-end/80876
21965 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21967 2017-05-30 Martin Liska <mliska@suse.cz>
21969 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21970 * dumpfile.h (struct dump_file_info): Remove ctors.
21972 2017-05-30 Martin Liska <mliska@suse.cz>
21974 * predict.def: Fix GNU coding style.
21976 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21978 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21979 Mark 'to' argument with ATTRIBUTE_UNUSED.
21981 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21983 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21984 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
21985 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
21988 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
21990 * doc/install.texi (Options specification): Restore entry of
21991 --enable-sjlj-exceptions.
21993 2017-05-27 Michael Eager <eager@eagercon.com>
21996 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
21998 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22000 * config/microblaze/microblaze.h
22001 (FIXED_REGISTERS): Update in macro.
22002 (CALL_USED_REGISTERS): Update in macro.
22004 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22006 * doc/install.texi: Add links to macOS binary distributions.
22008 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22012 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22014 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22016 2017-05-26 Martin Liska <mliska@suse.cz>
22018 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22020 2017-05-26 Martin Liska <mliska@suse.cz>
22022 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22023 always leading ';; '.
22024 (dump_bb_info): Likewise.
22025 (brief_dump_cfg): Likewise.
22026 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22027 * dumpfile.c: Remove usage of TDF_VERBOSE.
22028 * dumpfile.h (enum dump_kind): Likewise.
22029 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22030 * print-tree.c (debug_verbose): Remove.
22031 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22032 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22033 * tree-diagnostic.c (default_tree_printer): Replace
22034 TDF_DIAGNOSTIC with TDF_SLIM.
22036 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22038 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22039 in parameter loop, rather than loop_vinfo.
22040 (create_intersect_range_checks): Ditto.
22041 (vect_create_cond_for_alias_checks): Update call to above functions.
22043 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22045 PR tree-optimization/80815
22046 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22047 for merging runtime alias checks. Handle negative DR_STEPs.
22049 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22051 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22053 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22054 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22055 out code pruning runtime alias checks.
22056 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22057 factored out from above.
22058 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22060 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22062 (prune_runtime_alias_test_list): New decalaration.
22064 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22066 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22067 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22068 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22069 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22070 (operator==, comp_dr_with_seg_len_pair): Ditto.
22071 (vect_prune_runtime_alias_test_list): Ditto.
22073 2017-05-26 Martin Liska <mliska@suse.cz>
22076 * params.def: Bound partial-inlining-entry-probability param.
22078 2017-05-26 Marek Polacek <polacek@redhat.com>
22081 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22084 2017-05-26 Richard Biener <rguenther@suse.de>
22086 PR tree-optimization/80842
22087 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22090 2017-05-26 Richard Biener <rguenther@suse.de>
22092 PR tree-optimization/80844
22093 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22095 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22097 * doc/md.texi (Machine Constraints): Update x86 family
22098 machine constraints section to match 'config/i386/constraints.md'.
22100 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22102 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22104 2017-05-25 Nathan Sidwell <nathan@acm.org>
22106 * doc/invoke.texi (--enable-languages): Update documentation.
22108 2017-05-25 Martin Liska <mliska@suse.cz>
22110 * dumpfile.c: Add TDF_FOLDING.
22111 * dumpfile.h (enum dump_kind): Likewise.
22112 * genmatch.c (dt_simplify::gen_1): Use it.
22114 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22116 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22118 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22120 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22121 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22123 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22125 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22126 * match.pd (X == C): Rewrite it here.
22127 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22128 with_certain_nonzero_bits2): New predicates.
22129 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22131 2017-05-24 Nathan Sidwell <nathan@acm.org>
22133 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22136 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22139 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22141 * config/powerpcspe: New port. Files are copied from the rs6000
22142 port, with "rs6000" in filenames replaced by "powerpcspe".
22144 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22146 PR rtl-optimization/80754
22147 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22149 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22151 * config/sparc/sparc.md (length): Return the correct value for -mflat
22152 sibcalls to match output_sibcall.
22154 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22158 * config/rs6000/rs6000.c (struct machine_function): Add new field
22160 (rs6000_get_separate_components): Init that field, use it.
22161 (rs6000_components_for_bb): Use the field.
22163 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22165 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22167 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22169 PR middle-end/80823
22170 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22172 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22175 * config/s390/s390.c (s390_check_qrst_address): Check incoming
22176 address against address_operand predicate.
22177 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
22179 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
22181 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
22182 parameters passed indirectly.
22184 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
22186 * config/i386/i386.md (*movdi_internal): Remove SSE4
22187 alternative 18 (?r, *v). Update insn attributes.
22188 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
22189 Update insn attributes.
22190 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
22191 Update insn attributes.
22192 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
22193 alternative 1 (r, v). Remove isa attribute.
22194 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
22195 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
22196 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
22198 2017-05-23 Tom de Vries <tom@codesourcery.com>
22200 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
22203 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22205 * cgraphunit.c (symbol_table::process_new_functions): Update.
22206 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
22207 (inline_generate_summary): Rename to ...
22208 (ipa_fn_summary_generate): ... this one.
22209 (inline_read_summary): Rename to ...
22210 (ipa_fn_summary_read): ... this one.
22211 (inline_write_summary): Rename to ...
22212 (ipa_fn_summary_write): ... this one.
22213 (inline_free_summary): Rename to ...
22214 (ipa_free_fn_summary): ... this one.
22215 (pass_data_local_fn_summary, pass_local_fn_summary,
22216 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
22217 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
22218 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
22219 make_pass_ipa_fn_summary): New.
22220 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
22221 inline_write_summary, inline_free_summary): Remove.
22222 (ipa_free_fn_summary) : New.
22223 * ipa-inline.c (ipa_inline): Update.
22224 (pass_ipa_inline): Do not generate summaries.
22225 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
22227 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
22228 and add pass_ipa_fn_summary.
22229 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
22231 (make_pass_inline_parameters): Remove.
22233 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
22235 * omp-low.c (struct omp_context): Remove "default_kind" member.
22238 * omp-offload.c (execute_oacc_device_lower): Remove the
22239 parallelism dimensions function attributes for unparallelized
22240 OpenACC kernels constructs.
22242 2017-05-23 Martin Liska <mliska@suse.cz>
22244 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22246 (cgraph_edge::make_speculative): Likewise.
22247 (cgraph_edge::resolve_speculation): Likewise.
22248 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22249 (cgraph_node::dump): Likewise.
22250 * cgraph.h: Likewise.
22251 * cgraphunit.c (analyze_functions): Likewise.
22252 (symbol_table::compile): Likewise.
22253 * ipa-cp.c (print_all_lattices): Likewise.
22254 (determine_versionability): Likewise.
22255 (initialize_node_lattices): Likewise.
22256 (ipcp_verify_propagated_values): Likewise.
22257 (estimate_local_effects): Likewise.
22258 (update_profiling_info): Likewise.
22259 (create_specialized_node): Likewise.
22260 (perhaps_add_new_callers): Likewise.
22261 (decide_about_value): Likewise.
22262 (decide_whether_version_node): Likewise.
22263 (identify_dead_nodes): Likewise.
22264 (ipcp_store_bits_results): Likewise.
22265 * ipa-devirt.c (dump_targets): Likewise.
22266 (ipa_devirt): Likewise.
22267 * ipa-icf.c (sem_item::dump): Likewise.
22268 (sem_function::equals): Likewise.
22269 (sem_variable::equals): Likewise.
22270 (sem_item_optimizer::read_section): Likewise.
22271 (sem_item_optimizer::execute): Likewise.
22272 (congruence_class::dump): Likewise.
22273 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22274 (dump_inline_summary): Likewise.
22275 (estimate_node_size_and_time): Likewise.
22276 (inline_analyze_function): Likewise.
22277 * ipa-inline-transform.c (inline_call): Likewise.
22278 * ipa-inline.c (report_inline_failed_reason): Likewise.
22279 (want_early_inline_function_p): Likewise.
22280 (edge_badness): Likewise.
22281 (update_edge_key): Likewise.
22282 (inline_small_functions): Likewise.
22283 * ipa-profile.c (ipa_profile): Likewise.
22284 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22285 (ipa_make_edge_direct_to_target): Likewise.
22286 (remove_described_reference): Likewise.
22287 (ipa_impossible_devirt_target): Likewise.
22288 (propagate_controlled_uses): Likewise.
22289 (ipa_print_node_params): Likewise.
22290 (ipcp_transform_function): Likewise.
22291 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22292 (propagate_pure_const): Likewise.
22293 * ipa-reference.c (generate_summary): Likewise.
22294 (read_write_all_from_decl): Likewise.
22295 (propagate): Likewise.
22296 (ipa_reference_read_optimization_summary): Likewise.
22297 * ipa-utils.c (ipa_merge_profiles): Likewise.
22298 * ipa.c (walk_polymorphic_call_targets): Likewise.
22299 (symbol_table::remove_unreachable_nodes): Likewise.
22300 (ipa_single_use): Likewise.
22301 * passes.c (execute_todo): Likewise.
22302 * predict.c (drop_profile): Likewise.
22303 * symtab.c (symtab_node::get_dump_name): New function.
22304 (symtab_node::dump_name): Likewise.
22305 (symtab_node::dump_asm_name): Likewise.
22306 (symtab_node::dump_references): Likewise.
22307 (symtab_node::dump_referring): Likewise.
22308 (symtab_node::dump_base): Likewise.
22309 (symtab_node::debug_symtab): Likewise.
22310 * tree-sra.c (convert_callers_for_node): Likewise.
22311 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22312 * value-prof.c (init_node_map): Likewise.
22314 2017-05-23 Martin Liska <mliska@suse.cz>
22316 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22317 and symtab_node::debug_symtab to symbol_table::debug.
22318 * cgraphunit.c (analyze_functions): Use the renamed function.
22319 (symbol_table::compile): Likewise.
22320 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22321 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22322 * passes.c (execute_todo): Likewise.
22323 * symtab.c (symbol_table::dump): New function.
22324 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22326 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22328 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22329 that nonconst implies exec.
22331 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22333 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22334 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22335 (inline_edge_summary_vec): Turn into ...
22336 (ipa_call_summaries): ... this one.
22337 (redirect_to_unreachable, edge_set_predicate,
22338 evaluate_properties_for_edge, inline_summary_alloc,
22339 reset_ipa_call_summary, reset_inline_summary,
22340 inline_summary_t::duplicate): Update.
22341 (inline_edge_duplication_hook): Turn to ...
22342 (ipa_call_summary_t::duplicate): ... this one.
22343 (inline_edge_removal_hook): Turn to ...
22344 (ipa_call_summary_t::remove): ... this one.
22345 (dump_inline_edge_summary): Turn to ...
22346 (dump_ipa_call_summary): ... this one.
22347 (estimate_function_body_sizes): Update.
22348 (inline_update_callee_summaries): Update.
22349 (remap_edge_change_prob): Update.
22350 (remap_edge_summaries): Update.
22351 (inline_merge_summary): Update.
22352 (do_estimate_edge_time): Update.
22353 (inline_generate_summary): Update.
22354 (inline_read_section): Update.
22355 (inline_read_summary): Update.
22356 (inline_free_summary): Update.
22357 * ipa-inline.c (can_inline_edge_p): Update.
22358 (compute_inlined_call_time): Update.
22359 (want_inline_small_function_p): Update.
22360 (edge_badness): Update.
22361 (early_inliner): Update.
22362 * ipa-inline.h (inline_edge_summary): Turn to ...
22363 (ipa_call_summary): ... this one.
22364 (ipa_call_summary_t): New class.
22365 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22366 (ipa_call_summaries): New.
22367 (inline_edge_summary): Remove.
22368 (estimate_edge_growth): Update.
22369 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22370 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22371 * ipa-split.c (execute_split_functions): Update.
22372 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22374 2017-05-23 Tom de Vries <tom@codesourcery.com>
22376 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22377 attributes): Document rdrand effective target.
22379 2017-05-23 Tom de Vries <tom@codesourcery.com>
22381 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22382 attributes): Sort alphabetically.
22384 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22386 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22388 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22391 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22392 V2DF/V2DI splat into two separate patterns, one that handles
22393 registers, and the other that only handles memory. Drop support
22394 for splatting from a GPR on ISA 2.07 and then splitting the
22395 splat into direct move and splat.
22396 (vsx_splat_<mode>_reg): Likewise.
22397 (vsx_splat_<mode>_mem): Likewise.
22399 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22401 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22403 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22405 PR middle-end/80809
22406 * omp-low.c (finish_taskreg_remap): New function.
22407 (finish_taskreg_scan): If unit size of ctx->record_type
22408 is non-constant, unshare the size expression and replace
22409 decls in it with possible outer var refs.
22411 PR middle-end/80809
22412 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22413 GOVD_SHARED rather than GOVD_PRIVATE with it.
22414 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22415 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22417 PR middle-end/80853
22418 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22419 as last argument to build_outer_var_ref for pointer bases of array
22420 section reductions.
22422 2017-05-19 Martin Sebor <msebor@redhat.com>
22424 * print-tree.c (print_node): Print DECL_READ_P flag.
22426 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22428 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22429 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22430 * cgraph.c: Likewise.
22431 * cgraphunit.c: Likewise.
22432 * gengtype.c: Likewise.
22433 * ipa-cp.c: Likewise.
22434 * ipa-devirt.c: Likewise.
22435 * ipa-icf.c: Likewise.
22436 * ipa-predicate.c: Likewise.
22437 * ipa-profile.c: Likewise.
22438 * ipa-prop.c: Likewise.
22439 * ipa-split.c: Likewise.
22441 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22442 edge_predicate_pool, dump_inline_hints,
22443 inline_summary::account_size_time, redirect_to_unreachable,
22444 edge_set_predicate, set_hint_predicate,
22445 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22446 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22447 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22448 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22449 ipa_call_summary_t::remove, initialize_growth_caches,
22450 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22451 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22452 mark_modified, unmodified_parm_1, unmodified_parm,
22453 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22454 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22455 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22456 will_be_nonconstant_predicate, record_modified_bb_info,
22457 get_minimal_bb, record_modified, param_change_prob,
22458 phi_result_unknown_predicate, predicate_for_phi_result,
22459 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22460 estimate_function_body_sizes, compute_inline_parameters,
22461 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22462 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22463 inline_update_callee_summaries, remap_edge_change_prob,
22464 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22465 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22466 inline_analyze_function, inline_summary_t::insert,
22467 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22468 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22469 inline_free_summary): Move to ipa-fnsummary.h
22470 (predicate_t): Remove.
22471 * ipa-fnsummary.c: New file.
22472 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22473 (enum inline_hints_vals, inline_hints, agg_position_info,
22474 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22475 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22476 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22477 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22478 inline_read_summary, inline_write_summary, inline_free_summary,
22479 inline_analyze_function, initialize_inline_failed,
22480 inline_merge_summary, inline_update_overall_summary,
22481 compute_inline_parameters): Move to ipa-fnsummary.h
22482 * ipa-fnsummary.h: New file.
22483 * ipa-inline-transform.h: Include ipa-inline.h.
22484 * ipa-inline.c: LIkewise.
22486 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22488 * ipa-inline.c (edge_badness): Use inlined_time instead of
22489 inline_summaries->get.
22491 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22493 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22495 2017-05-22 Nathan Sidwell <nathan@acm.org>
22497 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22498 (fdump-lang): Document 'raw' option.
22499 * dumpfile.h (TDI_tu): Delete.
22500 * dumpfile.c (dump_files): Remove translation-unit.
22501 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22503 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22505 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22506 command option from $(AWK) call.
22507 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22508 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22509 [FORMAT]: Remove handling of variable.
22510 * config/avr/t-multilib: Regenerate.
22512 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22514 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22516 (dump_inline_summary): Do not print self_time.
22517 (estimate_function_body_sizes): Do not set self_time.
22518 (compute_inline_parameters): Likewise.
22519 (inline_read_section, inline_write_summary): Do not stream self_time.
22520 * ipa-inline.h (inline_summary): Drop self_time.
22522 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22524 * ipa-inline-analysis.c (account_size_time): Rename to ...
22525 (inline_summary::account_size_time): ... this one.
22526 (reset_ipa_call_summary): Turn to ...
22527 (ipa_call_summary::reset): ... this one.
22528 (reset_inline_summary): Turn to ...
22529 (inline_summary::reset): ... this one.
22530 (inline_summary_t::remove): Update.
22531 (inline_summary_t::duplicate): Update.
22532 (ipa_call_summary_t::remove): Update.
22533 (dump_inline_summary): Update.
22534 (estimate_function_body_sizes): Update.
22535 (compute_inline_parameters): Update.
22536 (estimate_node_size_and_time): Update.
22537 (inline_merge_summary): Update.
22538 (inline_update_overall_summary): Update.
22539 (inline_read_section): Update.
22540 (inline_write_summary): Update.
22541 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22542 add account_size_time and reset member functions.
22543 (ipa_call_summary): Add reset function.
22544 * ipa-predicate.h (predicate::operator &): Constify.
22546 2017-05-22 Richard Biener <rguenther@suse.de>
22548 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22550 2017-05-19 Jason Merrill <jason@redhat.com>
22552 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22554 2017-05-19 Marek Polacek <polacek@redhat.com>
22557 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22558 TYPE_OVERFLOW_WRAPS checks.
22560 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22562 * tree-core.h (enum omp_clause_default_kind): Add
22563 "OMP_CLAUSE_DEFAULT_PRESENT".
22564 * tree-pretty-print.c (dump_omp_clause): Handle it.
22565 * gimplify.c (enum gimplify_omp_var_data): Add
22566 "GOVD_MAP_FORCE_PRESENT".
22567 (gimplify_adjust_omp_clauses_1): Map it to
22568 "GOMP_MAP_FORCE_PRESENT".
22569 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22571 * gimplify.c (oacc_default_clause): Clarify.
22573 2017-05-19 Nathan Sidwell <nathan@acm.org>
22575 LANG_HOOK_REGISTER_DUMPS
22576 * toplev.c (general_init): Call register dump lang hook.
22577 * doc/invoke.texi: Document -fdump-lang option family.
22578 * dumpfile.c (dump_files): Remove class dump here.
22579 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22580 * dumpfile.h (tree_dump_index): Remove TDI_class.
22581 * langhooks-def.h (lhd_register_dumps): Declare.
22582 (LANG_HOOKS_REGISTER_DUMPS): Define.
22583 (LANG_HOOKS_INITIALIZER): Add it.
22584 * langhooks.c (lhd_register_dumps): Define.
22585 * langhooks.h (struct lang_hooks): Add register_dumps.
22587 2017-05-19 Nathan Sidwell <nathan@acm.org>
22589 * context.h (context::set_passes): New.
22590 * context.c (context::context): Do not create pass manager.
22591 * toplev.c (general_init): Create pass manager here.
22593 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22595 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22596 use this splitter if two add or or instructions would also work for
22597 the constant we want to generate.
22599 2017-05-19 Richard Biener <rguenther@suse.de>
22602 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22603 predicate evaluation.
22605 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22607 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22609 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22612 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22614 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22615 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22616 (inline_edge_summary_vec): Turn into ...
22617 (ipa_call_summaries): ... this one.
22618 (redirect_to_unreachable, edge_set_predicate,
22619 evaluate_properties_for_edge, inline_summary_alloc,
22620 reset_ipa_call_summary, reset_inline_summary,
22621 inline_summary_t::duplicate): Update.
22622 (inline_edge_duplication_hook): Turn to ...
22623 (ipa_call_summary_t::duplicate): ... this one.
22624 (inline_edge_removal_hook): Turn to ...
22625 (ipa_call_summary_t::remove): ... this one.
22626 (dump_inline_edge_summary): Turn to ...
22627 (dump_ipa_call_summary): ... this one.
22628 (estimate_function_body_sizes): Update.
22629 (inline_update_callee_summaries): Update.
22630 (remap_edge_change_prob): Update.
22631 (remap_edge_summaries): Update.
22632 (inline_merge_summary): Update.
22633 (do_estimate_edge_time): Update.
22634 (inline_generate_summary): Update.
22635 (inline_read_section): Update.
22636 (inline_read_summary): Update.
22637 (inline_free_summary): Update.
22638 * ipa-inline.c (can_inline_edge_p): Update.
22639 (compute_inlined_call_time): Update.
22640 (want_inline_small_function_p): Update.
22641 (edge_badness): Update.
22642 (early_inliner): Update.
22643 * ipa-inline.h (inline_edge_summary): Turn to ...
22644 (ipa_call_summary): ... this one.
22645 (ipa_call_summary_t): New class.
22646 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22647 (ipa_call_summaries): New.
22648 (inline_edge_summary): Remove.
22649 (estimate_edge_growth): Update.
22650 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22651 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22652 * ipa-split.c (execute_split_functions): Update.
22653 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22655 2017-05-19 Richard Biener <rguenther@suse.de>
22657 PR middle-end/80764
22658 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22660 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22662 * config/rs6000/rs6000.c (struct machine_function): Add field
22663 fpr_is_wrapped_separately.
22664 (rs6000_get_separate_components): Use 64 components. Handle the
22665 new FPR components.
22666 (rs6000_components_for_bb): Handle the FPR components.
22667 (rs6000_emit_prologue_components): Handle the FPR components.
22668 (rs6000_emit_epilogue_components): Handle the FPR components.
22669 (rs6000_set_handled_components): Handle the FPR components.
22670 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22671 that are already separately shrink-wrapped.
22672 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
22673 that are already separately shrink-wrapped.
22675 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22678 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
22681 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
22682 (define_peephole2 for Altivec d-form load): Add peepholes to catch
22683 cases where the register allocator uses a move and an offsettable
22684 memory operation to/from a FPR register on ISA 2.06/2.07.
22685 (define_peephole2 for Altivec d-form store): Likewise.
22687 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22690 * config/i386/mmx.md (*mov<mode>_internal): Enable
22691 alternatives 11, 12, 13 and 14 also for 32bit targets.
22692 Remove alternatives 15, 16, 17 and 18.
22693 * config/i386/sse.md (vec_concatv2di): Change
22694 alternative (!x, *y) to (x, ?!*Yn).
22696 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22698 * dumpfile.h (enum dump_kind): Remove stray comma.
22700 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22702 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22703 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22704 predicate::num_conditions
22705 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22706 (CHANGED): turn into predicate::changed.
22707 (agg_position_info): Move to ipa-predicate.h
22708 (add_condition, predicate::add_clause, predicate::operator &=,
22709 predicate::or_with, predicate::evaluate, predicate::probability,
22710 dump_condition, dump_clause, predicate::dump,
22711 predicate::remap_after_duplication, predicate::remap_after_inlining,
22712 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22713 (evaluate_conditions_for_known_args): Update.
22714 (set_cond_stmt_execution_predicate): Update.
22715 * ipa-inline.h: Include ipa-predicate.h
22716 (condition, inline_param_summary, conditions, agg_position_info,
22717 predicate): Move to ipa-predicate.h
22718 * ipa-predicate.c: New file.
22719 * ipa-predicate.h: New file.
22721 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22723 * final.c (leaf_function_p): Check we are not in a sequence.
22725 2017-05-18 Martin Liska <mliska@suse.cz>
22727 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22728 * dumpfile.c (dump_register): Use new enum dump_kind.
22729 (get_dump_file_name): Likewise.
22730 (dump_enable_all): Likewise.
22731 (dump_switch_p_1): Likewise.
22732 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22733 * dumpfile.h (enum dump_kind): New enum type.
22734 (struct dump_file_info): Create constructor and
22735 format fields and comments.
22736 * passes.c (pass_manager::register_one_dump_file):
22738 * statistics.c (statistics_early_init): Likewise.
22739 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22740 TDF_TREE with TDF_SLIM.
22741 (gather_memory_references_ref): Likewise.
22743 2017-05-18 Martin Liska <mliska@suse.cz>
22745 * vec.h (struct vnull): Use it.
22747 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22749 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22750 (true_predicate, false_predicate, true_predicate_p,
22751 false_predicate_p): Remove.
22752 (single_cond_predicate, not_inlined_predicate): Turn to member function
22754 (add_condition): Update.
22755 (add_clause): Turn to...
22756 (predicate::add_clause): ... this one; update; allow passing NULL
22758 (and_predicates): Turn to ...
22759 (predicate::operator &=): ... this one.
22760 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22761 (or_predicates): Turn to ...
22762 (predicate::or_with): ... this one.
22763 (evaluate_predicate): Turn to ...
22764 (predicate::evaluate): ... this one.
22765 (predicate_probability): Turn to ...
22766 (predicate::probability): ... this one.
22767 (dump_condition): Update.
22768 (dump_predicate): Turn to ...
22769 (predicate::dump): ... this one.
22770 (account_size_time): Update.
22771 (edge_set_predicate): Update.
22772 (set_hint_predicate): UPdate.
22773 (evaluate_conditions_for_known_args): Update.
22774 (evaluate_properties_for_edge): Update.
22775 (remap_predicate_after_duplication): Turn to...
22776 (predicate::remap_after_duplication): ... this one.
22777 (remap_hint_predicate_after_duplication): Update.
22778 (inline_summary_t::duplicate): UPdate.
22779 (dump_inline_edge_summary): Update.
22780 (dump_inline_summary): Update.
22781 (set_cond_stmt_execution_predicate): Update.
22782 (set_switch_stmt_execution_predicate): Update.
22783 (compute_bb_predicates): Update.
22784 (will_be_nonconstant_expr_predicate): Update.
22785 (will_be_nonconstant_predicate): Update.
22786 (phi_result_unknown_predicate): Update.
22787 (predicate_for_phi_result): Update.
22788 (array_index_predicate): Update.
22789 (estimate_function_body_sizes): Update.
22790 (estimate_node_size_and_time): Update.
22791 (estimate_ipcp_clone_size_and_time): Update.
22792 (remap_predicate): Rename to ...
22793 (predicate::remap_after_inlining): ... this one.
22794 (remap_hint_predicate): Update.
22795 (inline_merge_summary): Update.
22796 (inline_update_overall_summary): Update.
22797 (estimate_size_after_inlining): Update.
22798 (read_predicate): Rename to ...
22799 (predicate::stream_in): ... this one.
22800 (read_inline_edge_summary): Update.
22801 (write_predicate): Rename to ...
22802 (predicate::stream_out): ... this one.
22803 (write_inline_edge_summary): Update.
22804 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22805 (clause_t): Turn to uint32_t
22806 (predicate): Turn to class; implement constructor and operators
22808 (size_time_entry): Update.
22809 (inline_summary): Update.
22810 (inline_edge_summary): Update.
22812 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22814 * fold-const.c (fold_binary_loc): Move transformation...
22815 * match.pd (C - X CMP X): ... here.
22817 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22819 * config/sparc/sparc.c (sparc_option_override): Set function
22820 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22821 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22823 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22825 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22827 2017-05-18 Marek Polacek <polacek@redhat.com>
22830 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22831 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22832 gimple_assign_load_p.
22834 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22836 PR middle-end/80692
22837 * real.c (do_compare): Give decimal_do_compare preference over
22838 comparing just the signs.
22840 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22842 * doc/md.texi (Canonicalization of Instructions): Describe the
22843 canonical form of instructions that inherently set a condition
22846 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22848 PR middle-end/80775
22849 * tree-cfg.c: Move deletion of unreachable case statements to after
22850 the merging of consecutive case labels.
22852 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22854 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22855 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22856 restoring of callee-saved registers.
22858 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22860 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22861 * config/visium/visium.c (single_set_and_flags): Likewise.
22862 * config/visium/visium.md (Substitutions): Likewise.
22864 2017-05-17 Martin Liska <mliska@suse.cz>
22866 * cfg.c: Introduce dump_flags_t type and
22867 use it instead of int type.
22869 * cfghooks.c: Likewise.
22870 * cfghooks.h (struct cfg_hooks): Likewise.
22871 * cfgrtl.c: Likewise.
22872 * cfgrtl.h: Likewise.
22873 * cgraph.c (cgraph_node::get_body): Likewise.
22874 * coretypes.h: Likewise.
22875 * domwalk.c: Likewise.
22876 * domwalk.h: Likewise.
22877 * dumpfile.c (struct dump_option_value_info): Likewise.
22878 (dump_enable_all): Likewise.
22879 (dump_switch_p_1): Likewise.
22880 (opt_info_switch_p): Likewise.
22881 * dumpfile.h (enum tree_dump_index): Likewise.
22882 (struct dump_file_info): Likewise.
22883 * genemit.c: Likewise.
22884 * generic-match-head.c: Likewise.
22885 * gengtype.c (open_base_files): Likewise.
22886 * gimple-pretty-print.c: Likewise.
22887 * gimple-pretty-print.h: Likewise.
22888 * graph.c (print_graph_cfg): Likewise.
22889 * graphite-scop-detection.c (dot_all_sese): Likewise.
22890 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22891 * loop-unroll.c (report_unroll): Likewise.
22892 * passes.c (pass_manager::register_one_dump_file): Likewise.
22893 * print-tree.c: Likewise.
22894 * statistics.c: Likewise.
22895 * tree-cfg.c: Likewise.
22896 * tree-cfg.h: Likewise.
22897 * tree-dfa.c: Likewise.
22898 * tree-dfa.h: Likewise.
22899 * tree-dump.c (dump_function): Likewise.
22900 * tree-dump.h (struct dump_info): Likewise.
22901 * tree-pretty-print.c: Likewise.
22902 * tree-pretty-print.h: Likewise.
22903 * tree-ssa-live.c: Likewise.
22904 * tree-ssa-live.h: Likewise.
22905 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22906 * tree-vect-loop.c: Likewise.
22907 * tree-vect-slp.c: Likewise.
22909 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22910 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22912 PR tree-optimization/80457
22913 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22914 of all arguments to a statement as scalar_to_vec operations.
22915 (vectorizable_call): Adjust call to vect_model_simple_cost for
22917 (vectorizable_conversion): Likewise.
22918 (vectorizable_assignment): Likewise.
22919 (vectorizable_shift): Likewise.
22920 (vectorizable_operation): Likewise.
22921 (vectorizable_comparison): Likewise.
22922 (vect_is_simple_cond): Record the def types for operands.
22923 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22924 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22925 for statement argument count.
22927 2017-05-16 Carl Love <cel@us.ibm.com>
22929 * config/rs6000/rs6000-c: Add support for built-in functions
22930 vector unsigned long long vec_bperm (vector unsigned long long,
22931 vector unsigned char)
22932 vector signed long long vec_mule (vector signed int,
22934 vector unsigned long long vec_mule (vector unsigned int,
22935 vector unsigned int)
22936 vector signed long long vec_mulo (vector signed int,
22938 vector unsigned long long vec_mulo (vector unsigned int,
22939 vector unsigned int)
22940 vector signed char vec_sldw (vector signed char,
22941 vector signed char,
22943 vector unsigned char vec_sldw (vector unsigned char,
22944 vector unsigned char,
22946 vector signed short vec_sldw (vector signed short,
22947 vector signed short,
22949 vector unsigned short vec_sldw (vector unsigned short,
22950 vector unsigned short,
22952 vector signed int vec_sldw (vector signed int,
22955 vector unsigned int vec_sldw (vector unsigned int,
22956 vector unsigned int,
22958 vector signed long long vec_sldw (vector signed long long,
22959 vector signed long long,
22961 vector unsigned long long vec_sldw (vector unsigned long long,
22962 vector unsigned long long,
22964 * config/rs6000/rs6000-c: Add support for built-in functions
22965 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22966 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22967 * doc/extend.texi: Update the built-in documentation for the
22968 new built-in functions.
22970 2017-05-16 Marek Polacek <polacek@redhat.com>
22974 * tree.c (save_expr): Don't fold the expression.
22976 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22978 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22979 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22980 and (?*y,m). Update insn attributes.
22982 2017-05-16 Martin Liska <mliska@suse.cz>
22984 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
22985 flags argument of print_gimple_stmt, print_gimple_expr,
22986 print_generic_stmt and print_generic_expr.
22987 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
22988 * coretypes.h: Likewise.
22989 * except.c (dump_eh_tree): Likewise.
22990 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
22991 * gimple-pretty-print.h: Likewise.
22992 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
22993 (backprop::push_to_worklist): Likewise.
22994 (backprop::pop_from_worklist): Likewise.
22995 (backprop::process_use): Likewise.
22996 (backprop::intersect_uses): Likewise.
22997 (note_replacement): Likewise.
22998 * gimple-ssa-store-merging.c
22999 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23000 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23001 (pass_store_merging::execute): Likewise.
23002 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23003 (ssa_base_cand_dump_callback): Likewise.
23004 (dump_incr_vec): Likewise.
23005 (replace_refs): Likewise.
23006 (replace_mult_candidate): Likewise.
23007 (create_add_on_incoming_edge): Likewise.
23008 (create_phi_basis): Likewise.
23009 (insert_initializers): Likewise.
23010 (all_phi_incrs_profitable): Likewise.
23011 (introduce_cast_before_cand): Likewise.
23012 (replace_one_candidate): Likewise.
23013 * gimplify.c (gimplify_expr): Likewise.
23014 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23015 (set_rename): Likewise.
23016 (rename_uses): Likewise.
23017 (copy_loop_phi_nodes): Likewise.
23018 (add_close_phis_to_merge_points): Likewise.
23019 (copy_loop_close_phi_args): Likewise.
23020 (copy_cond_phi_args): Likewise.
23021 (graphite_copy_stmts_from_block): Likewise.
23022 (translate_pending_phi_nodes): Likewise.
23023 * graphite-poly.c (print_pdr): Likewise.
23024 (dump_gbb_cases): Likewise.
23025 (dump_gbb_conditions): Likewise.
23026 (print_scop_params): Likewise.
23027 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23028 (build_cross_bb_scalars_use): Likewise.
23029 (gather_bbs::before_dom_children): Likewise.
23030 * hsa-dump.c (dump_hsa_immed): Likewise.
23031 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23032 (get_replacement_map): Likewise.
23033 * ipa-inline-analysis.c (dump_condition): Likewise.
23034 (estimate_function_body_sizes): Likewise.
23035 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23036 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23037 * ipa-prop.c (ipa_dump_param): Likewise.
23038 (ipa_print_node_jump_functions_for_edge): Likewise.
23039 (ipa_modify_call_arguments): Likewise.
23040 (ipa_modify_expr): Likewise.
23041 (ipa_dump_param_adjustments): Likewise.
23042 (ipa_dump_agg_replacement_values): Likewise.
23043 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23044 * ipa-pure-const.c (check_stmt): Likewise.
23045 (pass_nothrow::execute): Likewise.
23046 * ipa-split.c (execute_split_functions): Likewise.
23047 * omp-offload.c (dump_oacc_loop_part): Likewise.
23048 (dump_oacc_loop): Likewise.
23049 * trans-mem.c (tm_log_emit): Likewise.
23050 (tm_memopt_accumulate_memops): Likewise.
23051 (dump_tm_memopt_set): Likewise.
23052 (dump_tm_memopt_transform): Likewise.
23053 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23054 (print_loop): Likewise.
23055 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23056 (chkp_gather_checks_info): Likewise.
23057 (chkp_get_check_result): Likewise.
23058 (chkp_remove_check_if_pass): Likewise.
23059 (chkp_use_outer_bounds_if_possible): Likewise.
23060 (chkp_reduce_bounds_lifetime): Likewise.
23061 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23062 (chkp_mark_completed_bounds): Likewise.
23063 (chkp_register_incomplete_bounds): Likewise.
23064 (chkp_mark_invalid_bounds): Likewise.
23065 (chkp_maybe_copy_and_register_bounds): Likewise.
23066 (chkp_build_returned_bound): Likewise.
23067 (chkp_get_bound_for_parm): Likewise.
23068 (chkp_build_bndldx): Likewise.
23069 (chkp_get_bounds_by_definition): Likewise.
23070 (chkp_generate_extern_var_bounds): Likewise.
23071 (chkp_get_bounds_for_decl_addr): Likewise.
23072 * tree-chrec.c (chrec_apply): Likewise.
23073 * tree-data-ref.c (dump_data_reference): Likewise.
23074 (dump_subscript): Likewise.
23075 (dump_data_dependence_relation): Likewise.
23076 (analyze_overlapping_iterations): Likewise.
23077 * tree-inline.c (expand_call_inline): Likewise.
23078 (tree_function_versioning): Likewise.
23079 * tree-into-ssa.c (dump_defs_stack): Likewise.
23080 (dump_currdefs): Likewise.
23081 (dump_names_replaced_by): Likewise.
23082 (dump_update_ssa): Likewise.
23083 (update_ssa): Likewise.
23084 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23085 * tree-parloops.c (build_new_reduction): Likewise.
23086 (try_create_reduction_list): Likewise.
23087 (ref_conflicts_with_region): Likewise.
23088 (oacc_entry_exit_ok_1): Likewise.
23089 (oacc_entry_exit_single_gang): Likewise.
23090 * tree-pretty-print.h: Likewise.
23091 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23092 (get_scalar_evolution): Likewise.
23093 (add_to_evolution): Likewise.
23094 (get_loop_exit_condition): Likewise.
23095 (analyze_evolution_in_loop): Likewise.
23096 (analyze_initial_condition): Likewise.
23097 (analyze_scalar_evolution): Likewise.
23098 (instantiate_scev): Likewise.
23099 (number_of_latch_executions): Likewise.
23100 (gather_chrec_stats): Likewise.
23101 (final_value_replacement_loop): Likewise.
23102 (scev_const_prop): Likewise.
23103 * tree-sra.c (dump_access): Likewise.
23104 (disqualify_candidate): Likewise.
23105 (create_access): Likewise.
23106 (reject): Likewise.
23107 (maybe_add_sra_candidate): Likewise.
23108 (create_access_replacement): Likewise.
23109 (analyze_access_subtree): Likewise.
23110 (analyze_all_variable_accesses): Likewise.
23111 (sra_modify_assign): Likewise.
23112 (initialize_constant_pool_replacements): Likewise.
23113 (find_param_candidates): Likewise.
23114 (decide_one_param_reduction): Likewise.
23115 (replace_removed_params_ssa_names): Likewise.
23116 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23117 * tree-ssa-copy.c (dump_copy_of): Likewise.
23118 (copy_prop_visit_cond_stmt): Likewise.
23119 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23120 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23121 (record_equivalences_from_stmt): Likewise.
23122 * tree-ssa-dse.c (compute_trims): Likewise.
23123 (delete_dead_call): Likewise.
23124 (delete_dead_assignment): Likewise.
23125 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23126 (forward_propagate_into_cond): Likewise.
23127 (pass_forwprop::execute): Likewise.
23128 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23129 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23131 (move_computations_worker): Likewise.
23132 (execute_sm): Likewise.
23133 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23134 (remove_exits_and_undefined_stmts): Likewise.
23135 (remove_redundant_iv_tests): Likewise.
23136 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23137 (adjust_iv_update_pos): Likewise.
23138 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23139 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23140 (value_replacement): Likewise.
23141 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23142 * tree-ssa-pre.c (print_pre_expr): Likewise.
23143 (get_representative_for): Likewise.
23144 (create_expression_by_pieces): Likewise.
23145 (insert_into_preds_of_block): Likewise.
23146 (eliminate_insert): Likewise.
23147 (eliminate_dom_walker::before_dom_children): Likewise.
23148 (eliminate): Likewise.
23149 (remove_dead_inserted_code): Likewise.
23150 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23151 * tree-ssa-reassoc.c (get_rank): Likewise.
23152 (eliminate_duplicate_pair): Likewise.
23153 (eliminate_plus_minus_pair): Likewise.
23154 (eliminate_not_pairs): Likewise.
23155 (undistribute_ops_list): Likewise.
23156 (eliminate_redundant_comparison): Likewise.
23157 (update_range_test): Likewise.
23158 (optimize_range_tests_var_bound): Likewise.
23159 (optimize_vec_cond_expr): Likewise.
23160 (rewrite_expr_tree): Likewise.
23161 (rewrite_expr_tree_parallel): Likewise.
23162 (linearize_expr): Likewise.
23163 (break_up_subtract): Likewise.
23164 (linearize_expr_tree): Likewise.
23165 (attempt_builtin_powi): Likewise.
23166 (attempt_builtin_copysign): Likewise.
23167 (transform_stmt_to_copy): Likewise.
23168 (transform_stmt_to_multiply): Likewise.
23169 (dump_ops_vector): Likewise.
23170 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23171 (print_scc): Likewise.
23172 (set_ssa_val_to): Likewise.
23173 (visit_reference_op_store): Likewise.
23174 (visit_use): Likewise.
23175 (sccvn_dom_walker::before_dom_children): Likewise.
23176 (run_scc_vn): Likewise.
23177 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
23179 (expr_hash_elt::print): Likewise.
23180 (const_and_copies::pop_to_marker): Likewise.
23181 (const_and_copies::record_const_or_copy_raw): Likewise.
23182 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
23183 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
23184 (dump_predicates): Likewise.
23185 (find_uninit_use): Likewise.
23186 (warn_uninitialized_phi): Likewise.
23187 (pass_late_warn_uninitialized::execute): Likewise.
23188 * tree-ssa.c (verify_vssa): Likewise.
23189 (verify_ssa): Likewise.
23190 (maybe_optimize_var): Likewise.
23191 * tree-vrp.c (dump_value_range): Likewise.
23192 (dump_all_value_ranges): Likewise.
23193 (dump_asserts_for): Likewise.
23194 (register_edge_assert_for_2): Likewise.
23195 (vrp_visit_cond_stmt): Likewise.
23196 (vrp_visit_switch_stmt): Likewise.
23197 (vrp_visit_stmt): Likewise.
23198 (vrp_visit_phi_node): Likewise.
23199 (simplify_cond_using_ranges_1): Likewise.
23200 (fold_predicate_in): Likewise.
23201 (evrp_dom_walker::before_dom_children): Likewise.
23202 (evrp_dom_walker::push_value_range): Likewise.
23203 (evrp_dom_walker::pop_value_range): Likewise.
23204 (execute_early_vrp): Likewise.
23206 2017-05-16 Richard Biener <rguenther@suse.de>
23208 * dwarf2out.c (loc_list_from_tree_1): Do not create
23209 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
23211 2017-05-16 Richard Biener <rguenther@suse.de>
23213 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
23215 (note_variable_value_in_expr): If we resolved the decl ref
23216 do not push to the stack.
23218 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
23220 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
23221 operations in fast-math mode.
23222 (vaddq_f16): Likewise.
23223 (vmul_f16): Likewise.
23224 (vmulq_f16): Likewise.
23225 (vsub_f16): Likewise.
23226 (vsubq_f16): Likewise.
23227 * config/arm/neon.md (add<mode>3): New.
23229 (fma:<VH:mode>3): New. Also remove outdated comment.
23232 2017-05-16 Martin Liska <mliska@suse.cz>
23236 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23237 (odr_types_equivalent_p): Likewise.
23239 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23241 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23243 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23246 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23247 non-interunit SSE move alternatives with '?'.
23248 (zero-extendsidi peephole2): New peephole to skip intermediate
23249 general register in SSE zero-extend sequence.
23251 2017-05-15 Jeff Law <law@redhat.com>
23253 * reorg.c (relax_delay_slots): Create a new variable to hold
23254 the temporary target rather than clobbering TARGET_LABEL.
23256 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23257 missing argument to extract_bit_field call.
23258 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23260 2017-05-15 Martin Liska <mliska@suse.cz>
23263 * gcc.c (process_command): Do not allow empty argument of -o option.
23265 2017-05-15 Renlin Li <renlin.li@arm.com>
23267 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23268 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23269 * config/aarch64/constraints.md (Usf): Add long call check.
23270 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23271 (call_value): Likewise.
23272 (sibcall): Likewise.
23273 (sibcall_value): Likewise.
23275 (call_value_insn): New.
23276 (sibcall_insn): Update rtx pattern.
23277 (sibcall_value_insn): Likewise.
23278 (call_internal): Remove.
23279 (call_value_internal): Likewise.
23280 (sibcall_internal): Likewise.
23281 (sibcall_value_internal): Likewise.
23282 (call_reg): Likewise.
23283 (call_symbol): Likewise.
23284 (call_value_reg): Likewise.
23285 (call_value_symbol): Likewise.
23287 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23290 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23292 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23294 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23295 compatible with CCGOCmode and with CCZmode.
23297 2017-05-14 Martin Sebor <msebor@redhat.com>
23299 PR middle-end/77671
23300 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23301 (gimple_fold_builtin_snprintf): Same.
23302 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23303 (gimple_fold_builtin_snprintf): Same.
23304 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23305 of character types.
23306 (is_call_safe): New function.
23307 (try_substitute_return_value): Call it.
23308 (try_simplify_call): New function.
23309 (pass_sprintf_length::handle_gimple_call): Call it.
23311 2017-05-14 Martin Sebor <msebor@redhat.com>
23313 PR middle-end/80669
23314 * builtins.c (expand_builtin_stpncpy): Simplify.
23316 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23318 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23319 * config/i386/i386.h
23320 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23321 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23322 (struct machine_function): Add new members call_ms2sysv,
23323 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23324 (struct machine_frame_state): New fields sp_realigned and
23325 sp_realigned_offset.
23326 * config/i386/i386.c
23327 (enum xlogue_stub): New enum.
23328 (enum xlogue_stub_sets): New enum.
23329 (class xlogue_layout): New class.
23330 (struct ix86_frame): New fields stack_realign_allocate_offset,
23331 stack_realign_offset and outlined_save_offset. Modify comments to
23332 detail stack layout when using out-of-line stubs.
23333 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23334 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23335 -mcall-ms2sysv-xlogues.
23336 (stub_managed_regs): New static variable.
23337 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23338 registers managed by out-of-line stub.
23339 (disable_call_ms2sysv_xlogues): New function.
23340 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23341 m->call_ms2sysv when appropriate and compute frame layout for
23343 (sp_valid_at, fp_valid_at): New inline functions.
23344 (choose_basereg): New function.
23345 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23347 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23348 Use align parameter of choose_baseaddr to generated aligned SSE movs
23350 (pro_epilogue_adjust_stack): Modify to track
23351 machine_frame_state::sp_realigned.
23352 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23353 (ix86_nsaved_sseregs): Likewise.
23354 (ix86_emit_save_regs): Likewise.
23355 (ix86_emit_save_regs_using_mov): Likewise.
23356 (ix86_emit_save_sse_regs_using_mov): Likewise.
23357 (get_scratch_register_on_entry): Likewise.
23358 (gen_frame_set): New function.
23359 (gen_frame_load): Likewise.
23360 (gen_frame_store): Likewise.
23361 (emit_outlined_ms2sysv_save): Likewise.
23362 (emit_outlined_ms2sysv_restore): Likewise.
23363 (ix86_expand_prologue): Modify stack re-alignment code and call
23364 emit_outlined_ms2sysv_save when appropriate.
23365 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23366 parameter rtx_insn *insn, which allows the function to be used to only
23367 generate the notes.
23368 (ix86_expand_epilogue): Modify validity checks of frame and stack
23369 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23370 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23371 * config/i386/predicates.md
23372 (save_multiple): New predicate.
23373 (restore_multiple): Likewise.
23374 * config/i386/sse.md
23375 (save_multiple<mode>): New pattern.
23376 (save_multiple_realign<mode>): Likewise.
23377 (restore_multiple<mode>): Likewise.
23378 (restore_multiple_and_return<mode>): Likewise.
23379 (restore_multiple_leave_return<mode>): Likewise.
23380 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23382 2017-05-14 Julia Koval <julia.koval@intel.com>
23384 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23385 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23386 (__builtin_ia32_xsetbv): New builtins.
23387 * config/i386/i386.c (ix86_expand_special_args_builtin):
23389 (ix86_expand_builtin): Special expand for new intrinsics.
23390 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23391 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23392 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23394 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23396 * cfganal.c (inverted_post_order_compute): Change argument type
23398 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23399 * df-core.c (rest_of_handle_df_initialize): Adjust.
23400 (rest_of_handle_df_finish): Likewise.
23401 (df_analyze_1): Likewise.
23402 (df_analyze): Likewise.
23403 (loop_inverted_post_order_compute): Change argument to be a vec *.
23404 (df_analyze_loop): Adjust.
23405 (df_get_n_blocks): Likewise.
23406 (df_get_postorder): Likewise.
23407 * df.h (struct df_d): Change field to be a vec.
23408 * lcm.c (compute_laterin): Adjust.
23409 (compute_available): Likewise.
23410 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23411 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23412 * tree-ssa-pre.c (compute_antic): Likewise.
23414 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23416 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23417 (depth_first_search::depth_first_search): Change structure init
23418 function to this constructor.
23419 (depth_first_search::add_bb): Rename function to this member.
23420 (depth_first_search::execute): Likewise.
23421 (flow_dfs_compute_reverse_finish): Adjust.
23423 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23425 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23426 (longest_simple_path): Likewise.
23427 * shrink-wrap.c (spread_components): Likewise.
23428 (disqualify_problematic_components): Likewise.
23429 (emit_common_heads_for_components): Likewise.
23430 (emit_common_tails_for_components): Likewise.
23431 (insert_prologue_epilogue_for_components): Likewise.
23433 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23435 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23438 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23440 * df-core.c (df_set_blocks): Start using auto_bitmap.
23441 (df_compact_blocks): Likewise.
23442 * df-problems.c (df_rd_confluence_n): Likewise.
23443 * df-scan.c (df_insn_rescan_all): Likewise.
23444 (df_process_deferred_rescans): Likewise.
23445 (df_update_entry_block_defs): Likewise.
23446 (df_update_exit_block_uses): Likewise.
23447 (df_entry_block_bitmap_verify): Likewise.
23448 (df_exit_block_bitmap_verify): Likewise.
23449 (df_scan_verify): Likewise.
23450 * lra-constraints.c (lra_constraints): Likewise.
23451 (undo_optional_reloads): Likewise.
23452 (lra_undo_inheritance): Likewise.
23453 * lra-remat.c (calculate_gen_cands): Likewise.
23454 (do_remat): Likewise.
23455 * lra-spills.c (assign_spill_hard_regs): Likewise.
23456 (spill_pseudos): Likewise.
23457 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23458 (bitmap_set_subtract_values): Likewise.
23460 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23462 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23463 management with auto_bitmap.
23464 (fix_inter_tick): Likewise.
23465 (fix_recovery_deps): Likewise.
23466 * ira.c (add_store_equivs): Likewise.
23467 (find_moveable_pseudos): Likewise.
23468 (split_live_ranges_for_shrink_wrap): Likewise.
23469 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23470 (rtx_reuse_manager::seen_def_p): Likewise.
23471 (rtx_reuse_manager::set_seen_def): Likewise.
23472 * print-rtl.h (class rtx_reuse_manager): Likewise.
23474 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23476 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23478 (migrate_btr_def): Likewise.
23479 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23480 * df-core.c (loop_post_order_compute): Likewise.
23481 (loop_inverted_post_order_compute): Likewise.
23482 * hsa-common.h: Likewise.
23483 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23484 * init-regs.c (initialize_uninitialized_regs): Likewise.
23485 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23486 (inline_small_functions): Likewise.
23487 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23488 * ira.c (combine_and_move_insns): Likewise.
23489 (build_insn_chain): Likewise.
23490 * loop-invariant.c (find_invariants): Likewise.
23491 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23492 * predict.c (tree_predict_by_opcode): Likewise.
23493 (predict_paths_leading_to): Likewise.
23494 (predict_paths_leading_to_edge): Likewise.
23495 (estimate_loops_at_level): Likewise.
23496 (estimate_loops): Likewise.
23497 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23498 (spread_components): Likewise.
23499 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23500 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23501 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23502 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23503 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23504 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23505 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23506 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23507 (mark_threaded_blocks): Likewise.
23508 (thread_through_all_blocks): Likewise.
23509 * tree-ssa.c (verify_ssa): Likewise.
23510 (execute_update_addresses_taken): Likewise.
23511 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23513 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23515 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23517 (post_order_compute): Likewise.
23518 (inverted_post_order_compute): Likewise.
23519 (pre_and_rev_post_order_compute_fn): Likewise.
23521 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23523 * genrecog.c (int_set::int_set): Explicitly construct our
23524 auto_vec base class.
23525 * vec.h (auto_vec::auto_vec): New constructor.
23527 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23529 * bitmap.h (class auto_bitmap): New constructor taking
23530 bitmap_obstack * argument.
23532 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23534 * bitmap.h (class auto_bitmap): Change type of m_bits to
23535 bitmap_head, and adjust ctor / dtor and member operators.
23537 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23539 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23540 when returned register mode doesn't match original mode.
23542 2017-05-12 Jeff Law <law@redhat.com>
23543 Jakub Jelinek <jakub@redhat.com>
23545 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23546 we look for cc setter after the compare-elim changes.
23547 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23548 within the vector to match what compare-elim now expects.
23549 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23550 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23552 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23553 after the compare-elim changes.
23554 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23555 the vector to match what compare-elim now expects.
23556 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23557 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23558 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23559 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23560 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23562 * config/visium/visium.c (single_set_and_flags): Fix where
23563 we look for cc setter after the compare-elim changes.
23564 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23565 with the vector to match what compare-elim now expects.
23566 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23567 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23568 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23569 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23570 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23571 (neg<mode>2_insn_set_overflow): Likewise.
23573 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23575 PR middle-end/79794
23576 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23577 maybe_expand_insn call, set ops[0].target. If still set after call,
23578 set alt_rtl. Add extra arg to recursive calls.
23579 (extract_bit_field): Add alt_rtl argument. Pass to
23581 * expmed.h (extract_bit_field): Fix prototype.
23582 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23583 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23584 to extract_bit_field_calls.
23585 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23586 Pass alt_rtl to extract_bit_field calls.
23587 * calls.c (store_unaligned_arguments_into_psuedos)
23588 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23589 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23591 * optabs.h (struct expand_operand): Add target bitfield.
23593 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23595 * compare-elim.c (try_eliminate_compare): Canonicalize
23596 operation with embedded compare to
23597 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23598 (set (reg) (operation)].
23600 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23602 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23605 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23606 cost of adding a carry flag for ADC instruction.
23607 [case MINUS]: Ignore the cost of subtracting a carry flag
23608 for SBB instruction.
23610 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23612 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23614 * config/rs6000/bmiintrin.h: New file.
23615 * config/rs6000/bmi2intrin.h: New file.
23616 * config/rs6000/x86intrin.h: New file.
23618 2017-05-12 Jeff Law <law@redhat.com>
23620 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23623 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23625 PR middle-end/80707
23626 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23628 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23630 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23631 early expansion of vector divide builtins.
23632 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23633 builtins identified as having unsigned arguments.
23635 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23637 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23638 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23639 expansion of vector logical operations (and, andc, or, xor,
23642 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23644 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23645 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23647 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23649 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23650 early GIMPLE expansion of vector multiplies.
23652 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23654 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23655 TARGET_HAVE_MOVT conditional.
23656 (movt splitter): Likewise.
23658 2017-05-12 Richard Biener <rguenther@suse.de>
23660 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23661 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23662 Fold all stmts not inplace.
23664 2017-05-12 Richard Biener <rguenther@suse.de>
23666 PR tree-optimization/80713
23667 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23668 inserted_exprs bit for not removed stmts.
23670 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23672 PR middle-end/69921
23673 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
23674 parallelized" attribute for parallelized OpenACC kernels.
23675 * omp-offload.c (execute_oacc_device_lower): Use it.
23677 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
23678 Set "oacc kernels" attribute.
23679 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
23680 parameter. Adjust all users.
23681 (oacc_fn_attrib_kernels_p): Remove function.
23682 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
23683 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
23684 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
23685 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
23686 assert "oacc kernels" attribute is set.
23688 2017-05-11 Carl Love <cel@us.ibm.com>
23690 * config/rs6000/rs6000-c: Add support for built-in functions
23691 vector unsigned char vec_popcnt (vector signed char)
23692 vector unsigned char vec_popcnt (vector unsigned char)
23693 vector unsigned short vec_popcnt (vector signed short)
23694 vector unsigned short vec_popcnt (vector unsigned short)
23695 vector unsigned int vec_popcnt (vector signed int)
23696 vector unsigned int vec_popcnt (vector unsigned int)
23697 vector unsigned long long vec_popcnt (vector signed long long)
23698 vector unsigned long long vec_popcnt (vector unsigned long long)
23699 vector signed long long vec_slo (vector signed long long,
23700 vector signed char)
23701 vector signed long long vec_slo (vector signed long long,
23702 vector unsigned char)
23703 vector unsigned long long vec_slo (vector unsigned long long,
23704 vector signed char)
23705 vector unsigned long long vec_slo (vector unsigned long long,
23706 vector unsigned char)
23707 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23708 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23709 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23710 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23711 * doc/extend.texi: Update the built-in documentation file for the
23712 new built-in functions.
23714 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23716 * attribs.h (sorted_attr_string): Move machine independent
23717 functions for target clone support from the i386 port to common
23718 code. Rename ix86_function_versions to common_function_versions.
23719 Rename make_name to make_unique_name.
23720 (common_function_versions): Likewise.
23721 (make_unique_name): Likewise.
23722 (make_dispatcher_decl): Likewise.
23723 (is_function_default_version): Likewise.
23724 * attribs.c (attr_strcmp): Likewise.
23725 (sorted_attr_string): Likewise.
23726 (common_function_versions): Likewise.
23727 (make_unique_name): Likewise.
23728 (make_dispatcher_decl): Likewise.
23729 (is_function_default_version): Likewise.
23730 * config/i386/i386.c (attr_strcmp): Likewise.
23731 (sorted_attr_string): Likewise.
23732 (ix86_function_versions): Likewise.
23733 (make_name): Likewise.
23734 (make_dispatcher_decl): Likewise.
23735 (is_function_default_version): Likewise.
23736 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23738 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23741 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23742 Account for direct move costs for vec_construct of integer
23745 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23748 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23749 (UNSPEC_STX_ATOMIC): Ditto.
23750 (loaddi_via_sse): New insn.
23751 (storedi_via_sse): Ditto.
23752 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23753 Update corresponding peephole2 patterns.
23754 (atomic_storedi_fpu): Ditto.
23756 2017-05-11 Julia Koval <julia.koval@intel.com>
23758 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23759 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23761 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23762 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23763 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23765 2017-05-11 Nathan Sidwell <nathan@acm.org>
23767 * graphite-poly.c: Include dumpfile.h.
23769 * dumpfle.h (dump_function): Declare here ...
23770 * tree-dump.h (dump_function): ... not here.
23771 * dumpfile.c: #include tree-cfg.h.
23772 (dump_function): Move here from ...
23773 * tree-dump.c (dump_function): ... here.
23774 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23775 * graphite-poly.c: Don't include tree-dump.h.
23776 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23777 * print-tree.c: Likewise.
23778 * stor-layout.c: Likewise.
23779 * tree-nested.c: Likewise.
23781 * dumpfile.c (dump_start): Use TDF_FLAGS.
23782 (dump_enable_all): Fix TDF_KIND check thinko.
23784 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23786 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23787 array entries to represent two legal parameterizations of the
23788 overloaded __builtin_cmpb function, as represented by the
23789 P6_OV_BUILTIN_CMPB constant.
23790 (altivec_resolve_overloaded_builtin): Add special case handling
23791 for the __builtin_cmpb function, as represented by the
23792 P6_OV_BUILTIN_CMPB constant.
23793 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23794 (BU_P6_64BIT_2): New macro.
23795 (BU_P6_OVERLOAD_2): New macro
23796 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23797 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23798 (CMPB): Add overload support to represent both 32-bit and 64-bit
23799 compare-bytes function.
23800 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23801 support for TARGET_CMPB.
23802 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23803 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23804 documentation of the __builtin_cmpb overloaded built-in function.
23806 2017-05-11 Richard Biener <rguenther@suse.de>
23808 PR tree-optimization/80705
23809 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23810 bases are not vectorizable.
23812 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23814 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23815 when counting register pressure.
23817 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23819 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23820 (iv_ca_more_deps): Renamed to ...
23821 (iv_ca_compare_deps): ... this.
23822 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23824 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23826 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23828 (determine_group_iv_costs): ... here.
23829 (find_inv_vars_cb): Record inv var if it's not recorded before.
23831 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23833 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23834 (get_shiftadd_cost): Ditto.
23836 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23838 * tree-ssa-address.c: Include header file.
23839 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23841 (add_to_parts): Refactor.
23842 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23843 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23846 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23848 PR tree-optimization/53090
23849 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23851 (extract_cond_operands): Detect condition with IV on both sides
23852 and return COMP_IV_EXPR_2.
23853 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23854 (rewrite_use_compare): Simplify by removing call to function
23855 extract_cond_operands.
23857 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23859 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23860 (extract_cond_operands): Detect condition comparing against non-
23861 invariant bound and return appropriate enum value.
23862 (find_interesting_uses_cond): Update use of extract_cond_operands.
23863 Handle its return value accordingly.
23864 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23866 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23868 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23869 nonlinear iv_use computation in loop invariant sensitive way.
23871 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23873 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23874 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23876 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23878 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23879 (dump_cand): Support iv_cand.inv_exprs.
23880 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23882 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23885 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23887 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23889 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23890 as local function. Include necessary header files.
23891 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23893 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23895 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23897 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23899 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23900 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23901 RSHIFT_EXPR and BIT_NOT_EXPR.
23903 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23905 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23906 (adjust_setup_cost): New parameter supporting round up adjustment.
23907 (struct address_cost_data): Delete.
23908 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23909 (split_address_cost, ptr_difference_cost): Delete.
23910 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23911 (struct ainc_cost_data): New struct.
23912 (get_address_cost_ainc): New function.
23913 (get_address_cost, get_computation_cost): Reimplement.
23914 (determine_group_iv_cost_address): Record inv_expr for all uses of
23916 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23917 (iv_ca_has_deps): Reimplemented to ...
23918 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23920 (iv_ca_extend): Call iv_ca_more_deps.
23922 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23924 * tree-ssa-address.c (struct mem_address): Move to header file.
23925 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23926 * tree-ssa-address.h (struct mem_address): Move from C file.
23927 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23929 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23931 * tree-affine.h (aff_combination_type): New interface.
23932 (aff_combination_zero_p): Remove static.
23933 (aff_combination_const_p): New interface.
23934 (aff_combination_singleton_var_p): New interfaces.
23936 2017-05-11 Richard Biener <rguenther@suse.de>
23938 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23939 Skip unreachable blocks and destinations.
23940 (eliminate): Move stmt removal and fixup ...
23941 (fini_eliminate): ... here. Skip inserted exprs.
23942 (pass_pre::execute): Move fini_pre after fini_eliminate.
23943 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23944 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23945 PRE to get rid of dead code that has invalid SSA form and
23946 split critical edges again.
23948 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23950 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23952 2017-05-11 Richard Biener <rguenther@suse.de>
23954 * passes.c (execute_function_todo): Verify loops if they are
23955 said to be up-to-date.
23956 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23957 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23959 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23962 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23963 handle calling assemble_external ourself.
23966 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23967 modes with zero size. Enhance comment.
23969 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23971 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23972 built-ins for vec_xl and vec_xst with short and char pointer
23975 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23977 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23978 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23979 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23980 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23981 (_mm_maskz_min_round_ss): New intrinsics.
23982 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23983 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23984 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
23985 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
23986 (__builtin_ia32_minss_mask_round): New builtins.
23987 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23988 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23989 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
23991 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
23992 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
23994 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
23997 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23999 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24000 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24001 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24002 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24003 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24004 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24005 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24006 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24007 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24008 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24009 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24010 (__builtin_ia32_mulss_mask_round): New builtins.
24011 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24012 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24013 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24015 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24016 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24018 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24021 2017-05-10 Julia Koval <julia.koval@intel.com>
24023 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24024 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24025 (_mm256_setr_m128i): New intrinsics.
24027 2017-05-10 Julia Koval <julia.koval@intel.com>
24029 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24030 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24031 (_mm_maskz_rcp14_ss): New intrinsics.
24032 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24033 (__builtin_ia32_rcp14ss_mask): New builtins.
24034 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24036 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24038 PR tree-optimization/51513
24039 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24040 (assert_unreachable_fallthru_edge_p): Use it.
24041 (group_case_labels_stmt): Likewise.
24042 * tree-cfg.h: Prototype it.
24043 * stmt.c: Include cfghooks.h and tree-cfg.h.
24044 (emit_case_dispatch_table) <gap_label>: New local variable.
24045 Use it to fill dispatch table gaps.
24046 Test for default_label before updating probabilities.
24047 (expand_case) <default_label>: Remove unneeded initialization.
24048 Test for unreachable default case statement and remove its edge.
24049 Set default_label accordingly.
24050 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24052 2017-05-10 Carl Love <cel@us.ibm.com>
24054 * config/rs6000/rs6000-c: Add support for built-in functions
24055 vector signed char vec_neg (vector signed char)
24056 vector signed short int vec_neg (vector short int)
24057 vector signed int vec_neg (vector signed int)
24058 vector signed long long vec_neg (vector signed long long)
24059 vector float vec_neg (vector float)
24060 vector double vec_neg (vector double)
24061 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24063 * config/rs6000/altivec.h: Add define for vec_neg
24064 * doc/extend.texi: Update the built-in documentation for the
24065 new built-in functions.
24067 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24069 PR tree-optimization/77644
24070 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24072 2017-05-10 Nathan Sidwell <nathan@acm.org>
24074 * dumpfile.h (TDI_lang_all): New.
24075 (TDF_KIND): New. Renumber others
24076 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24078 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24080 (get_dump_file_name): Adjust suffix generation.
24081 (dump_enable_all): Use TDF_KIND.
24082 * doc/invoke.texi (-fdump-lang-all): Document.
24084 * dumpfile.h: Tabify.
24086 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24089 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24090 Move member access before delete.
24092 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24094 * tree-inline.c (expand_call_inline): Split block at stmt
24097 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24100 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24101 are now unused after splitting mov{sf,sd}_hardfloat.
24102 (f32_lr2): Likewise.
24103 (f32_lm): Likewise.
24104 (f32_lm2): Likewise.
24105 (f32_li): Likewise.
24106 (f32_li2): Likewise.
24107 (f32_lv): Likewise.
24108 (f32_sr): Likewise.
24109 (f32_sr2): Likewise.
24110 (f32_sm): Likewise.
24111 (f32_sm2): Likewise.
24112 (f32_si): Likewise.
24113 (f32_si2): Likewise.
24114 (f32_sv): Likewise.
24115 (f32_dm): Likewise.
24116 (f32_vsx): Likewise.
24117 (f32_av): Likewise.
24118 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24119 For movsf, order stores so the VSX stores occur before the GPR
24120 store which encourages the register allocator to use a traditional
24121 FPR instead of a GPR. For movsd, order the stores so that the GPR
24122 store comes before the VSX stores to allow the power6 to work.
24123 This is due to the power6 not having a 32-bit integer store
24124 instruction from a FPR.
24125 (movsf_hardfloat): Likewise.
24126 (movsd_hardfloat): Likewise.
24128 2017-05-09 Martin Sebor <msebor@redhat.com>
24130 PR translation/80280
24131 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24134 PR translation/80280
24135 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24136 data member added in r247778.
24137 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24139 2017-05-09 Nathan Sidwell <nathan@acm.org>
24141 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24143 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24146 2017-05-09 Marek Polacek <polacek@redhat.com>
24148 * doc/invoke.texi: Fix typo.
24150 2017-05-09 Richard Biener <rguenther@suse.de>
24152 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24153 (vrp_val_is_min): Likewise.
24154 (set_value_range_to_value): Likewise.
24155 (set_value_range_to_nonnegative): Likewise.
24156 (gimple_assign_nonzero_p): Likewise.
24157 (gimple_stmt_nonzero_p): Likewise.
24158 (vrp_int_const_binop): Likewise. Remove unreachable case.
24159 (adjust_range_with_scev): Adjust comments.
24160 (compare_range_with_value): Likewise.
24161 (extract_range_from_phi_node): Likewise.
24162 (test_for_singularity): Likewise.
24164 2017-05-09 Richard Biener <rguenther@suse.de>
24166 * tree-vrp.c (get_single_symbol): Add assert that we don't
24167 get overflowed constants as invariant part.
24168 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24169 checks. Use wi::cmp instead of recursing for integer constants.
24170 (compare_values): Just ignore whether we assumed undefined
24171 overflow instead of failing the compare.
24172 (extract_range_for_var_from_comparison_expr): Add comment before the
24173 TREE_NO_WARNING sets.
24174 (test_for_singularity): Likewise.
24175 (extract_range_from_comparison): Do not disable optimization
24176 when we assumed undefined overflow.
24177 (extract_range_basic): Remove init of unused var.
24179 2017-05-09 Richard Biener <rguenther@suse.de>
24181 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
24182 (extract_range_from_multiplicative_op_1): Adjust.
24183 (extract_range_from_binary_expr_1): Use int_const_binop.
24185 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
24188 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
24189 rs6000_store_data_bypass_p in seven define_bypass directives and
24190 in several comments.
24191 * config/rs6000/rs6000-protos.h: Add prototype for
24192 rs6000_store_data_bypass_p function.
24193 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
24194 function implements slightly different (rs6000-specific) semantics
24195 than store_data_bypass_p, returning false rather than aborting
24196 with assertion error when arguments do not satisfy the
24197 requirements of store data bypass.
24198 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
24199 rs6000_store_data_bypass_p.
24201 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
24203 * config/xtensa/xtensa-protos.h
24204 (xtensa_initial_elimination_offset): New declaration.
24205 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24206 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
24207 macro definition, add case for FRAME_POINTER_REGNUM when
24208 FRAME_GROWS_DOWNWARD.
24209 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
24210 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
24211 xtensa_initial_elimination_offset.
24213 2017-05-08 Nathan Sidwell <nathan@acm.org>
24215 * doc/invoke.texi: Alphabetize -fdump options.
24217 2017-05-08 Martin Sebor <msebor@redhat.com>
24219 PR translation/80280
24220 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
24222 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
24224 * target.def (compute_frame_layout): New optional target hook.
24225 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
24226 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
24227 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
24229 * reload1.c (verify_initial_elim_offsets): Likewise.
24230 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
24231 (use_simple_return_p): Call arm_compute_frame_layout if needed.
24232 (arm_get_frame_offsets): Split up into this ...
24233 (arm_compute_frame_layout): ... and this function.
24235 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24237 * config/aarch64/constraints.md (Usa): New constraint.
24238 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24240 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24242 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24243 with_multilib_list after it has been checked.
24245 2017-05-08 Richard Biener <rguenther@suse.de>
24247 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24248 (bitmap_set_subtract_values): Likewise.
24250 2017-05-08 Richard Biener <rguenther@suse.de>
24252 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24253 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24255 (gimple_stmt_nonzero_warnv_p): Rename to ...
24256 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24258 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24259 (extract_range_basic): Adjust, do not disable propagation on
24260 strict overflow sensitive simplification.
24261 (vrp_visit_cond_stmt): Likewise.
24263 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24265 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24266 body size unconditionally.
24268 2017-05-07 Jeff Law <law@redhat.com>
24271 2017-05-06 Jeff Law <law@redhat.com>
24272 PR tree-optimization/78496
24273 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24276 PR tree-optimization/78496
24277 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24278 (simplify_stmt_using_ranges): Call it.
24279 (vrp_dom_walker::before_dom_children): Extract equivalences
24280 from an ASSERT_EXPR with an equality comparison against a
24283 2017-05-06 Jeff Law <law@redhat.com>
24285 PR tree-optimization/78496
24286 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24289 PR tree-optimization/78496
24290 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24291 (simplify_stmt_using_ranges): Call it.
24292 (vrp_dom_walker::before_dom_children): Extract equivalences
24293 from an ASSERT_EXPR with an equality comparison against a
24296 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24298 * lra-constraints.c (lra_copy_reg_equiv): New function.
24299 (split_reg): Use it to copy equivalence information from the
24300 original register to the spill register.
24302 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24304 PR rtl-optimization/75964
24305 * simplify-rtx.c (simplify_const_relational_operation): Remove
24306 invalid handling of comparisons of integer ABS.
24308 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24310 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24311 initialize to zero.
24312 (init_regs): Remove declaration.
24313 (function_arg_advance_32): Initialize error_p as boolean variable.
24315 2017-05-05 Nathan Sidwell <nathan@acm.org>
24317 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24318 lines. Use for (;;).
24320 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24322 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24323 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24324 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24325 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24326 VF=2 that require versioning.
24328 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24330 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24333 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24335 * diagnostic.h (diagnostic_override_option_index): Convert from
24336 macro to inline function.
24338 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24340 * diagnostic.c (last_module_changed_p): New function.
24341 (set_last_module): New function.
24342 (diagnostic_report_current_module): Convert macro usage to
24343 the above functions.
24344 * diagnostic.h (diagnostic_context::last_module): Strengthen
24345 from const line_map * to const line_map_ordinary *.
24346 (diagnostic_last_module_changed): Delete macro.
24347 (diagnostic_set_last_module): Delete macro.
24349 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24351 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24352 with diagnostic_report_diagnostic.
24353 (diagnostic_n_impl_richloc): Likewise.
24354 * diagnostic.h (report_diagnostic): Delete macro.
24355 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24356 with diagnostic_report_diagnostic.
24357 * substring-locations.c (format_warning_va): Likewise.
24359 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24361 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24362 save/restor of format_spec. Move option-printing code to...
24363 (print_option_information): ...this new function, and
24364 reimplement by simply printing to the pretty_printer,
24365 rather than appending to the format string.
24367 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24369 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24370 handling logic into...
24371 (update_effective_level_from_pragmas): ...this new function.
24373 2017-05-04 Andrew Waterman <andrew@sifive.com>
24375 * config/riscv/riscv.opt (mstrict-align): New option.
24376 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24377 (SLOW_UNALIGNED_ACCESS): Define.
24378 (riscv_slow_unaligned_access): Declare.
24379 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24381 (riscv_slow_unaligned_access): New variable.
24382 (rocket_tune_info): Set slow_unaligned_access to true.
24383 (optimize_size_tune_info): Set slow_unaligned_access to false.
24384 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24385 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24386 (riscv_option_override): Set riscv_slow_unaligned_access.
24387 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24389 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24391 * config/riscv/riscv.md: Unify indentation.
24393 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24398 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24400 (extendsi<mode>2): Add support for doing sign extension via
24401 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24402 don't have ISA 3.0 instructions.
24403 (extendsi<mode>2 splitter): Likewise.
24404 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24405 generate the normal insns since SImode can now go in vector
24406 registers. Disallow the special UNSPECs needed for previous
24407 machines to hide SImode being used. Add new insns
24408 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24409 (fix_trunc<mode>si2_stfiwx): Likewise.
24410 (fix_trunc<mode>si2_internal): Likewise.
24411 (fixuns_trunc<mode>si2): Likewise.
24412 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24413 (fctiw<u>z_<mode>_smallint): Likewise.
24414 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24415 of floating point to 32-bit integer from doing a direct move to
24416 the GPR registers to do a store.
24417 (fctiwz_<mode>): Break long line.
24419 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24421 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24422 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24423 (addr_list, addr_offset_valid_p): New.
24424 (split_address_groups): Check offset validity with above function.
24425 (gt-tree-ssa-loop-ivopts.h): Include header file.
24427 2017-05-05 Nathan Sidwell <nathan@acm.org>
24429 * config.gcc (arm*-*-*): Add missing 'fi'.
24431 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24433 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24434 included in -fopt-info does not matter.
24435 * doc/optinfo.texi (-fopt-info): Fix description of default
24436 behavour. Explicitly say order of options included in -fopt-info
24439 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24441 * config.gcc: Allow combinations of aprofile and rmprofile values for
24442 --with-multilib-list.
24443 * config/arm/t-multilib: New file.
24444 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24445 variables. Remove setting of ISA and floating-point ABI in
24446 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24447 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24448 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24449 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24451 * config/arm/t-rmprofile: Likewise except for the matches changes.
24452 * doc/install.texi (--with-multilib-list): Document the combination of
24453 aprofile and rmprofile values and warn about pitfalls in doing that.
24455 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24457 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24458 (movdi_aarch64): Likewise.
24460 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24462 PR tree-optimization/80632
24463 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24465 (build_arrays): Initialize it for virtual phis.
24466 (fix_phi_nodes): Use it for virtual phis.
24468 PR tree-optimization/80558
24469 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24470 [x, y] op z into [x op, y op z] for op & or | if conditions
24473 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24474 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24477 * config/arm/arm.md (use_literal_pool): Remove.
24478 (64-bit immediate split): No longer takes cost into consideration
24479 if arm_disable_literal_pool is enabled.
24480 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24481 used when arm_disable_literal_pool is enabled.
24482 (arm_max_const_double_inline_cost): Remove use of
24483 arm_disable_literal_pool.
24484 (push_minipool_fix): Add assert.
24485 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24486 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24487 (no_literal_pool_sf_immediate): New.
24489 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24491 PR tree-optimization/80613
24492 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24493 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24495 2017-05-05 Richard Biener <rguenther@suse.de>
24497 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24499 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24501 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24502 of this flag from insn conditions due to removal from r247495.
24504 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24506 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24508 (arm_early_store_addr_dep_ptr): Likewise.
24509 * config/arm/aarch-common-protos.h
24510 (arm_early_load_addr_dep_ptr): Add prototype.
24511 (arm_early_store_addr_dep_ptr): Likewise.
24512 * config/arm/cortex-a53.md: Add new bypasses.
24514 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24516 * tree.c (next_type_uid): Change type to unsigned.
24517 (type_hash_canon): Decrement back next_type_uid if
24518 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24519 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24522 2017-05-04 Martin Sebor <msebor@redhat.com>
24524 * builtins.c: Fix a trivial typo in a comment.
24526 PR middle-end/79234
24527 * builtins.c (check_sizes): Adjust to handle reading past the end.
24528 Avoid printing excessive upper bound of ranges. Use %E to print
24529 tree nodes instead of converting them to %wu.
24530 (expand_builtin_memchr): New function.
24531 (compute_dest_size): Rename...
24532 (compute_objsize): ...to this.
24533 (expand_builtin_memcpy): Adjust.
24534 (expand_builtin_mempcpy): Adjust.
24535 (expand_builtin_strcat): Adjust.
24536 (expand_builtin_strcpy): Adjust.
24537 (check_strncat_sizes): Adjust.
24538 (expand_builtin_strncat): Adjust.
24539 (expand_builtin_strncpy): Adjust and simplify.
24540 (expand_builtin_memset): Adjust.
24541 (expand_builtin_bzero): Adjust.
24542 (expand_builtin_memcmp): Adjust.
24543 (expand_builtin): Handle memcmp.
24544 (maybe_emit_chk_warning): Check strncat just once.
24546 2017-05-04 Martin Sebor <msebor@redhat.com>
24548 PR preprocessor/79214
24549 PR middle-end/79222
24550 PR middle-end/79223
24551 * builtins.c (check_sizes): Add inlining context and issue
24552 warnings even when -Wno-system-headers is set.
24553 (check_strncat_sizes): Same.
24554 (expand_builtin_strncat): Same.
24555 (expand_builtin_memmove): New function.
24556 (expand_builtin_stpncpy): Same.
24557 (expand_builtin): Handle memmove and stpncpy.
24559 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24561 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24562 which is not used any more.
24564 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24566 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24568 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24570 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24571 (cortexa53_tunings): Likewise.
24572 (cortexa57_tunings): Likewise.
24573 (cortexa72_tunings): Likewise.
24574 (cortexa73_tunings): Likewise.
24576 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24578 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24579 Set loop alignment to 8.
24581 2017-05-04 Martin Sebor <msebor@redhat.com>
24583 PR translation/80280
24584 * builtins.c (expand_builtin_object_size): Add missing quoting to
24585 %D and like directives.
24586 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24587 (hsa_type_for_tree_type): Same.
24588 (verify_function_arguments): Same.
24589 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24590 * varasm.c (get_section): Same.
24593 2017-05-04 Martin Sebor <msebor@redhat.com>
24595 PR translation/80280
24596 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24598 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24600 * config/aarch64/aarch64.c (generic_addrcost_table):
24601 Change HI/TI mode setting.
24603 2017-05-04 Martin Jambor <mjambor@suse.cz>
24605 PR tree-optimization/80622
24606 * tree-sra.c (comes_initialized_p): New function.
24607 (build_accesses_from_assign): Only set write lazily when
24608 comes_initialized_p is false.
24609 (analyze_access_subtree): Use comes_initialized_p.
24610 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24611 instead of testing for PARM_DECL.
24613 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24615 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24616 constraint on operand 0 to allow more general addressing modes.
24617 Adjust output template.
24618 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24620 * config/aarch64/aarch64-protos.h
24621 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24622 * config/aarch64/constraints.md (Dp): New address constraint.
24623 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24626 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24628 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24629 update use of estimate_ipcp_clone_size_and_time.
24630 (estimate_local_effects): Update use of
24631 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24632 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24633 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24634 Return nonspecialized time.
24636 2017-05-04 Richard Biener <rguenther@suse.de>
24638 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24639 for the last VUSE which def dominates the PHI. Directly call
24641 (get_continuation_for_phi_1): Remove.
24643 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24645 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24646 to explain the use of truncating division. Cap the number of
24647 iterations to the maximum given by nb_iterations_upper_bound,
24650 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24652 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24653 * configure: Regenerate.
24654 * config.in: Regenerate.
24655 * config/i386/driver-mingw32.c: new file.
24656 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24657 * config.host: Link driver-mingw32.o on MinGW host.
24658 * doc/install.texi: Document new --enable-mingw-wildcard configure
24661 2017-05-04 Marek Polacek <polacek@redhat.com>
24663 PR tree-optimization/80612
24664 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24666 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24667 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24669 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24670 (movt splitter): Likewise.
24671 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24672 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
24673 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
24674 block for Thumb-1 with MOVT.
24675 (thumb2_legitimate_address_p): Move code block ...
24676 (can_avoid_literal_pool_for_label_p): ... into this new function.
24677 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
24679 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
24680 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
24681 "M-profile targets with the MOVT instruction".
24683 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24685 * config/arm/arm-builtins.c (arm_init_builtins): Rename
24686 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
24687 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24689 2017-05-04 Martin Liska <mliska@suse.cz>
24691 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24692 variable cond_code.
24694 2017-05-04 Richard Biener <rguenther@suse.de>
24696 * tree.c (array_at_struct_end_p): Handle arrays at struct
24697 end with flexarrays more conservatively. Refactor and treat
24698 arrays of arrays or aggregates more strict. Fix
24699 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24700 * tree.c (array_at_struct_end_p): Adjust prototype.
24701 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24702 * gimple-fold.c (get_range_strlen): Likewise.
24703 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24705 2017-05-04 Richard Biener <rguenther@suse.de>
24707 PR tree-optimization/31130
24708 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24710 (supports_overflow_infinity): Likewise.
24711 (is_negative_overflow_infinity): Likewise.
24712 (is_positive_overflow_infinity): Likewise.
24713 (is_overflow_infinity): Likewise.
24714 (stmt_overflow_infinity): Likewise.
24715 (overflow_infinity_range_p): Likewise.
24716 (usable_range_p): Remove as always returning true.
24717 (make_overflow_infinity): Remove.
24718 (negative_overflow_infinity): Likewise.
24719 (positive_overflow_infinity): Likewise.
24720 (avoid_overflow_infinity): Likewise.
24721 (set_value_range): Adjust accordingly.
24722 (set_value_range_to_nonnegative): Likewise, remove now unused
24723 overflow_infinity arg.
24724 (vrp_operand_equal_p): Adjust.
24725 (update_value_range): Likewise.
24726 (range_int_cst_singleton_p): Likewise.
24727 (operand_less_p): Likewise.
24728 (compare_values_warnv): Likewise.
24729 (extract_range_for_var_from_comparison_expr): Likewise.
24730 (vrp_int_const_binop): Likewise.
24731 (zero_nonzero_bits_from_vr): Likewise.
24732 (extract_range_from_multiplicative_op_1): Likewise.
24733 (extract_range_from_binary_expr_1): Likewise.
24734 (extract_range_from_unary_expr): Likewise.
24735 (extract_range_from_comparison): Likewise.
24736 (extract_range_basic): Likewise.
24737 (adjust_range_with_scev): Likewise.
24738 (compare_ranges): Likewise.
24739 (compare_range_with_value): Likewise.
24740 (dump_value_range): Likewise.
24741 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24743 (simplify_cond_using_ranges): Adjust.
24745 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24747 * brig-builtins.def: Added a builtin for class_f64.
24748 * builtin-types.def: Added a builtin type needed by class_f64.
24750 2017-05-03 Jason Merrill <jason@redhat.com>
24752 * timevar.def: Add TV_CONSTEXPR.
24754 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24756 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24758 2017-05-03 Martin Jambor <mjambor@suse.cz>
24760 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24761 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24762 * ipa-cp.c (ipcp_propagate_stage): Do not call
24763 ipa_update_after_lto_read.
24764 * ipa-inline.c (ipa_inline): Likewise.
24766 2017-05-03 Martin Jambor <mjambor@suse.cz>
24768 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24769 tag. Added a default constructor and a destructor.
24770 (ipa_edge_args_sum_t): New class;
24771 (ipa_edge_args_sum): Declare.
24772 (ipa_edge_args_vector): Remove declaration.
24773 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24774 (ipa_free_edge_args_substructures): Remove declaration.
24775 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24776 (ipa_edge_args_info_available_for_edge_p): Likewise.
24777 * ipa-prop.c (ipa_edge_args_vector): Removed.
24778 (edge_removal_hook_holder): Likewise.
24779 (edge_duplication_hook_holder): Likewise.
24780 (ipa_edge_args_sum): New variable.
24781 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24782 ipa_edge_args_vector.
24783 (ipa_free_edge_args_substructures): Likewise.
24784 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24785 ipa_edge_args_vector.
24786 (ipa_edge_removal_hook): Turned into method
24787 ipa_edge_args_sum_t::remove.
24788 (ipa_edge_duplication_hook): Turned into method
24789 ipa_edge_args_sum_t::duplicate.
24790 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24791 registering edge hooks.
24792 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24793 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24794 ipa_edge_args_sum instead of ipa_edge_args_vector.
24795 * ipa-profile.c (ipa_profile): Likewise.
24797 2017-05-03 Martin Jambor <mjambor@suse.cz>
24799 * symbol-summary.h (function_summary): New method exists.
24800 (function_summary::symtab_removal): Deallocate through release.
24801 (call_summary): New class.
24802 (gt_ggc_mx): New overload.
24803 (gt_pch_nx): Likewise.
24804 (gt_pch_nx): Likewise.
24806 2017-05-03 Jeff Law <law@redhat.com>
24808 PR tree-optimization/78496
24809 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24810 from simplify_cond_using_ranges. Split off code to walk
24811 backwards through casts into ...
24812 (simplify_cond_using_ranges_2): New function.
24813 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24814 (execute_vrp): After identifying jump threads, call
24815 simplify_cond_using_ranges_2.
24817 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24820 * ipa-inline.h (inline_summary): Add ctor.
24821 (create_ggc): Do not use ggc_cleared_alloc.
24823 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24824 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24826 * gcc.c (handle_braces): Support escaping in switch matching
24828 * doc/invoke.texi (Spec Files): Document it.
24829 Remove superfluous @code markup in items.
24831 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24833 * diagnostic-show-locus.c (struct column_range): New struct.
24834 (get_affected_columns): New function.
24835 (get_printed_columns): New function.
24836 (struct correction): New struct.
24837 (correction::ensure_capacity): New function.
24838 (correction::ensure_terminated): New function.
24839 (struct line_corrections): New struct.
24840 (line_corrections::~line_corrections): New dtor.
24841 (line_corrections::add_hint): New function.
24842 (layout::print_trailing_fixits): Reimplement in terms of the new
24844 (selftest::test_overlapped_fixit_printing): New function.
24845 (selftest::diagnostic_show_locus_c_tests): Call it.
24847 2017-05-03 Nathan Sidwell <nathan@acm.org>
24849 Canonicalize canonical type hashing
24850 * tree.h (type_hash_canon_hash): Declare.
24851 * tree.c (type_hash_list, attribute_hash_list): Move into
24852 type_hash_canon_hash.
24853 (build_type_attribute_qual_variant): Break out hash code calc into
24854 type_hash_canon_hash.
24855 (type_hash_canon_hash): New. Generic type hash computation.
24856 (build_range_type_1, build_array_type_1, build_function_type,
24857 build_method_type_directly, build_offset_type, build_complex_type,
24858 make_vector_type): Call it.
24860 2017-05-03 Richard Biener <rguenther@suse.de>
24862 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24863 When all DRs have unknown misaligned do not always peel
24864 when there is a store but apply the same costing model as if
24865 there were only loads.
24867 2017-05-03 Richard Biener <rguenther@suse.de>
24870 PR tree-optimization/80492
24871 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24872 compare_base_decls returning dont-know properly.
24874 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24876 * config/arm/iterators.md (CCSI): New mode iterator.
24877 (arch): New mode attribute.
24878 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24879 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24880 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24881 code iterator for success result mode.
24882 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24883 the corresponding new insn generators.
24885 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24888 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24889 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24891 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24893 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24894 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24895 (DDR_A): Wrap DDR argument in brackets.
24896 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24897 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24898 (DDR_REVERSED_P): Likewise.
24900 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24902 PR tree-optimization/79472
24903 * tree-switch-conversion.c (struct switch_conv_info): Add
24904 contiguous_range and default_case_nonstandard fields.
24905 (collect_switch_conv_info): Compute contiguous_range and
24906 default_case_nonstandard fields, don't clear final_bb if
24907 contiguous_range and only the default case doesn't have the required
24909 (check_all_empty_except_final): Set default_case_nonstandard instead
24910 of failing if contiguous_range and the default case doesn't have empty
24912 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24913 and only the default case doesn't have the required constants. Skip
24915 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24916 if default_case_nonstandard.
24917 (build_constructors): Build constant 1 just once. Assert that default
24918 values aren't inserted in between cases if contiguous_range. Skip
24920 (build_arrays): Skip virtual phis.
24921 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24922 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24923 Handle virtual phis.
24924 (gen_inbound_check): Handle default_case_nonstandard case.
24925 (process_switch): Adjust check_final_bb caller. Call
24926 gather_default_values with the first non-default case instead of
24927 default case if default_case_nonstandard.
24929 2017-05-02 Nathan Sidwell <nathan@acm.org>
24931 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24932 check. Fix formatting.
24934 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24936 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24937 errors when comparing specialized and unspecialized times.
24939 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24941 * diagnostic-show-locus.c
24942 (layout::should_print_annotation_line_p): Make private.
24943 (layout::print_annotation_line): Make private.
24944 (layout::annotation_line_showed_range_p): Make private.
24945 (layout::show_ruler): Make private.
24946 (layout::print_source_line): Make private. Pass in line and
24947 line_width, rather than calling location_get_source_line. Drop
24949 (layout::print_leading_fixits): New method.
24950 (layout::print_any_fixits): Rename to...
24951 (layout::print_trailing_fixits): ...this, and make private.
24952 Don't print newline fixits.
24953 (diagnostic_show_locus): Move logic for printing one row into...
24954 (layout::print_line): ...this new function. Move the
24955 location_get_source_line call and error-handling from
24956 print_source_line to here. Call print_leading_fixits, and rename
24957 print_any_fixits to print_trailing_fixits.
24958 (selftest::test_fixit_insert_containing_newline): Update now that
24959 newlines are partially supported.
24960 (selftest::test_fixit_insert_containing_newline_2): New test.
24961 (selftest::test_fixit_replace_containing_newline): Update comments.
24962 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24963 * edit-context.c (class added_line): New class.
24964 (class edited_line): Describe newline handling in comment.
24965 (edited_line::actually_edited_p): New method.
24966 (edited_line::print_content): Delete redundant decl.
24967 (edited_line::m_predecessors): New field.
24968 (edited_file::print_content): Call edited_line::print_content.
24969 (edited_file::print_diff): Update to support newlines.
24970 (edited_file::print_diff_hunk): Likewise.
24971 (edited_file::print_run_of_changed_lines): New function.
24972 (edited_file::print_diff_line): Convert to...
24973 (print_diff_line): ...this.
24974 (edited_file::get_effective_line_count): New function.
24975 (edited_line::edited_line): Initialize new field m_predecessors.
24976 (edited_line::~edited_line): Clean up m_predecessors.
24977 (edited_line::apply_fixit): Handle newlines.
24978 (edited_line::get_effective_line_count): New function.
24979 (edited_line::print_content): New function.
24980 (edited_line::print_diff_lines): New function.
24981 (selftest::test_applying_fixits_insert_containing_newline): New
24983 (selftest::test_applying_fixits_replace_containing_newline): New
24985 (selftest::insert_line): New function.
24986 (selftest::test_applying_fixits_multiple_lines): Add example of
24988 (selftest::edit_context_c_tests): Call the new tests.
24990 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24992 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
24993 parameter cand. Update dump information.
24994 (get_computation_cost): Update uses.
24996 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24998 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
24999 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25000 (get_computation_at, rewrite_use_address): Update use of
25001 get_computation_aff.
25003 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25005 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25006 (get_computation): Delete.
25007 (get_computation_cost): Implement like get_computation_cost_at.
25008 Use get_computation_at.
25009 (get_computation_cost_at): Delete.
25010 (rewrite_use_nonlinear_expr): Use get_computation_at.
25011 (rewrite_use_compare, remove_unused_ivs): Ditto.
25013 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25015 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25017 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25019 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25020 (ivopts_global_cost_for_size): Rename parameter and update uses.
25021 (iv_ca_recount_cost): Update uses.
25022 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25023 candidates seperately in n_invs and n_cands.
25024 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25026 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25028 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25029 (find_inv_vars_cb): New.
25030 (find_depends): Renamed to ...
25031 (find_inv_vars): ... this.
25032 (add_candidate_1, force_var_cost): Call find_inv_vars.
25033 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25035 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25037 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25038 inv_vars. Add inv_exprs.
25039 (struct iv_cand): Rename depends_on to inv_vars.
25040 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25041 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25042 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25043 (dump_cand): Dump inv_vars.
25044 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25045 (record_invariant, find_depends, add_candidate_1): Ditto.
25046 (set_group_iv_cost, force_var_cost): Ditto.
25047 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25048 (get_computation_cost_at, get_computation_cost): Ditto.
25049 (determine_group_iv_cost_generic): Ditto.
25050 (determine_group_iv_cost_address): Ditto.
25051 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25052 (determine_group_iv_costs): Ditto.
25053 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25054 (iv_ca_set_remove_invariants): Renamed to ...
25055 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25056 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25057 (iv_ca_set_add_invariants): Renamed to ...
25058 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25059 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25060 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25061 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25062 (create_new_ivs): Remove useless dump.
25064 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25066 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25068 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25069 (iv_ca_set_no_cp, create_new_iv): Ditto.
25071 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25073 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25075 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25077 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25078 function tree_check2.
25080 2017-05-02 Martin Liska <mliska@suse.cz>
25082 * doc/gcov.texi: Add missing preposition.
25083 * gcov.c (function_info::function_info): Properly fill up
25084 all member variables.
25086 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25088 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25090 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25092 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25094 2017-05-02 Martin Liska <mliska@suse.cz>
25097 * lto-streamer-in.c (lto_read_tree_1): Remove
25098 LTO_STREAMER_DEBUG.
25099 * lto-streamer.c (struct tree_hash_entry): Likewise.
25100 (struct tree_entry_hasher): Likewise.
25101 (tree_entry_hasher::hash): Likewise.
25102 (tree_entry_hasher::equal): Likewise.
25103 (lto_streamer_init): Likewise.
25104 (lto_orig_address_map): Likewise.
25105 (lto_orig_address_get): Likewise.
25106 (lto_orig_address_remove): Likewise.
25107 * lto-streamer.h: Likewise.
25108 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25109 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25111 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25113 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25114 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25115 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25116 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25117 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25118 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25119 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25120 (mm_maskz_sub_ss): New intrinsics.
25121 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25122 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25123 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25124 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25125 (__builtin_ia32_subss_mask_round): New builtins.
25126 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25127 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25128 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25130 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25131 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25133 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25136 2017-05-02 Martin Jambor <mjambor@suse.cz>
25138 PR tree-optimization/78687
25139 * tree-sra.c (access): New field parent.
25140 (process_subtree_disqualification): New function.
25141 (disqualify_candidate): Call it.
25142 (build_accesses_from_assign): Reset write flag if creating an
25144 (build_access_subtree): Fill in parent field and also prpagate
25145 down grp_write flag.
25146 (create_artificial_child_access): New parameter set_grp_write, set
25147 grp_write to its value.
25148 (propagate_subaccesses_across_link): Also propagate grp_write flag
25150 (propagate_all_subaccesses): Push the closest parent back to work
25151 queue if add_access_to_work_queue returned true.
25153 2017-05-02 Richard Biener <rguenther@suse.de>
25155 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25156 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25157 -fstrict-overflow documentation.
25158 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25159 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25160 flag_strict_overflow.
25161 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25162 * lto-opts.c (lto_write_options): Do not stream it.
25163 * lto-wrapper.c (merge_and_complain): Do not handle it.
25164 * opts.c (default_options_table): Do not set -fstrict-overflow.
25165 (finish_options): Likewise do not clear it when sanitizing.
25166 * simplify-rtx.c (simplify_const_relational_operation): Do not
25167 test flag_strict_overflow.
25169 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25171 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25172 using enabled attribute.
25173 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
25174 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
25175 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
25176 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
25177 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
25178 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
25179 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
25180 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
25181 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
25182 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
25184 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25186 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
25188 2017-05-02 Richard Biener <rguenther@suse.de>
25190 PR tree-optimization/80591
25192 2017-04-10 Richard Biener <rguenther@suse.de>
25194 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25197 2017-05-02 Richard Biener <rguenther@suse.de>
25199 PR tree-optimization/80549
25200 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
25201 (cleanup_tree_cfg_noloop): Create forwarders to known loop
25202 headers if they do not have a preheader.
25204 2017-05-02 Martin Liska <mliska@suse.cz>
25207 * common.opt: Fix typo.
25208 * doc/invoke.texi: Likewise.
25210 2017-05-01 Jan Beulich <jbeulich@suse.com>
25212 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
25213 swapping, add (x,x,m,x,n) alternative.
25215 2017-05-01 Nathan Sidwell <nathan@acm.org>
25217 * calls.c (combine_pending_stack_adjustment_and_call): Remove
25218 unnecessary unadjusted_alignment check.
25220 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25223 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
25225 * gimplify.c (gimplify_cilk_detach): New function.
25226 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
25227 * tree-core.h: Document EXPR_CILK_SPAWN.
25228 * tree.h (EXPR_CILK_SPAWN): Define.
25230 2017-05-01 David Malcolm <dmalcolm@redhat.com>
25232 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
25233 to use new fixit_hint representation, using the "replace" logic.
25234 (get_line_span_for_fixit_hint): Likewise.
25235 (layout::print_any_fixits): Likewise.
25236 (selftest::test_one_liner_many_fixits): Rename to...
25237 (selftest::test_one_liner_many_fixits_1): ...this, and update
25238 comment and expected output to reflect that the multiple fix-it
25239 hints are now consolidated into one insertion.
25240 (selftest::test_one_liner_many_fixits_2): New test.
25241 (selftest::test_diagnostic_show_locus_one_liner): Update for
25243 (selftest::test_fixit_consolidation): Update for fix-it API
25245 * diagnostic.c (print_parseable_fixits): Likewise.
25246 * edit-context.c (edited_line::m_line_events): Convert from
25247 auto_vec <line_event *> to auto_vec <line_event>.
25248 (class line_event): Convert from abstract base class to a concrete
25249 class, taking over the role of replace_event.
25250 (class insert_event): Delete.
25251 (class replace_event): Rename to class line_event. Convert to
25253 (edit_context::add_fixits): Reimplement.
25254 (edit_context::apply_insert): Delete.
25255 (edit_context::apply_replace): Rename to...
25256 (edit_context::apply_fixit): ...this. Convert to half-open range.
25257 (edited_file::apply_insert): Delete.
25258 (edited_file::apply_replace): Rename to...
25259 (edited_file::apply_fixit): ...this.
25260 (edited_line::~edited_line): Drop deletion of events.
25261 (edited_line::apply_insert): Delete.
25262 (edited_line::apply_replace): Rename to...
25263 (edited_line::apply_fixit): ...this. Convert to half-open range.
25264 Update for change to type of m_line_events.
25265 * edit-context.h (edit_context::apply_insert): Delete.
25266 (edit_context::apply_replace): Rename to...
25267 (edit_context::apply_fixit): ...this.
25269 2017-05-01 Martin Sebor <msebor@redhat.com>
25271 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25274 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25277 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25278 __get_cpuid_max returns 0.
25279 (__get_cpuid_count): Ditto.
25281 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25283 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25284 replacement expression is another instance of one of its arguments.
25286 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25289 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25290 check for stack push/pop autoinc.
25291 * config/i386/i386.c (ix86_agi_dependent): Return false
25292 if the only reason why modified_in_p returned true is that
25293 addr is SP based and set_insn is a push or pop.
25295 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25297 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25300 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25303 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25304 (account_size_time): Use two predicates - exec_pred and
25306 (evaluate_conditions_for_known_args): Compute both clause and
25308 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25309 (inline_summary_t::duplicate): Update.
25310 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25312 (compute_inline_parameters): Likewise.
25313 (estimate_edge_size_and_time): Update caluclation of time.
25314 (estimate_node_size_and_time): Compute both time and nonspecialized
25316 (estimate_ipcp_clone_size_and_time): Update.
25317 (inline_merge_summary): Update.
25318 (do_estimate_edge_time): Update.
25319 (do_estimate_edge_size): Update.
25320 (do_estimate_edge_hints): Update.
25321 (inline_read_section, inline_write_summary): Stream both new predicates.
25322 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25324 (compute_inlined_call_time): Cleanup.
25325 (big_speedup_p): Update.
25326 (edge_badness): Update.
25327 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25328 (size_time_entry): Replace predicate by exec_predicate and
25329 nonconst_predicate.
25330 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25331 (estimate_edge_time): Return also nonspec_time.
25332 (reset_edge_growth_cache): Update.
25334 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25336 PR rtl-optimization/80491
25337 * ifcvt.c (noce_process_if_block): When looking for x setter
25338 with missing else_bb, don't check only the insn right before
25339 cond_earliest, but look for the last insn that x is modified in
25340 within the same bb.
25342 PR rtl-optimization/80491
25343 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25345 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25347 PR tree-optimization/80487
25348 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25350 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25352 PR tree-optimization/79697
25353 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25354 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25355 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25357 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25358 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25360 2017-04-28 Martin Sebor <msebor@redhat.com>
25362 PR tree-optimization/80523
25363 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25364 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25366 (maybe_warn, format_directive, parse_directive): Use new functions.
25367 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25369 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25371 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25373 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25375 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25376 target_header_dir): Set correctly.
25377 * configure: Regenerated.
25378 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25379 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25380 instead of SYSTEM_HEADER_DIR.
25382 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25384 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25385 (estimate_local_effects): Likewise.
25386 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25387 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25388 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25389 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25390 do_estimate_edge_time, estimate_edge_time): Likewise.
25391 * ipa-inline-analysis.c (estimate_node_size_and_time,
25392 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25393 (estimate_time_after_inlining): Remove.
25395 2017-04-28 Martin Liska <mliska@suse.cz>
25397 * doc/gcov.texi: Enhance documentation of gcov.
25399 2017-04-28 Martin Liska <mliska@suse.cz>
25401 * doc/gcov.texi: Sort options in alphabetic order.
25402 * doc/gcov-dump.texi: Likewise.
25403 * doc/gcov-tool.texi: Likewise.
25404 * gcov.c (print_usage): Likewise.
25405 * gcov-dump.c (print_usage): Likewise.
25406 * gcov-tool.c (print_merge_usage_message): Likewise.
25407 (print_rewrite_usage_message): Likewise.
25408 (print_overlap_usage_message): Likewise.
25410 2017-04-28 Martin Liska <mliska@suse.cz>
25412 PR gcov-profile/53915
25413 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25415 2017-04-28 Martin Liska <mliska@suse.cz>
25417 PR gcov-profile/79891
25418 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25419 is marked by compiler as living on a line.
25420 (get_cycles_count): Remove usage of the union.
25421 (output_intermediate_file): Likewise.
25422 (find_source): Fix GNU coding style.
25423 (accumulate_line_counts): Remove old non-all block mode.
25424 (output_lines): Remove usage of the union.
25425 * profile.c (output_location): Include all BBs, even if
25426 belonging to a same line (and file) as a previous BB.
25428 2017-04-28 Martin Liska <mliska@suse.cz>
25430 * gcov.c (process_args): Handle new argument 'w'.
25431 (read_graph_file): Assign ID to BBs.
25432 (output_branch_count): Display BB # if verbose flag is set.
25433 (output_lines): Likewise for arcs.
25434 (print_usage): Add '--verbose' option help.
25435 * doc/gcov.texi: Document --verbose (-w) option.
25437 2017-04-28 Martin Liska <mliska@suse.cz>
25439 * gcov.c (struct block_location_info): New struct.
25440 (process_file): Fill up the new structure.
25441 (read_graph_file): Replace usage of encoding by the newly added
25443 (add_line_counts): Likewise.
25444 (accumulate_line_counts): Remove usage of the union.
25445 (function_info::function_info): New function.
25446 (function_info::~function_info): Likewise.
25447 (process_file): Call delete instead of release_function.
25448 (release_function): Release the function.
25449 (release_structures): Call delete instead of release_function.
25450 (solve_flow_graph): Replace usage of num_blocks.
25451 (find_exception_blocks): Likewise.
25452 (output_lines): Fix GNU coding style.
25454 2017-04-28 Martin Liska <mliska@suse.cz>
25457 * coverage.c (coverage_remove_note_file): New function.
25458 * coverage.h: Declare the function.
25459 * toplev.c (finalize): Clean if an error has been seen.
25461 2017-04-28 Martin Liska <mliska@suse.cz>
25463 PR gcov-profile/80031
25464 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25465 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25466 * gcov.c (read_graph_file): Read just number of blocks.
25467 * profile.c (branch_prob): Do not stream 0 flags per a basic
25470 2017-04-28 Martin Liska <mliska@suse.cz>
25472 * gcov-dump.c (tag_*): Add new argument to declarations.
25473 (dump_gcov_file): Likewise.
25474 (tag_blocks): Add and use new argument depth.
25475 (tag_arcs): Likewise.
25476 (tag_lines): Likewise.
25477 (tag_counters): Likewise.
25478 (tag_summary): Likewise.
25479 (dump_working_sets): Use depth to do a proper indentation.
25481 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25484 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25485 * symtab.c (symtab_node::debug_symtab): Move definition here.
25487 2017-04-28 Richard Biener <rguenther@suse.de>
25489 * lto-streamer.h (LTO_major_version): Bump to 7.
25491 2017-04-28 Richard Biener <rguenther@suse.de>
25493 * tree-vrp.c (assert_info): New struct.
25494 (add_assert_info): New helper.
25495 (register_edge_assert_for_2): Refactor to add asserts to a vector
25497 (register_edge_assert_for_1): Likewise.
25498 (register_edge_assert_for): Likewise.
25499 (finish_register_edge_assert_for): New helper actually registering
25500 asserts where live on edge.
25501 (find_conditional_asserts): Adjust.
25502 (find_switch_asserts): Likewise.
25503 (evrp_dom_walker::try_find_new_range): Generalize.
25504 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25506 2017-04-27 Marek Polacek <polacek@redhat.com>
25509 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25510 arg10 and arg11 to itype.
25512 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25514 * doc/extend.texi (Object Size Checking): Improve grammar.
25516 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25519 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25520 that the logic for permitting reciprocal estimates matches that
25523 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25526 * tree.c (type_cache_hasher::equal): Only compare
25527 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25528 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25529 non-aggregate element types.
25530 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25531 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25533 2017-04-27 Richard Biener <rguenther@suse.de>
25535 PR middle-end/80533
25536 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25537 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25538 keeping a reference to a trailing array.
25540 2017-04-27 Richard Biener <rguenther@suse.de>
25542 PR middle-end/80539
25543 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25544 being in loop-closed SSA form conservatively.
25545 (chrec_fold_multiply_poly_poly): Likewise.
25547 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25549 PR middle-end/79665
25550 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25551 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25553 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25556 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25557 (aarch64_function_arg_alignment): Return unsigned int again, but still
25558 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25559 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25560 Don't emit -Wpsabi note.
25561 (aarch64_function_arg_boundary): Likewise.
25562 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25565 2017-04-26 Nathan Sidwell <nathan@acm.org>
25567 * tree.h (crc32_unsigned_n): Declare.
25568 (crc32_unsigned, crc32_unsigned): Make inline.
25569 * tree.c (crc32_unsigned_bits): Replace with ...
25570 (crc32_unsigned_n): ... this.
25571 (crc32_unsigned, crc32_byte): Remove.
25572 (crc32_string): Remove unnecessary braces.
25574 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25576 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25577 * ipa-inline-analysis.c (MAX_TIME): Remove.
25578 (account_size_time): Use sreal for time.
25579 (dump_inline_summary): Update.
25580 (estimate_function_body_sizes): Update.
25581 (estimate_edge_size_and_time): Update.
25582 (estimate_calls_size_and_time): Update.
25583 (estimate_node_size_and_time): Update.
25584 (inline_merge_summary): Update.
25585 (inline_update_overall_summary): Update.
25586 (estimate_time_after_inlining): Update.
25587 (inline_read_section): Update.
25588 (inline_write_summary): Update.
25589 * ipa-inline.c (compute_uninlined_call_time): Update.
25590 (compute_inlined_call_time): Update.
25591 (recursive_inlining): Update.
25592 (inline_small_functions): Update.
25593 (dump_overall_stats): Update.
25594 * ipa-inline.h: Include sreal.h.
25595 (size_time_entry): Turn time to sreal.
25596 (inline_summary): Turn self_time nad time to sreal.
25598 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25600 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25602 (sreal::stream_out, sreal::stream_in): New.
25603 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25605 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25607 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25610 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25613 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25615 (dimode_scalar_chain::compute_convert_gain): Ditto.
25616 (dimode_scalar_chain::make_vector_copies): Ditto.
25617 (dimode_scalar_chain::convert_reg): Ditto.
25618 (dimode_scalar_chain::convert_insn): Ditto.
25619 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25620 (VI248_AVX512BW_1): New mode iterator.
25621 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25622 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25625 2017-04-25 Martin Sebor <msebor@redhat.com>
25627 PR tree-optimization/80497
25628 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25629 constants are representable in HOST_WIDE_INT.
25630 (parse_directive): Ditto.
25632 2017-04-25 Martin Sebor <msebor@redhat.com>
25635 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25636 (new_zero_array): Adjust signature.
25637 (dom_info::dom_init): Used unsigned rather that size_t.
25638 (dom_info::dom_info): Same.
25640 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25641 Jakub Jelinek <jakub@redhat.com>
25644 * config/arm/arm.c: Include gimple.h.
25645 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25646 returns negative, increment ncrn only if it returned positive.
25647 (arm_needs_doubleword_align): Return int instead of bool,
25648 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25649 members, but if there is any such non-FIELD_DECL
25650 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25651 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25652 returns negative, increment nregs only if it returned positive.
25653 (arm_setup_incoming_varargs): Likewise.
25654 (arm_function_arg_boundary): Emit -Wpsabi note if
25655 arm_needs_doubleword_align returns negative, return
25656 DOUBLEWORD_ALIGNMENT only if it returned positive.
25658 2017-04-25 Marek Polacek <polacek@redhat.com>
25661 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25662 first argument to type.
25664 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25667 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25668 type checks to test for compatibility instead of equality.
25670 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25671 Jakub Jelinek <jakub@redhat.com>
25674 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
25676 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
25677 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
25678 the alignment computation, but return their maximum in warn_alignment.
25679 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25680 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
25682 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
25683 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25686 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25688 * config/arc/simdext.md (dmpyh): Fix typo.
25690 2017-04-25 Richard Biener <rguenther@suse.de>
25692 PR tree-optimization/80492
25693 * alias.c (compare_base_decls): Handle registers with asm
25694 specification conservatively.
25695 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25696 compare_base_decls returning dont-know properly.
25698 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25700 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25701 (legitimate_offset_address_p): New function.
25702 (arc_legitimate_address_p): Use above function.
25704 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25706 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25708 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25710 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25711 ACCH registers whenever they are available.
25713 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25715 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25716 double regs fix when not used.
25718 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25720 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25722 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25723 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25725 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25727 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25728 when emitting short ADD instructions.
25730 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25732 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25734 (cmpsi_cc_c_insn): Likewise.
25735 (cbranchsi4_scratch): Compute proper instruction length using
25736 compact_hreg_operand.
25737 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25739 2017-04-25 Richard Biener <rguenther@suse.de>
25741 PR middle-end/80509
25742 * passes.c (pass_manager::pass_manager): Initialize
25743 m_name_to_pass_map.
25745 2017-04-25 Richard Biener <rguenther@suse.de>
25747 PR tree-optimization/79201
25748 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25750 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25753 * config/s390/vector.md: Split MEM->GPR vector moves for
25754 non-s_operand addresses.
25756 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25759 * config/s390/predicates.md (reload_const_wide_int_operand): New
25761 * config/s390/s390.md ("movti"): Remove d/P alternative.
25762 ("movti_bigconst"): New pattern definition.
25764 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25767 * s390-protos.h (s390_expand_cs_hqi): Removed.
25768 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25769 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25770 modes as well as CCZ1mode and CCZmode.
25771 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25772 signature of s390_emit_compare_and_swap.
25773 (s390_expand_cs_hqi): Likewise, make static.
25774 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25775 compare-and-swap, in some cases.
25776 (s390_expand_cs): Wrapper function.
25777 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25779 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25780 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25781 patterns for small and large integers. Forbid symref memory operands.
25782 Move expander to s390.c. Require cc register.
25783 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25784 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25785 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25786 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25787 symref memory operands. Remove CC mode and call s390_match_ccmode
25789 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25791 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25793 * config/s390/s390.md (define_peephole2): New peephole to help
25794 combining the load-and-test pattern with volatile memory.
25796 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25798 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25799 with CCZmode for TARGET_Z196.
25801 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25803 PR rtl-optimization/80501
25804 * combine.c (make_compound_operation_int): Set subreg_code to SET
25805 even for AND with mask of the sign bit of mode.
25807 PR rtl-optimization/80500
25808 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25809 sum's initial value.
25811 2017-04-25 Julian Brown <julian@codesourcery.com>
25812 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25814 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25816 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25818 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25820 2017-04-25 Julian Brown <julian@codesourcery.com>
25821 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25823 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25824 (thunderx2t99_sha): New Reservation.
25826 2017-04-25 Julian Brown <julian@codesourcery.com>
25827 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25829 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25830 type for 1-element load.
25832 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25834 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25836 2017-04-24 Martin Jambor <mjambor@suse.cz>
25838 PR tree-optimization/80293
25839 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25840 char arrays not totally scalarizable if it is false.
25841 (analyze_all_variable_accesses): Pass correct value in the new
25842 parameter. Add a statistics counter.
25844 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25846 PR middle-end/79931
25847 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25849 2017-04-24 Richard Biener <rguenther@suse.de>
25851 PR tree-optimization/80494
25852 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25853 out for complex types.
25855 2017-04-24 Richard Biener <rguenther@suse.de>
25857 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25858 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25859 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25860 (DFS): Adjust and never fail.
25861 (sccvn_dom_walker::fail): Remove.
25862 (sccvn_dom_walker::before_dom_children): Adjust.
25863 (run_scc_vn): Likewise and never fail.
25864 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25865 (pass_fre::execute): Likewise.
25867 2017-04-24 Richard Biener <rguenther@suse.de>
25869 PR tree-optimization/79725
25870 * tree-ssa-sink.c (statement_sink_location): Return whether
25871 failure reason was zero uses. Move that check later.
25872 (sink_code_in_bb): Deal with zero uses by removing the stmt
25875 2017-04-24 Richard Biener <rguenther@suse.de>
25878 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25879 pointer-based references.
25881 2017-04-24 Richard Biener <rguenther@suse.de>
25884 * pass_manager.h (pass_manager::operator new): Remove.
25885 (pass_manager::operator delete): Likewise.
25886 * passes.c (pass_manager::operator new): Remove.
25887 (pass_manager::operator delete): Likewise.
25888 (pass_manager::pass_manager): Zero individual pass members.
25890 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25893 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25894 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25895 Check "XEXP (src, 1)" operand here.
25896 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25897 Check "XEXP (src, 1)" operand here.
25898 (dimode_scalar_chain::make_vector_copies): Detect count register
25899 of a shift instruction. Zero extend count register from QImode
25900 to DImode to satisfy vector shift pattern count operand predicate.
25901 Substitute vector shift count operand with a DImode copy.
25902 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25905 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25907 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25908 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25909 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25910 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25911 (UNSPEC_NOREX_MEM): Remove definition.
25913 2017-04-21 Richard Biener <rguenther@suse.de>
25915 PR tree-optimization/79547
25916 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25917 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25918 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25919 without any constraints.
25921 2017-04-21 Richard Biener <rguenther@suse.de>
25923 PR tree-optimization/78847
25924 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25926 2017-04-21 Richard Biener <rguenther@suse.de>
25928 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25929 (build_distinct_type_copy): Likewise.
25930 (build_variant_type_copy): Likewise.
25931 * tree.c (build_qualified_type): Pass down mem-stat info.
25932 (build_distinct_type_copy): Likewise.
25933 (build_variant_type_copy): Likewise.
25935 2017-04-21 Richard Biener <rguenther@suse.de>
25937 PR tree-optimization/80237
25938 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25940 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25941 for a simplified result.
25943 2016-04-21 Richard Biener <rguenther@suse.de>
25945 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25946 sth as strict as a simple_iv but a chrec without symbols and an
25947 operand defined in the loop we are peeling (and not some subloop).
25948 (propagate_constants_for_unrolling): Propagate all constants.
25950 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25953 * config/i386/i386.c (print_reg): Remove assert for disalowed
25954 regno values, call output_operand_lossage instead.
25956 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25959 * config/i386/constraints.md (Yc): New register constraint.
25960 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25961 Use Yc constraint for alternative 2 of operand 0. Remove
25962 preferred_for_speed attribute.
25964 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25966 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25967 lastprivate clauses in SIMT case.
25969 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25971 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25973 2017-04-20 Richard Biener <rguenther@suse.de>
25975 PR tree-optimization/57796
25976 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25977 as N scalar stores.
25978 (vect_model_load_cost): Cost gathers as N scalar loads.
25980 2017-04-20 Richard Biener <rguenther@suse.de>
25982 * ggc-page.c (ggc_allocated_p): Rename to ...
25983 (safe_lookup_page_table_entry): ... this and return the lookup
25985 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
25987 2017-04-20 Richard Biener <rguenther@suse.de>
25989 PR tree-optimization/80453
25990 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
25991 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
25992 from the conditions.
25993 (vn_phi_eq): Pass them down.
25994 (vn_phi_lookup): Record them.
25995 (vn_phi_insert): Likewise.
25997 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
25999 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26000 uninitialized variable warning to avoid buffer overrun.
26002 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26005 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26006 is suppressed for '{ 0 }' in C.
26008 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26010 * BASE-VER: Set to 8.0.0.
26012 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26014 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26015 priority .init_array and .fini_array section with SECTION_NOTYPE
26018 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26020 PR middle-end/80423
26021 * tree.h (build_array_type): Add typeless_storage default argument.
26022 * tree.c (type_cache_hasher::equal): Also compare
26023 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26024 (build_array_type): Add typeless_storage argument, set
26025 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26027 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26028 (build_array_type): Likewise. Add typeless_storage argument.
26030 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26031 Jakub Jelinek <jakub@redhat.com>
26033 PR tree-optimization/80426
26034 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26035 operation on symbolic operands, also compute the overflow for the
26036 invariant part when the operation degenerates into a negation.
26038 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26041 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26042 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26045 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26047 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26050 * config/avr/avr.c (tree.h): Include it.
26051 (cgraph.h): Include it.
26052 (avr_encode_section_info): Don't warn for uninitialized progmem
26053 variable if it's just an alias.
26055 2017-04-19 Richard Biener <rguenther@suse.de>
26058 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26059 when needed by AutoPGO.
26061 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26064 * doc/lto.texi: Remove an extra 'that'.
26066 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26068 PR rtl-optimization/80429
26069 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26070 are only used in debug insns.
26072 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26073 Vladimir Makarov <vmakarov@redhat.com>
26075 * config/sparc/predicates.md (input_operand): Add comment. Return
26076 true for any memory operand when LRA is in progress.
26077 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26079 2017-04-18 Jeff Law <law@redhat.com>
26082 * mips.md ({return,simple_return}_internal): Do not overwrite
26085 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26087 PR tree-optimization/80443
26088 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26089 instead of adding 1, subtract -1 and similarly instead of subtracting
26092 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26094 PR rtl-optimization/80357
26095 * haifa-sched.c (tmp_bitmap): New variable.
26096 (model_recompute): Handle duplicate use records.
26097 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26098 (free_global_sched_pressure_data): Free it.
26100 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26103 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26104 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26105 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26106 instead of SYSTEM_HEADER_DIR.
26108 2017-04-18 Jeff Law <law@redhat.com>
26110 PR middle-end/80422
26111 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26112 predecessors after walking up the insn chain.
26114 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26117 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26118 sizetype type into debug info.
26120 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26123 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26124 unneeded test for TARGET_UPPER_REGS_SF.
26125 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26127 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26130 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26131 instead of gsi_after_labels.
26133 2017-04-18 Jeff Law <law@redhat.com>
26135 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26139 2017-04-13 Jeff Law <law@redhat.com>
26140 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26141 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26143 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26146 * config/avr/avr.c (intl.h): Include it.
26147 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26149 2017-04-18 Martin Liska <mliska@suse.cz>
26151 PR gcov-profile/78783
26152 * gcov-tool.c (gcov_output_files): Validate that destination
26153 file is either removed by the tool or by a user.
26155 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26156 Guy Benyei <guybe@mellanox.com>
26158 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26159 block, and do not negate it, the stored id is already negative.
26161 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26163 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26165 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26168 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26169 masks of options that should be turned off if the VSX vector
26170 options are turned off.
26171 (OTHER_P8_VECTOR_MASKS): Likewise.
26172 (OTHER_VSX_VECTOR_MASKS): Likewise.
26173 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
26174 rs6000_disable_incompatible_switches to validate no type switches
26176 (rs6000_incompatible_switch): New function to disallow turning on
26177 other vector options if -mno-vsx, -mno-power8-vector, or
26178 -mno-power9-vector are specified.
26180 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26182 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
26184 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26186 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
26187 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
26188 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
26189 (ARG_POINTER_CFA_OFFSET): Likewise.
26191 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26193 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
26194 conditions to take advantage of various optimizations.
26196 2017-04-13 Jeff Law <law@redhat.com>
26198 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26199 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26200 (zero_extendsidi2_dext): Likewise.
26202 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26205 * fold-const.c (fold_ternary_loc): Revert
26206 use op0 instead of fold_convert_loc (loc, type, arg0) part of
26209 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
26211 PR rtl-optimization/80343
26212 * lra-remat.c (update_scratch_ops): Assign original hard reg to
26213 new scratch pseudo.
26215 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
26218 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
26219 to ubsan_encode_value.
26221 2017-04-13 Jeff Law <law@redhat.com>
26223 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
26224 appearing in DEBUG_INSNs.
26226 2017-04-13 Martin Liska <mliska@suse.cz>
26228 PR gcov-profile/80413
26229 * gcov-io.c (gcov_write_string): Copy to buffer just when
26230 allocated size is greater than zero.
26232 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26235 * dwarf2out.c (decls_for_scope): Ignore declarations of
26236 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26238 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26241 * ipa-visibility.c (non_local_p): Fix typos.
26242 (localize_node): When localizing symbol in same comdat group,
26243 dissolve the group only when we know external symbols are going
26245 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26247 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26249 PR tree-optimization/79390
26250 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26251 order does not result in usable sequence, retry with reversed operand
26257 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26258 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26259 op0 instead of fold_convert_loc (loc, type, arg0).
26261 2017-04-12 Jeff Law <law@redhat.com>
26263 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26264 has a delay slot in the generated code.
26266 * config/cris/cris.md (cris_preferred_reload_class): Return
26267 GENNONACR_REGS rather than GENERAL_REGS.
26269 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26272 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26273 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26274 signedness of the result type.
26276 2017-04-12 Richard Biener <rguenther@suse.de>
26277 Jeff Law <law@redhat.com>
26279 PR tree-optimization/80359
26280 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26281 trim stores to TARGET_MEM_REFs.
26283 2017-04-12 Richard Biener <rguenther@suse.de>
26285 PR tree-optimization/79390
26286 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26287 threading case even more.
26289 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26292 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26293 for quad_address_p for TImode, instead of just not indexed_address.
26295 2017-04-12 Richard Biener <rguenther@suse.de>
26296 Bernd Edlinger <bernd.edlinger@hotmail.de>
26298 PR middle-end/79671
26299 * alias.c (component_uses_parent_alias_set_from): Handle
26300 TYPE_TYPELESS_STORAGE.
26301 (get_alias_set): Likewise.
26302 * tree-core.h (tree_type_common): Add typeless_storage flag.
26303 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26304 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26305 for types containing members with TYPE_TYPELESS_STORAGE.
26306 (place_field): Likewise.
26307 (layout_type): Likewise for ARRAY_TYPE.
26308 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26309 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26310 TYPE_TYPELESS_STORAGE.
26311 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26313 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26316 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26317 first argument to type.
26319 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26323 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26324 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26325 (rs6000_expand_binop_builtin): Likewise.
26326 (rs6000_expand_ternop_builtin): Likewise; also add missing
26327 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26328 vshasigma built-ins.
26329 * doc/extend.texi: Document that vec_xxpermdi's third argument
26330 must be a constant.
26332 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26334 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26335 Use shift_const cost parameter when calculating gain of STV shifts.
26337 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26339 PR rtl-optimization/70478
26340 * lra-constraints.c (process_alt_operands): Check memory for
26341 disfavoring memory insn operand.
26343 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26345 PR middle-end/80100
26346 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26347 left shift in unsigned HOST_WIDE_INT type.
26349 PR rtl-optimization/80385
26350 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26351 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26354 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26355 if they have any depend clauses.
26357 2017-04-11 Martin Liska <mliska@suse.cz>
26360 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26361 * ipa-split.c (split_function): Create a local comdat symbol
26362 if caller is in a comdat group.
26364 2017-04-11 Martin Liska <mliska@suse.cz>
26367 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26370 2017-04-11 Martin Sebor <msebor@redhat.com>
26372 PR middle-end/80364
26373 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26374 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26376 (directive::set_width, directive::set_precision, format_character):
26378 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26381 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26384 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26385 conflict, set target->arch_name instead of target->cpu_name.
26387 2017-04-11 Richard Biener <rguenther@suse.de>
26389 PR tree-optimization/80374
26390 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26391 build_zero_cst, remove fold_convertible_p check again.
26393 2017-04-11 Martin Liska <mliska@suse.cz>
26396 * ubsan.c (instrument_object_size): Do not instrument register
26399 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26402 * config/i386/i386-builtin-types.def
26403 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26404 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26405 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26406 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26407 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26408 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26409 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26410 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26411 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26412 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26413 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26414 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26415 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26416 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26417 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26418 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26419 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26420 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26421 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26422 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26423 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26424 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26425 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26426 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26427 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26428 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26429 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26430 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26431 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26432 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26433 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26434 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26435 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26436 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26437 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26438 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26439 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26440 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26441 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26442 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26443 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26444 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26445 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26446 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26447 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26448 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26450 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26451 flag to second_arg_count, handle 4 argument function type _COUNT
26452 aliases, handle second_arg_count on second argument rather than last.
26454 2017-04-10 Jeff Law <law@redhat.com>
26456 PR tree-optimization/80374
26457 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26458 record anything if we can not convert integer_zero_node to the
26461 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26464 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26465 Enhance special handling given to the TARGET_P9_MINMAX option in
26466 relation to certain other options.
26468 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26470 PR tree-optimization/80153
26471 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26472 remove POINTER_PLUS_EXPR's base part directly, rather than through
26475 2017-04-10 Richard Biener <rguenther@suse.de>
26476 Bin Cheng <bin.cheng@arm.com>
26478 PR tree-optimization/80153
26479 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26480 the first element of pointer type aff_tree. Build result expr in
26482 (add_elt_to_tree): Convert to type unconditionally. Remove other
26483 fold_convert calls.
26484 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26485 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26487 2017-04-10 Richard Biener <rguenther@suse.de>
26489 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26492 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26494 PR rtl-optimization/70478
26495 * lra-constraints.c (curr_small_class_check): New.
26496 (update_and_check_small_class_inputs): New.
26497 (process_alt_operands): Update curr_small_class_check. Disfavor
26498 alternative insn memory operands. Check available regs for small
26501 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26504 * config/mips/mips.opt (-mvirt): Update description.
26505 * doc/invoke.texi (-mvirt): Likewise.
26507 2017-04-10 Richard Biener <rguenther@suse.de>
26509 PR middle-end/80362
26510 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26511 looking for NEGATE_EXPR in -A / -B to A / B folding.
26513 2017-04-10 Martin Liska <mliska@suse.cz>
26515 PR gcov-profile/80224
26516 * gcov.c (print_usage): Fix usage string.
26517 (get_gcov_intermediate_filename): Remove.
26518 (output_gcov_file): Use both for normal and intermediate format.
26519 (generate_results): Do not initialize special file for
26520 intermediate format.
26522 2017-04-10 Richard Biener <rguenther@suse.de>
26524 PR tree-optimization/80304
26525 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26528 2017-04-10 Nathan Sidwell <nathan@acm.org>
26531 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26532 (rs6000_init_builtins): Use it.
26534 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26536 * config/arm/arm.md (<mrc>): Add mode to SET source.
26537 (<mrrc>): Likewise.
26539 2017-04-10 Richard Biener <rguenther@suse.de>
26541 PR middle-end/80344
26542 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26544 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26547 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26548 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26549 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26550 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26551 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26552 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26553 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26554 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26555 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26556 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26557 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26558 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26559 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26560 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26561 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26562 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26563 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26564 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26565 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26566 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26567 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26568 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26569 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26571 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26573 PR rtl-optimization/70478
26574 * lra-constraints.c: Reverse the last patch.
26576 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26578 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26579 Add comment for WCHAR_T.
26581 2017-04-08 Martin Liska <mliska@suse.cz>
26584 2017-04-07 Martin Liska <mliska@suse.cz>
26587 * ipa-split.c (split_function): Add function part to a same comdat
26590 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26593 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26595 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26597 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26598 (vec_pairing): ...to this.
26599 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26600 (rs6000_sched_init): Adjust for name change.
26601 (struct rs6000_sched_context): Likewise.
26602 (rs6000_init_sched_context): Likewise.
26603 (rs6000_set_sched_context): Likewise.
26605 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26611 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26613 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26614 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26615 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26617 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26619 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26621 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26623 PR rtl-optimization/70703
26624 * ira-color.c (update_conflict_hard_regno_costs): Use
26625 int64_t instead of HOST_WIDE_INT.
26627 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26629 PR rtl-optimization/70478
26630 * lra-constraints.c (process_alt_operands): Disfavor alternative
26631 insn memory operands.
26633 2017-04-07 Jeff Law <law@redhat.com>
26635 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26636 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26638 2017-04-07 Martin Liska <mliska@suse.cz>
26641 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26642 Show error message instead of an ICE.
26644 2017-04-07 Martin Liska <mliska@suse.cz>
26647 * ipa-split.c (split_function): Add function part to a same comdat
26650 2017-04-07 Richard Biener <rguenther@suse.de>
26652 PR middle-end/80341
26653 * tree.c (get_unwidened): Also handle ! for_type case for
26655 * convert.c (do_narrow): Split out from ...
26656 (convert_to_integer_1): ... here. Do not pass final truncation
26657 type to get_unwidened for TRUNC_DIV_EXPR.
26659 2017-04-07 Richard Biener <rguenther@suse.de>
26661 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26663 (aff_combination_const): Adjust.
26664 (aff_combination_scale): Likewise.
26665 (aff_combination_add_elt): Likewise.
26666 (aff_combination_add_cst): Likewise.
26667 (aff_combination_convert): Likewise.
26668 (add_elt_to_tree): Likewise. Remove unused argument.
26669 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26671 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
26673 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
26675 * config/arm/arm.c (arm_default_short_enums): Use
26676 ARM_DEFAULT_SHORT_ENUMS.
26677 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
26679 2017-04-06 Jakub Jelinek <jakub@redhat.com>
26682 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
26683 members with redundant out-of-class redeclaration.
26685 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26688 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26689 * config/i386/i386.md (*zero_extendsidi2):
26690 Add (?*x,*x) and (?*v,*v) alternatives.
26692 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26695 * config/i386/i386.c (ix86_expand_builtin)
26696 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26697 mode from insn data. Convert operands to insn operand mode.
26698 Copy operands that don't satisfy insn predicate to a register.
26700 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26702 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26705 2017-04-06 Richard Biener <rguenther@suse.de>
26707 PR tree-optimization/80334
26708 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26709 preserve alignment of accesses.
26711 2017-04-06 Richard Biener <rguenther@suse.de>
26713 PR tree-optimization/80262
26714 * tree-sra.c (build_ref_for_offset): Preserve address-space
26716 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26717 Drop useless address-space information on MEM_REF offsets.
26719 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26721 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26723 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26725 PR rtl-optimization/70703
26726 * ira-color.c (update_conflict_hard_regno_costs): Use
26727 HOST_WIDE_INT instead of long.
26729 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26732 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26733 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26735 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26736 for x86_64 target. Handle -m3dnowa option.
26738 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26740 PR rtl-optimization/70703
26741 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26742 (update_conflict_hard_regno_costs): Use long instead of unsigned
26743 arithmetic for cost calculation.
26745 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26746 Bernd Edlinger <bernd.edlinger@hotmail.de>
26749 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26752 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26755 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26756 ptr_mode with Pmode throughout.
26757 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26758 into probe_stack_range and use DImode.
26760 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26763 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26764 call_eh_return is true.
26766 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26768 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26769 Initialize last_match_fntype_index.
26771 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26774 * tree-nvr.c: Include internal-fn.h.
26775 (pass_return_slot::execute): Ignore internal calls without
26778 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26779 Richard Biener <rguenther@suse.de>
26782 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26783 captures used multiple times, except for the last use.
26784 * generic-match-head.c: Include gimplify.h.
26786 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26788 PR tree-optimization/79390
26789 * target.h (struct noce_if_info): Declare.
26790 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26791 * target.def (noce_conversion_profitable_p): New target hook.
26792 * ifcvt.h (struct noce_if_info): New type, moved from ...
26793 * ifcvt.c (struct noce_if_info): ... here.
26794 (noce_conversion_profitable_p): Renamed to ...
26795 (default_noce_conversion_profitable_p): ... this. No longer
26797 (noce_try_store_flag_constants, noce_try_addcc,
26798 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26799 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26800 instead of noce_conversion_profitable_p.
26801 * config/i386/i386.c: Include ifcvt.h.
26802 (ix86_option_override_internal): Don't override
26803 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26804 (ix86_noce_conversion_profitable_p): New function.
26805 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26806 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26807 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26808 * doc/tm.texi: Regenerated.
26810 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26812 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26815 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26818 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26819 instructions for small multiply cores.
26821 2017-04-04 Jeff Law <law@redhat.com>
26823 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26825 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26826 that are not set by the loop over the elements.
26828 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26831 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26832 int mode, convert_modes it to mode as unsigned, otherwise use
26833 lowpart_subreg to mode rather than SImode.
26834 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26835 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26836 Use DImode instead of SImode for the shift count operand.
26837 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26840 2017-04-04 Richard Biener <rguenther@suse.de>
26842 PR middle-end/80281
26843 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26844 arithmetic done for the negate or the plus. Simplify.
26845 (A - (-B) -> A + B): Likewise.
26846 * fold-const.c (split_tree): Make sure to not negate pointers.
26848 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26850 PR rtl-optimization/60818
26851 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26852 a compare of comparisons with the thing compared if this results
26853 in a different machine mode.
26855 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26857 * alias.c (base_alias_check): Fix typo in comment.
26858 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26859 * cgraphunit.c (symbol_table::compile): Likewise.
26860 * collect2.c (maybe_run_lto_and_relink): Likewise.
26861 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26862 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26863 * config/avr/avr.c (avr_map_op_t): Likewise.
26864 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26865 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26866 * config/epiphany/epiphany.md (movcc): Likewise.
26867 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26868 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26870 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26871 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26872 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26873 * config/sh/sh.c (sh_rtx_costs): Likewise.
26874 * fold-const.c (fold_truth_andor): Likewise.
26875 * genautomata.c (collapse_flag): Likewise.
26876 * gengtype.h (struct type::u::s): Likewise.
26877 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26878 * input.c (FORMAT_AMOUNT): Likewise.
26879 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26880 (known_aggs_to_agg_replacement_list): Likewise.
26881 * ipa-inline-analysis.c: Likewise.
26882 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26883 * ipa-polymorphic-call.c
26884 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26885 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26886 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26888 * modulo-sched.c (apply_reg_moves): Likewise.
26889 * omp-expand.c (build_omp_regions_1): Likewise.
26890 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26891 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26892 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26893 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26894 * value-prof.c: Likewise.
26895 * var-tracking.c (val_reset): Likewise.
26897 2017-04-03 Richard Biener <rguenther@suse.de>
26899 PR tree-optimization/80275
26900 * fold-const.c (split_address_to_core_and_offset): Handle
26903 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26905 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26906 descriptors is at least equal to that of functions.
26908 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26910 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26912 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26915 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26916 (mov<IMOD4:mode>): New expander.
26917 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26919 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26921 PR rtl-optimization/79405
26922 * fwprop.c (propagations_left): New variable.
26923 (forward_propagate_into): Decrement it.
26924 (fwprop_init): Initialize it.
26925 (fw_prop): If the variable has reached zero, stop propagating.
26926 (fwprop_addr): Ditto.
26928 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26931 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26932 a FUNCTION_DECL, pass it as decl instead of origin to
26935 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26937 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26940 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26943 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26944 TARGET_VSX_SMALL_INTEGER.
26946 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26948 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26949 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26951 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26953 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26954 extraction from odd-numbered MSA register.
26956 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26958 PR middle-end/80173
26959 * expmed.c (store_bit_field_1): Don't attempt to create
26960 a word subreg out of hard registers wider than word if they
26961 have HARD_REGNO_NREGS of 1 for their mode.
26963 PR middle-end/80163
26964 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26965 conversions to integer types wider than word and pointer.
26968 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26969 (rtx_equal_for_cselib_p): Pass 0 to it.
26970 * cselib.c (cselib_hasher::equal): Likewise.
26971 (rtx_equal_for_cselib_1): Add depth argument. If depth
26972 is 128, don't look up VALUE locs and punt. Increment
26973 depth in recursive calls when walking VALUE locs.
26975 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26977 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26978 (make_gcov_file_name): Use the canonical path name for generating
26980 (read_line): Fix handling of files with ascii null bytes.
26982 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26984 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
26985 to initialise a vector register instead
26986 of using a const_int.
26988 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26990 PR translation/80189
26991 * gimplify.c (omp_default_clause): Use %qs instead of %s in
26992 diagnostic messages.
26994 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
26997 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
26998 (dfp_diex_<mode>): Update mode of operand 1.
26999 * doc/extend.texi (dxex, dxexq): Document change to return type.
27000 (diex, diexq): Document change to argument type.
27002 2017-03-30 Martin Jambor <mjambor@suse.cz>
27005 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27006 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27007 it reflects the signature changes performed at the callee side.
27008 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27009 to cgraph_build_function_type_skip_args.
27010 (build_function_decl_skip_args): Adjust call to the above function.
27012 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27015 * config/i386/sse.md
27016 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27017 register as dest whenever it is a MEM not rtx_equal_p to the
27018 corresponding dup operand, and when forcing into reg move the
27019 reg into the memory afterwards.
27020 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27021 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27022 for the force_reg mode.
27023 (avx512vl_vextractf128<mode>): Use register as dest either
27024 always when a MEM, or when it is a MEM not rtx_equal_p to the
27025 corresponding dup operand, or even not when it is a CONST_VECTOR
27026 depending on the mode and lo vs. hi.
27027 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27029 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27030 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27031 Likewise. Require that operands[2] is even.
27032 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27033 Remove extraneous parens. Require that operands[2] is a multiple
27035 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27036 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27037 disallow memory then.
27039 2017-03-30 Richard Biener <rguenther@suse.de>
27041 PR tree-optimization/77498
27042 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27043 to non-constants over backedges.
27045 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27047 PR rtl-optimization/80233
27048 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27049 as last_combined_insn. Do not test for BARRIER_P separately.
27051 2017-03-29 Andreas Schwab <schwab@suse.de>
27054 * calls.c (prepare_call_address): Convert funexp to Pmode before
27055 copying to temp reg.
27057 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27059 PR tree-optimization/80158
27060 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27061 Handle possible future case of more than one alternate
27063 (replace_rhs_if_not_dup): Likewise.
27064 (replace_one_candidate): Likewise.
27066 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27068 PR rtl-optimization/80193
27069 * ira.c (ira): Do not check allocation for LRA.
27071 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27073 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27074 (nvptx_output_simt_exit): Declare.
27075 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27076 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27077 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27078 (nvptx_declare_function_name): ...here. Emit declaration of local
27079 memory space buffer for omp_simt_enter insn.
27080 (nvptx_output_unisimt_switch): New.
27081 (nvptx_output_softstack_switch): New.
27082 (nvptx_output_simt_enter): New.
27083 (nvptx_output_simt_exit): New.
27084 * config/nvptx/nvptx.h (struct machine_function): New fields
27085 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27086 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27087 (UNSPECV_SIMT_EXIT): Ditto.
27088 (omp_simt_enter_insn): New insn.
27089 (omp_simt_enter): New expansion.
27090 (omp_simt_exit): New insn.
27091 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27093 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27094 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27095 (expand_GOMP_SIMT_EXIT): New.
27096 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27097 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27098 (GOMP_SIMT_EXIT): Ditto.
27099 * target-insns.def (omp_simt_enter): New insn.
27100 (omp_simt_exit): Ditto.
27101 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27103 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27104 (lower_rec_input_clauses): Likewise.
27105 (lower_lastprivate_clauses): Handle SIMT privatization.
27107 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27108 (ompdevlow_adjust_simt_enter): New.
27109 (find_simtpriv_var_op): New.
27110 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27111 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27113 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27114 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27115 (copy_decl_for_dup_finish): Ditto.
27117 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27119 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27122 * config/i386/i386.c (ix86_option_override_internal): Always
27123 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27125 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27127 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27129 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27131 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27132 mark new edge's irreducible flag accordign to it.
27133 (vect_do_peeling): Check loop preheader edge's irreducible flag
27134 and pass it to function slpeel_add_loop_guard.
27136 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27138 PR tree-optimization/80218
27139 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27140 Update block frequencies and counts.
27142 2017-03-28 Richard Biener <rguenther@suse.de>
27144 PR tree-optimization/78644
27145 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27146 of a simplification result we may not use it at all.
27148 2017-03-28 Richard Biener <rguenther@suse.de>
27151 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27152 without arguments, generate default definition of a SSA name.
27154 2017-03-28 Richard Biener <rguenther@suse.de>
27156 PR middle-end/80222
27157 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27158 TYPE_REF_CAN_ALIAS_ALL references.
27159 * fold-const.c (fold_indirect_ref_1): Likewise.
27161 2017-03-28 Martin Liska <mliska@suse.cz>
27164 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27165 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27167 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27168 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27170 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27171 (EXTRA_SPECS): Define.
27172 (SUBTARGET_EXTRA_SPECS): Likewise.
27173 (SUBTARGET_CPP_SPEC): Likewise.
27174 * config/arc/elf.h (EXTRA_SPECS): Renamed to
27175 SUBTARGET_EXTRA_SPECS.
27176 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
27178 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27180 * config/arc/simdext.md (vst64_insn): Update pattern.
27181 (vld32wh_insn): Likewise.
27182 (vld32wl_insn): Likewise.
27183 (vld64_insn): Likewise.
27184 (vld32_insn): Likewise.
27186 2017-03-28 Marek Polacek <polacek@redhat.com>
27189 * fold-const.c (fold_comparison): Use protected_set_expr_location
27190 instead of SET_EXPR_LOCATION.
27192 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
27194 * tree.c (add_expr): Avoid name lookup warning.
27196 2017-03-27 Jeff Law <law@redhat.com>
27198 PR tree-optimization/80216
27199 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
27200 function name. Limit recursion depth.
27201 (record_temporary_equivalences): Corresponding changes.
27203 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
27205 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
27208 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27211 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
27213 * cfgcleanup.c (reg_note_cfa_p): New array.
27214 (insns_have_identical_cfa_notes): New function.
27215 (old_insns_match_p): Don't cross-jump in between /f
27216 and non-/f instructions. If both i1 and i2 are frame related,
27217 verify all CFA notes, their order and content.
27219 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
27222 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
27223 HImode and SImode with zero extend to DImode to one insn.
27224 (bswap<mode>2_extenddi): Likewise.
27225 (bswapsi2_extenddi): Likewise.
27226 (bswaphi2_extendsi): Likewise.
27227 (bswaphi2): Combine bswap HImode and SImode into one insn.
27228 Separate memory insns from swapping register.
27229 (bswapsi2): Likewise.
27230 (bswap<mode>2): Likewise.
27231 (bswaphi2_internal): Delete, no longer used.
27232 (bswapsi2_internal): Likewise.
27233 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
27234 store, and gpr<-gpr swap insns.
27235 (bswap<mode>2_store): Likewise.
27236 (bswaphi2_reg): Register only splitter, combine with the splitter.
27237 (bswaphi2 splitter): Likewise.
27238 (bswapsi2_reg): Likewise.
27239 (bswapsi2 splitter): Likewise.
27240 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27241 the insns into load, store, and register/register insns.
27242 (bswapdi2_ldbrx): Likewise.
27243 (bswapdi2_load): Likewise.
27244 (bswapdi2_store): Likewise.
27245 (bswapdi2_reg): Likewise.
27247 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27249 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27250 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27252 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27255 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27257 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27258 special handling for target option conflicts between dform
27259 options (-mpower9-dform, -mpower9-dform-vector,
27260 -mpower9-dform-scalar) and -mno-direct-move.
27262 2017-03-27 Richard Biener <rguenther@suse.de>
27264 PR tree-optimization/80181
27265 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27267 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27269 * config/arc/predicates.md (move_double_src_operand): Replace the
27270 call to move_double_src_operand with a call to address_operand.
27272 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27274 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27275 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27276 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27278 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27280 * config/arc/predicates.md (long_immediate_loadstore_operand):
27281 Consider scaled addresses cases.
27283 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27285 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27286 restored when in interrupt.
27287 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27288 doesn't have delay slot.
27290 2017-03-27 Richard Biener <rguenther@suse.de>
27293 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27294 inlined thunk clones.
27296 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27299 * asan.c (instrument_derefs): Copy over last operand from
27300 original COMPONENT_REF to the new COMPONENT_REF with
27301 DECL_BIT_FIELD_REPRESENTATIVE.
27302 * ubsan.c (instrument_object_size): Likewise.
27304 2017-03-27 Richard Biener <rguenther@suse.de>
27306 PR tree-optimization/80170
27307 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27308 sure DR/SCEV didnt fold in constants we do not see when looking
27309 at the reference base alignment.
27311 2017-03-27 Richard Biener <rguenther@suse.de>
27313 PR middle-end/80171
27314 * gimple-fold.c (fold_ctor_reference): Properly guard against
27315 NULL return value from canonicalize_constructor_val.
27317 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27320 * config/i386/i386.c (ix86_expand_builtin)
27321 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27322 flags reg setting and flags reg using instructions.
27323 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27324 clobbering instructions to zero extend op2.
27326 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27328 * doc/install.texi (Configuration) <--with-aix-soname>:
27329 Update link to AIX ld.
27331 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27333 PR rtl-optimization/80160
27334 PR rtl-optimization/80159
27335 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27336 reg_alternate_class into account.
27338 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27341 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27342 to consider in curr_insn_transform.
27344 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27346 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27347 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27348 and emit_mode_inner.
27350 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27352 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27353 argument to the overloaded builtin variants. Use the new flag to
27354 deprecate certain builtin variants.
27355 * config/s390/s390-builtin-types.def: Add new builtin types.
27356 * config/s390/s390-builtins.h: Support new flags field for
27357 overloaded builtins.
27358 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27359 (s390_macro_to_expand): Enable vector float data type.
27360 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27361 builtins by incrementing the __VEC__ version number.
27362 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27364 (s390_resolve_overloaded_builtin): Emit error messages depending
27365 on the builtin flags.
27366 * config/s390/s390.c (s390_expand_builtin): Support additional
27367 flags argument. Change error message to match the messages
27368 emitted in s390-c.c.
27369 * config/s390/s390.md: New UNSPEC_* constants.
27370 (op_type): Add new instruction types.
27371 * config/s390/vecintrin.h: Add new builtins and test data class
27373 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27374 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27375 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27376 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27377 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27378 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27380 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27381 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27382 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27383 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27385 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27386 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27387 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27388 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27390 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27391 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27392 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27393 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27394 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27395 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27396 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27398 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27399 ("vec_scatter_element<V_HW_4:mode>_DI")
27400 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27401 ("vec_fpint<mode>", "vflls")
27402 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27403 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27404 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27405 ("*vec_cmphe<mode>_cc"): ... these.
27407 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27408 mode constant instead of magic value.
27410 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27412 * config/s390/s390.c (s390_expand_vec_compare): Support other
27413 vector floating point modes than just V2DF.
27414 (s390_expand_vcond): Likewise.
27415 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27416 (s390_cannot_change_mode_class): Prevent mode changes between TF
27417 and V1TF in vector registers.
27418 * config/s390/s390.md (DF, SF): New mode attributes.
27419 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27420 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27421 SFmode support for VRs.
27422 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27424 (VFT, VF_HW): New mode iterators.
27425 (vw, sdx): New mode attributes.
27426 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27427 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27428 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27429 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27430 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27431 also the new vector floating point modes. Renaming to ...
27433 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27434 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27435 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27436 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27437 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27438 ("vec_unordered<mode>"): ... these.
27440 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27441 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27442 ("*vec_extendv2df"): New insn definitions.
27444 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27446 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27447 ("mulditi3_2", "*muldi3_sign"): New patterns.
27448 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27449 rename the pattern definition.
27451 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27453 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27455 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27457 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27459 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27460 instruction if possible.
27461 * config/s390/vector.md (vec_halfnumelts): New mode
27463 ("*vec_vllezlf<mode>"): New pattern.
27465 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27467 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27468 ("popcountv4si2", "popcountv2di2"): Rename to ...
27469 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27470 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27472 ("popcount<mode>2_vxe"): New pattern.
27474 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27476 * common/config/s390/s390-common.c (processor_flags_table): Add
27478 * config.gcc: Add arch12.
27479 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27480 Default to arch12 for unknown CPU model numbers.
27481 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27482 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27483 PROCESSOR_max sanity check.
27484 * config/s390/s390-opts.h (enum processor_type): Add
27486 * config/s390/s390.c (processor_table): Add arch12.
27487 (s390_expand_builtin): Add check for B_VXE flag.
27488 (s390_issue_rate): Add PROCESSOR_ARCH12.
27489 (s390_get_sched_attrmask): Likewise.
27490 (s390_get_unit_mask): Likewise.
27491 (s390_sched_score): Enable z13 scheduling for arch12.
27492 (s390_sched_reorder): Likewise.
27493 (s390_sched_variable_issue): Likewise.
27494 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27496 (s390_tune_attr): Use z13 scheduling also for arch12.
27497 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27498 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27499 (TARGET_VXE_P): New macros.
27500 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27501 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27502 * config/s390/s390.opt: Add arch12 as processor_type.
27504 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27506 * config/s390/s390.md
27507 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27508 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27509 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27511 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27512 Rename expanders to ...
27514 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27515 ("fixuns_truncdddi2_emu"): ... these.
27517 ("fixuns_trunc<mode>si2_emu"): New expander.
27519 ("*fixuns_truncdfdi2_z13"): Rename to ...
27520 ("*fixuns_truncdfdi2_vx"): ... this.
27522 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27524 * config/s390/2964.md: Remove the single element vector compare
27525 instructions which are no longer used.
27526 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27528 (s390_canonicalize_comparison): Remove handling of DFmode
27530 (s390_expand_vec_compare_scalar): Remove function.
27531 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27532 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27534 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27536 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27538 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27540 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27541 will anyway by matched by mov<mode>_64dfp.
27543 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27545 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27546 vlef/vstef. Add missing operand to vleif.
27548 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27550 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27551 pair for all vector types with 64 bit elements.
27552 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27553 * config/s390/vector.md (V_HW_64): ... here.
27554 (V_128_NOSINGLE): New mode iterator.
27555 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27556 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27557 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27558 ("*vec_load_pairv2di"): Change to ...
27559 ("*vec_load_pair<mode>"): ... this one.
27561 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27563 * config/s390/constraints.md: Add comments.
27564 (jKK): Reject element sizes > 8 bytes.
27565 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27567 * config/s390/s390.md: Add the s_operand checks formerly in
27568 s390_split_ok_p to various splitters where they are still
27570 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27571 for 128 bit vectors. Plus two splitters.
27573 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27575 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27578 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27581 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27582 error if the boundary argument is not constant.
27584 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27586 PR rtl-optimization/80112
27587 * loop-doloop.c (doloop_condition_get): Don't check condition
27588 if cmp isn't SET with IF_THEN_ELSE src.
27590 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27592 PR tree-optimization/80158
27593 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27594 replacing a candidate statement, also replace it for the
27595 candidate's alternate interpretation.
27596 (replace_rhs_if_not_dup): Likewise.
27597 (replace_one_candidate): Likewise.
27599 2017-03-24 Richard Biener <rguenther@suse.de>
27601 PR tree-optimization/80167
27602 * graphite-isl-ast-to-gimple.c
27603 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27605 (translate_isl_ast_to_gimple::get_rename): Likewise.
27607 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27609 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27610 handling of certain combinations of target options, including the
27611 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27612 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27614 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27617 * config/arm/arm.md (*load_multiple): Add reload_completed to
27618 matching condition.
27620 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27621 Richard Biener <rguenth@suse.de>
27623 PR tree-optimization/79908
27624 PR tree-optimization/80136
27625 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27626 been cast away, gimplify_and_add suffices.
27628 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27630 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27632 2017-03-23 Richard Biener <rguenther@suse.de>
27634 PR tree-optimization/80032
27635 * gimplify.c (gimple_push_cleanup): Forced unconditional
27636 cleanups still have to go to the conditional_cleanups
27639 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27641 PR tree-optimization/80072
27642 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27644 (next_operand_entry_id): Change type to unsigned int.
27645 (sort_by_operand_rank): Make sure to return the right return value
27646 even if unsigned fields are bigger than INT_MAX.
27647 (struct oecount): Change cnt and id type to unsigned int.
27648 (oecount_hasher::equal): Formatting fix.
27649 (oecount_cmp): Make sure to return the right return value
27650 even if unsigned fields are bigger than INT_MAX.
27651 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27654 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27655 TREE_READONLY on result if writing it more than once.
27658 * doc/invoke.texi (-fsanitize=thread): Document that with
27659 -fnon-call-exceptions atomics are not able to throw
27663 * tsan.c: Include tree-eh.h.
27664 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27665 maybe_clean_or_replace_eh_stmt where needed.
27666 (instrument_memory_accesses): Add cfg_changed argument.
27667 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27668 if it returned true.
27669 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27671 PR rtl-optimization/63191
27672 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
27673 wrapper function, moved the whole old content into ...
27674 (ix86_delegitimize_address_1): ... this. New inline function.
27675 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
27676 true as last argument instead of ix86_delegitimize_address.
27678 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27680 * config/aarch64/aarch64.c (generic_branch_cost): Copy
27681 cortexa57_branch_cost.
27683 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27685 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
27687 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27690 * doc/md.texi (Constraints): Document wA constraint.
27691 * config/rs6000/constraints.md (wA): New.
27692 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27693 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27694 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27695 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27697 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27700 * gimplify.c (is_oacc_declared): New function.
27701 (oacc_default_clause): Use it to set default flags for acc declared
27702 variables inside parallel regions.
27703 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27704 declared variables.
27705 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27706 declare attribute to any decl as necessary.
27708 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27711 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27712 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27713 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27714 (arm_arch_lpae): This.
27715 * config/arm/arm.c (arm_arch7ve): Rename into ...
27716 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27717 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27720 2017-03-22 Martin Liska <mliska@suse.cz>
27723 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27724 error message instead of an ICE.
27726 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27728 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27730 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27732 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27734 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27737 2017-03-21 Martin Sebor <msebor@redhat.com>
27739 * doc/extend.texi: Use "cannot" instead of "can't."
27740 * doc/hostconfig.texi: Same.
27741 * doc/install.texi: Same.
27742 * doc/invoke.texi: Same.
27743 * doc/loop.texi: Same.
27744 * doc/md.texi: Same.
27745 * doc/objc.texi: Same.
27746 * doc/rtl.texi: Same.
27747 * doc/tm.texi: Same.
27748 * doc/tm.texi.in: Same.
27749 * doc/trouble.texi: Same.
27751 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27754 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27755 (collect_checksum_attributes): Set it.
27756 (die_checksum_ordered): Use it.
27758 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27760 PR tree-optimization/79908
27761 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27762 change: For a VA_ARG whose LHS has been cast away, use
27763 force_gimple_operand to construct the side effects.
27765 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27767 PR translation/80001
27768 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27769 more amenable to translation.
27770 (oacc_loop_auto_partitions): Likewise.
27772 2017-03-21 Marek Polacek <polacek@redhat.com>
27773 Martin Sebor <msebor@redhat.com>
27775 PR tree-optimization/80109
27776 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27777 on INTEGRAL_TYPE_P.
27779 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27780 Segher Boessenkool <segher@kernel.crashing.org>
27783 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27784 check reg_used_between_p between insn and one of succ or succ2
27785 depending on if succ is artificial insn not inserted into insn
27788 2017-03-21 Martin Liska <mliska@suse.cz>
27790 PR gcov-profile/80081
27791 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27792 * doc/gcc.texi: Include gcov-dump stuff.
27793 * doc/gcov-dump.texi: New file.
27795 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27797 PR rtl-optimization/79150
27798 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27799 conditional jump, if the jump is the last insn of the loop.
27801 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27802 Richard Biener <rguenth@suse.de>
27804 PR tree-optimization/79908
27805 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27806 been cast away, use force_gimple_operand to construct the side
27809 2017-03-21 Martin Liska <mliska@suse.cz>
27811 PR libfortran/79956
27812 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27815 2017-03-21 Brad Spengler <spender@grsecurity.net>
27818 * plugin.c (htab_hash_plugin): New function.
27819 (add_new_plugin): Use it and adjust.
27820 (parse_plugin_arg_opt): Adjust.
27821 (init_one_plugin): Likewise.
27823 2017-03-21 Richard Biener <rguenther@suse.de>
27825 PR tree-optimization/80032
27826 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27827 if set force the cleanup to happen unconditionally.
27828 (gimplify_target_expr): Push inserted clobbers with force_uncond
27829 to avoid them being removed by control-dependent DCE.
27831 2017-03-21 Richard Biener <rguenther@suse.de>
27833 PR tree-optimization/80122
27834 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27835 va_arg_pack_len when the inlined call stmt requires pack
27837 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27839 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27842 * tsan.c (instrument_builtin_call): If the memory model argument
27843 is not a constant, assume it is valid.
27846 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27849 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27851 PR rtl-optimization/79910
27852 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27853 if its dest is used by an insn before I2 (other than the combined
27854 insns themselves, which are properly handled already).
27856 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27859 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27861 * combine.c (record_used_regs): New static function.
27862 (try_combine): Handle situations where there is an additional
27863 instruction between I2 and I3 which needs to have a LOG_LINK
27867 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27869 * combine.c (try_combine): Delete redundant i1 test. Call
27870 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27872 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27875 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27876 alternatives 13/14.
27878 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27880 PR tree-optimization/80054
27881 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27882 the optimization if a PHI or any of its arguments is not dominated
27883 by the candidate's basis. Use gphi* rather than gimple* as
27885 (replace_profitable_candidates): Clean up a gimple* variable that
27886 should be a gphi* variable.
27888 2017-03-20 Martin Sebor <msebor@redhat.com>
27891 * doc/extend.texi (attribute constructor): Document present limitation.
27893 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27896 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27897 __POWER9_VECTOR__ #ifdef control, change template definition to
27898 use Power9-specific built-in function.
27899 (vec_any_eq): Likewise.
27900 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27901 to control outcomes from this test.
27902 (vector_ae_<mode>p): For VEC_F modes, likewise.
27904 2017-03-20 Ian Lance Taylor <iant@google.com>
27906 * config/i386/i386.c (ix86_function_regparm): Save an extra
27907 register for -fsplit-stack with DECL_STATIC_CHAIN.
27909 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27912 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27913 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27915 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27917 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27919 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27922 2017-03-20 Marek Polacek <polacek@redhat.com>
27925 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27927 2017-03-20 Richard Biener <rguenther@suse.de>
27929 PR tree-optimization/80113
27930 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27931 allocate extra SSA name for PHI def.
27932 (add_close_phis_to_outer_loops): Likewise.
27933 (add_close_phis_to_merge_points): Likewise.
27934 (copy_loop_close_phi_args): Likewise.
27935 (copy_cond_phi_nodes): Likewise.
27937 2017-03-20 Martin Liska <mliska@suse.cz>
27939 PR middle-end/79753
27940 * tree-chkp.c (chkp_build_returned_bound): Do not build
27941 returned bounds for a LHS that's not a BOUNDED_P type.
27943 2017-03-20 Martin Liska <mliska@suse.cz>
27947 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27948 COMPLEX_CST and VECTOR_CST.
27950 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27953 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27954 target operand. A new splitter adds the clobber statement in case
27955 the target operand is dead anyway.
27957 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27959 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27960 to age-old versions of binutils and glibc.
27962 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27964 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27966 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27968 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27970 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27972 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27973 requirement for binutils 2.13.
27975 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27977 * combine.c (try_combine): Delete redundant i1 test. Call
27978 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27980 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27982 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27983 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
27985 <riscv64-*-elf>: Re-arrange section
27986 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
27987 <riscv32-*-linux>: Likewise.
27988 <riscv64-*-elf>: Likewise
27989 <riscv64-*-linux>: Likewise.
27991 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
27994 * aarch64.opt(verbose-cost-dump): Fix typo.
27996 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
27999 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28000 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28002 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28004 * reload.c (find_reloads): When reloading a nonoffsettable address,
28005 use RELOAD_OTHER for it and its address reloads.
28007 PR rtl-optimization/79910
28008 * combine.c (record_used_regs): New static function.
28009 (try_combine): Handle situations where there is an additional
28010 instruction between I2 and I3 which needs to have a LOG_LINK
28013 2017-03-17 Jeff Law <law@redhat.com>
28015 PR tree-optimization/71437
28016 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28017 conditional in the hash table first.
28018 (vrp_dom_walker::before_dom_children): Extract condition from
28019 ASSERT_EXPR. Record condition, its inverion and any implied
28020 conditions as well.
28022 2017-03-17 Marek Polacek <polacek@redhat.com>
28023 Markus Trippelsdorf <markus@trippelsdorf.de>
28025 PR tree-optimization/80079
28026 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28029 2017-03-17 Richard Biener <rguenther@suse.de>
28031 PR middle-end/80075
28032 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28033 Properly verify the LHS before the RHS possibly claims to be
28035 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28038 2017-03-17 Martin Jambor <mjambor@suse.cz>
28040 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28041 (List of -O2 options): Likewise.
28042 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28045 2017-03-17 Tom de Vries <tom@codesourcery.com>
28047 * gcov-dump.c (print_usage): Print bug_report_url.
28049 2017-03-17 Richard Biener <rguenther@suse.de>
28051 PR middle-end/80050
28052 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28053 (parser::peek): Likewise.
28055 2017-03-17 Richard Biener <rguenther@suse.de>
28057 PR tree-optimization/80048
28058 * sese.c (free_sese_info): Properly release rename_map and
28059 copied_bb_map elements.
28061 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28063 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28064 Add linked-list forward and backlinks. Insert on
28065 construction, remove on destruction.
28066 (class pass_store_merging): Add m_stores_head field.
28067 (pass_store_merging::terminate_and_process_all_chains):
28068 Iterate over m_stores_head list.
28069 (pass_store_merging::terminate_all_aliasing_chains):
28071 (pass_store_merging::execute): Check for debug stmts first.
28072 Push new chains onto the m_stores_head stack.
28074 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28077 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28078 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28079 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28081 2017-03-16 Jeff Law <law@redhat.com>
28083 PR tree-optimization/71437
28084 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28085 member function. Implementation moved into after_dom_children
28086 member function and into the threader's thread_outgoing_edges
28088 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28089 some code into new thread_outgoing_edges.
28090 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28091 definition. Simplify marker handling (do it here). Assume we always
28092 have the available expression and the const/copies tables.
28093 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28095 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28096 * tree-vrp.c (equiv_stack): No longer file scoped.
28097 (vrp_dom_walker): New class.
28098 (vrp_dom_walker::before_dom_children): New member function.
28099 (vrp_dom_walker::after_dom_children): Likewise.
28100 (identify_jump_threads): Setup domwalker. Use it rather than
28101 walking edges in a random order by hand. Simplify setup/finalization.
28102 (finalize_jump_threads): Remove.
28103 (vrp_finalize): Do not call identify_jump_threads here.
28104 (execute_vrp): Do it here instead and call thread_through_all_blocks
28107 PR tree-optimization/71437
28108 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28110 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28112 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28113 (dom_opt_dom_walker::thread_across_edge): Remove
28114 handle_dominating_asserts argument. All callers changed.
28115 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28116 changes. Remove calls to lhs_of_dominating_assert. Other
28117 uses of handle_dominating_asserts turn into unconditional code
28118 (simplify_control_stmt_condition_1): Likewise.
28119 (simplify_control_stmt_condition): Likewise.
28120 (thread_through_normal_block, thread_across_edge): Likewise.
28121 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28122 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28123 object if it is not an SSA_NAME.
28124 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28125 before calling into the VRP specific simplifiers.
28126 (identify_jump_threads): Remove handle_dominating_asserts
28129 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28132 * tree-diagnostic.c (default_tree_printer): No longer static.
28133 * tree-diagnostic.h (default_tree_printer): New prototype.
28135 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28137 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28138 Change ins into fmov.
28140 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28142 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28143 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28144 Use h_con constraint for operand 1.
28145 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28146 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28148 2017-03-15 Jeff Law <law@redhat.com>
28150 PR tree-optimization/71437
28151 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28152 (record_temporary_equivalences): Use it.
28154 PR tree-optimization/71437
28155 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28156 tree-ssa-scopedtables.
28157 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28158 (record_conditions, record_cond, vuse_eq): Likewise.
28159 (record_edge_info): Adjust to API tweak of record_conditions.
28160 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28161 (record_temporary_equivalences, optimize_stmt): Likewise.
28162 (eliminate_redundant_computations): Likewise.
28163 (record_equivalences_from_stmt): Likewise.
28164 * tree-ssa-scopedtables.c: Include options.h and params.h.
28165 (vuse_eq): New function, moved from tree-ssa-dom.c
28166 (build_and_record_new_cond): Likewise.
28167 (record_conditions): Likewise. Accept vector of conditions rather
28168 than edge_equivalence structure for first argument.
28169 for the first argument.
28170 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28171 from tree-ssa-dom.c.
28172 (avail_exprs_stack::record_cond): Likewise.
28173 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
28174 from tree-ssa-dom.c.
28175 (avail_exprs_stack): Add new member functions lookup_avail_expr
28177 (record_conditions): Declare.
28179 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
28182 * lra-constraints.c (process_alt_operands): Increase reject for
28183 reloading an input/output operand.
28185 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28188 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
28189 insns to convert from signed/unsigned char/short to IEEE 128-bit
28191 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
28193 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
28196 * config/i386/i386.c (ix86_vector_duplicate_value): Create
28197 subreg of inner mode for values already in registers.
28199 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
28201 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
28202 iteration reg is used after the loop.
28204 2017-03-14 Martin Sebor <msebor@redhat.com>
28206 PR tree-optimization/79800
28207 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
28208 precision in negative-positive range.
28209 (format_floating): Call non-const overload with adjusted precision.
28211 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28214 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
28217 2017-03-14 Martin Sebor <msebor@redhat.com>
28219 PR middle-end/80020
28220 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
28221 * builtins.def (aligned_alloc): Use it.
28224 * Makefile.in (GTFILES): Add calls.c.
28225 * calls.c: Include "gt-calls.h".
28227 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
28229 PR rtl-optimization/79728
28230 * regs.h (struct target_regs): New field
28231 x_contains_allocatable_regs_of_mode.
28232 (contains_allocatable_regs_of_mode): New macro.
28233 * reginfo.c (init_reg_sets_1): Initialize it, and change
28234 contains_reg_of_mode so it includes global regs as well.
28235 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28236 rather than contains_regs_of_mode.
28238 2017-03-14 Martin Liska <mliska@suse.cz>
28240 * doc/invoke.texi: Document options that can't be combined with
28241 -fcheck-pointer-bounds.
28243 2017-03-14 Martin Liska <mliska@suse.cz>
28245 PR middle-end/79831
28246 * doc/invoke.texi (-Wchkp): Document the option.
28248 2017-03-14 Martin Liska <mliska@suse.cz>
28250 * Makefile.in: Install gcov-dump.
28252 2017-03-14 Martin Liska <mliska@suse.cz>
28254 * multiple_target.c (expand_target_clones): Bail out for
28255 an invalid attribute.
28257 2017-03-14 Richard Biener <rguenther@suse.de>
28259 * alias.c (struct alias_set_entry): Pack properly.
28260 * cfgloop.h (struct loop): Likewise.
28261 * cse.c (struct set): Likewise.
28262 * ipa-utils.c (struct searchc_env): Likewise.
28263 * loop-invariant.c (struct invariant): Likewise.
28264 * lra-remat.c (struct cand): Likewise.
28265 * recog.c (struct change_t): Likewise.
28266 * rtl.h (struct address_info): Likewise.
28267 * symbol-summary.h (function_summary): Likewise.
28268 * tree-loop-distribution.c (struct partition): Likewise.
28269 * tree-object-size.c (struct object_size_info): Likewise.
28270 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28271 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28272 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28273 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28274 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28275 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28276 (struct _stmt_vec_info): Likewise.
28278 2017-03-14 Martin Liska <mliska@suse.cz>
28281 * multiple_target.c (create_dispatcher_calls): Check that
28282 a target can create a function dispatcher.
28284 2017-03-14 Martin Liska <mliska@suse.cz>
28287 * multiple_target.c (expand_target_clones): Drop local.local
28288 flag for default implementation.
28290 2017-03-14 Richard Biener <rguenther@suse.de>
28292 PR tree-optimization/80030
28293 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28295 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28297 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28298 gcc_fallthrough() instead of __attribute__((fallthrough));
28300 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28302 * doc/gcc.texi: Remove "up" link to (DIR).
28303 * doc/gccint.texi: Ditto.
28305 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28307 * doc/install.texi (Specific) <avr>: Remove reference to
28310 2017-03-13 Jeff Law <law@redhat.com>
28312 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28313 attribute rather than comments.
28315 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28316 match_scratch operand is highest.
28318 2017-03-13 Martin Liska <mliska@suse.cz>
28320 PR middle-end/78339
28321 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28322 is a CHKP clone, use original declaration.
28324 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28326 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28327 (arc_conditional_register_usage): Use a different allocation order
28328 when optimizing for size.
28329 * common/config/arc/arc-common.c (arc_option_optimization_table):
28330 Section anchors default on when optimizing for size.
28332 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28334 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28336 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28338 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28339 * config/arc/arc.md (cpu_facility): Add cd variant.
28340 (*movqi_insn): Add code density variant.
28341 (*movhi_insn): Likewise.
28342 (*movqi_insn): Likewise.
28343 (*addsi3_mixed): Likewise.
28344 (subsi3_insn): Likewise.
28346 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28348 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28350 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28352 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28353 expressions with MINUS and UNARY ops.
28355 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28358 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28360 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28361 between vec_select and vector argument.
28362 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28363 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28364 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28365 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28366 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28367 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28369 2017-03-13 Richard Biener <rguenther@suse.de>
28372 * params.def (vect-max-peeling-for-alignment): Fix typo.
28374 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28376 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28377 issue that only occurred with binutils below 2.18.
28379 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28381 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28382 refer to binutils 2.11/2.12 minimum.
28384 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28386 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28387 ftp.kernel.org and simplify binutils requirement.
28389 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28391 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28393 (Optimize Options): Ditto. Also remove redundancy.
28395 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28397 PR translation/79848
28398 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28400 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28401 to G_ to avoid double translation.
28403 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28405 PR translation/79923
28406 * auto-profile.c (get_combined_location): Convert leading
28407 character of diagnostics to lower case and remove trailing period.
28408 (read_profile): Likewise for various diagnostics.
28409 * config/arm/arm.c (arm_option_override): Remove trailing period
28410 from various diagnostics.
28411 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28412 (msp430_expand_delay_cycles): Likewise.
28414 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28417 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28418 full command-line argument, rather than just "str".
28419 (aarch64_validate_march): Likewise.
28420 (aarch64_validate_mtune): Likewise.
28422 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28424 PR rtl-optimization/78911
28425 * lra-assigns.c (must_not_spill_p): New function.
28426 (spill_for): Use it.
28428 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28430 PR tree-optimization/79981
28431 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28432 ATOMIC_COMPARE_EXCHANGE ifn result.
28433 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28434 IFN_ATOMIC_COMPARE_EXCHANGE.
28436 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28439 * opts.c (parse_sanitizer_options): Add missing question mark to
28440 "did you mean" message.
28442 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28444 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28446 (VMULEUH_UNS): Likewise.
28447 (VMULOUB_UNS): Likewise.
28448 (VMULOUH_UNS): Likewise.
28449 * config/rs6000/rs6000.c (builtin_function_type): Remove
28450 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28452 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28455 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28456 x with result of extra_parsing_for_operand_code_0.
28457 (function_reader::extra_parsing_for_operand_code_0): Convert
28458 return type from void to rtx, returning x. When reading
28459 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28460 larger size containing struct block_symbol.
28462 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28464 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28465 -mfloat128-hardware without -m64.
28467 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28470 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28471 entries to the case statement that marks unsigned arguments to
28472 overloaded functions.
28474 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28476 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28477 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28479 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28482 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28483 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28485 2017-03-10 Martin Liska <mliska@suse.cz>
28489 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28490 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28493 2017-03-10 Olivier Hainque <hainque@adacore.com>
28495 * tree-switch-conversion (array_value_type): Start by resetting
28496 candidate type to it's main variant.
28498 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28500 PR rtl-optimization/79909
28501 * combine.c (try_combine): Use simplify_replace_rtx on individual
28502 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28503 of the whole CALL_INSN_FUNCTION_USAGE.
28505 PR tree-optimization/79972
28506 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28507 get_range_info on SSA_NAMEs. Formatting fixes.
28509 2017-03-10 Richard Biener <rguenther@suse.de>
28510 Jakub Jelinek <jakub@redhat.com>
28512 PR tree-optimization/77975
28513 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28514 edge to be constant.
28515 (get_val_for): For constant x return it. Formatting fix.
28516 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28517 would use the same bases as the current one.
28519 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28521 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28522 instead of vec_select for V1TImode.
28523 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28525 (VSX_LE_128): Add V1TI to this mode iterator.
28526 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28527 (*vsx_le_perm_store_<mode>): Likewise.
28528 (pre-reload splitter for VSX stores): Likewise.
28529 (post-reload splitter for VSX stores): Likewise.
28530 (*vsx_xxpermdi2_le_<mode>): Likewise.
28531 (*vsx_lxvd2x2_le_<mode>): Likewise.
28532 (*vsx_stxvd2x2_le_<mode>): Likewise.
28534 2017-03-09 Michael Eager <eager@eagercon.com>
28536 Correct failures with --enable-checking=yes,rtl.
28538 * config/microblaze/microblaze.c (microblaze_expand_shift):
28539 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28540 test for const0_rtx.
28541 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28542 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28544 2017-03-09 Richard Biener <rguenther@suse.de>
28546 PR tree-optimization/79977
28547 * graphite-scop-detection.c (scop_detection::merge_sese):
28548 Handle the case of extra exits to blocks dominating the entry.
28550 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28552 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28555 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28557 PR rtl-optimization/79949
28558 * lra-constraints.c (process_alt_operands): Check memory when
28559 trying to predict a cycle. Print about the overall increase.
28561 2017-03-09 Richard Biener <rguenther@suse.de>
28563 PR middle-end/79971
28564 * gimple-expr.c (useless_type_conversion_p): Preserve
28565 TYPE_SATURATING for fixed-point types.
28567 2017-03-09 Richard Biener <rguenther@suse.de>
28570 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28571 alignment of BLKmode params.
28573 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28576 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28577 (VALL_NO_V2Q): Likewise.
28579 * config/aarch64/aarch64-simd.md
28580 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28582 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28583 VALL_NO_V2Q mode iterator.
28584 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28586 2017-03-09 Martin Liska <mliska@suse.cz>
28588 PR tree-optimization/79631
28589 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28590 tree_int_cst_sign_bit just for INTEGER constants.
28592 2017-03-09 Martin Liska <mliska@suse.cz>
28596 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28599 2017-03-09 Marek Polacek <polacek@redhat.com>
28602 * tree.c (inchash::add_expr): Handle TREE_VEC.
28604 2017-03-09 Martin Liska <mliska@suse.cz>
28607 (chkp_narrow_size_and_offset): New function.
28608 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28609 (void chkp_parse_bit_field_ref): New function.
28610 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28611 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28613 2017-03-09 Martin Liska <mliska@suse.cz>
28616 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28617 (chkp_find_bounds_1): Remove gcc_unreachable.
28619 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28622 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28623 BUILT_IN_SYNC*, determine the access type from the size suffix and
28624 always build a MEM_REF with that type. Handle forgotten
28625 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28628 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28629 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28630 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28631 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28632 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28633 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28634 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28635 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28636 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28637 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28638 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28639 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28640 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28641 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28642 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28643 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28644 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28645 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28646 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28647 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28648 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28649 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28650 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28651 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28652 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28653 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28654 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28655 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28656 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28657 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28658 definitions outside of __OPTIMIZE__ guarded section.
28661 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28662 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28663 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28664 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28667 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28669 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28670 ("vfenez<mode>"): Add missing constraints.
28672 2017-03-08 Martin Sebor <msebor@redhat.com>
28675 * config/nds32/nds32.c (nds32_option_override):
28676 Fix misspelled diagnostic.
28678 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28681 * gimplify.c (gimplify_omp_for): Replace index var in outer
28682 taskloop statement with an artificial variable and add
28683 OMP_CLAUSE_PRIVATE clause for it.
28685 2017-03-08 Richard Biener <rguenther@suse.de>
28687 PR tree-optimization/79955
28688 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28689 for accesses that are completely outside of the variable.
28691 2017-03-08 Andrew Haley <aph@redhat.com>
28693 PR tree-optimization/79943
28694 * tree-ssa-loop-split.c (compute_new_first_bound): When
28695 calculating the new upper bound, (END-BEG) should be added, not
28698 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28700 * config/avr/avr.md (setmemhi): Make sure match_dup
28701 operand number comes before match_scratch.
28703 2017-03-08 Richard Biener <rguenther@suse.de>
28705 PR tree-optimization/79920
28706 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28707 with ncopies == 1 to ...
28708 (vect_transform_slp_perm_load): ... here. Properly compute
28709 all element loads by iterating VF times over the group. Do
28710 not handle ncopies (computed in a broken way) in
28711 vect_create_mask_and_perm.
28713 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28716 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28717 is a uniform vector, use uniform_vector_p return value instead of
28718 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28720 2017-03-07 Marek Polacek <polacek@redhat.com>
28722 PR middle-end/79809
28723 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28724 (alloca_call_type): Likewise.
28726 2017-03-07 Martin Liska <mliska@suse.cz>
28728 * gcov.c (process_args): Put comment to correct location.
28730 2017-03-07 Martin Liska <mliska@suse.cz>
28732 PR middle-end/68270
28733 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28734 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28735 (chkp_narrow_bounds_for_field): Likewise.
28736 (chkp_parse_array_and_component_ref): Pass one more argument to
28739 2017-03-07 Richard Biener <rguenther@suse.de>
28741 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28744 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28746 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28747 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28749 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28752 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28753 to end of description.
28754 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28756 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28758 PR rtl-optimization/79901
28759 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28761 (*avx512f_<code><mode>3<mask_name>): ... this.
28762 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28763 iterator instead of VI8_AVX2_AVX512BW.
28765 PR rtl-optimization/79901
28766 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28767 min/max expander, expand it using expand_vec_cond_expr.
28770 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28773 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28776 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28777 to void * for PCH reasons.
28778 * dwarf2out.c (output_loc_operands, output_die): Cast
28779 v.val_vec.array to unsigned char *.
28781 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28784 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28787 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28789 PR rtl-optimization/79571
28790 * lra-constraints.c (process_alt_operands): Calculate static
28791 reject and subtract it from overall when only addresses will be
28794 2017-03-06 Julia Koval <julia.koval@intel.com>
28797 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28798 incoming stack boundary to 128 for 64-bit targets.
28800 2017-03-06 Richard Biener <rguenther@suse.de>
28802 PR tree-optimization/79894
28803 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28804 to NULL after folding it.
28806 2017-03-06 Richard Biener <rguenther@suse.de>
28808 PR tree-optimization/79824
28809 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28810 check disabling peeling for gaps.
28812 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28814 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28815 attributes): Document gettimeofday.
28817 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28819 * config/s390/s390.c (s390_option_override_internal): Set
28820 PARAM_MIN_VECT_LOOP_BOUND
28822 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28824 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28825 * config/s390/s390.md: Likewise.
28827 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28830 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28831 (<avx2_avx512>_perm<mode>): Rename to ...
28832 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28834 (<avx512>_perm<mode>_mask): Rename to ...
28835 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28837 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28838 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28839 instead of VI8F_256_512.
28840 (avx512f_perm<mode>): New define_expand.
28841 (avx512f_perm<mode>_mask): Likewise.
28842 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28843 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28845 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28847 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28848 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28850 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28852 2017-03-06 Martin Liska <mliska@suse.cz>
28855 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28856 when having a SSA NAME w/o VAR_DECL assigned to it.
28858 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28860 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28861 msa_dpsub_<su>_d): Fix MODE for vec_select.
28863 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28865 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28867 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28869 2017-03-06 Richard Biener <rguenther@suse.de>
28871 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28872 * plugin.c (register_plugin_info): Likewise.
28873 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28875 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28877 * config/i386/sse.md (sse_storehps, sse_storelps,
28878 avx_<castmode><avxsizesuffix>_<castmode>,
28879 avx512f_<castmode><avxsizesuffix>_<castmode>,
28880 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28881 in condition that at least one operand is not a MEM.
28883 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28885 PR middle-end/79805
28886 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28887 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28889 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28890 gimple_call_nothrow_p flag based on whether original builtin can throw.
28891 If it can, emit following stmts on the fallthrough edge.
28892 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28893 don't create new bb if inserting just debug stmts on the edge, try to
28894 insert them on the fallthru bb or just reset debug stmts.
28896 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28899 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28900 restore recog_data (including the operand rtxes inside it) around
28901 the call to get_insn_template.
28903 2017-03-03 Martin Sebor <msebor@redhat.com>
28905 PR tree-optimization/79699
28906 * context.c (context::~context): Free MPFR caches to avoid
28907 a memory leak on program exit.
28909 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28911 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28912 Use wide_int::ulow () instead of .elt (0).
28914 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28916 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28917 (*pushxf): Limit oF constraint to 32bit targets and add oC
28918 constraint for 64bit targets.
28919 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28920 (*pushdf): Change rmF constraint to rmC.
28922 2017-03-03 Martin Liska <mliska@suse.cz>
28924 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28925 Remove unused variable.
28927 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28930 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28931 is a memory operand, increase num_memory.
28932 (ix86_expand_args_builtin): Likewise.
28934 2017-03-03 Jan Hubicka <jh@suse.cz>
28937 * ipa-devirt.c (maybe_record_node): Properly handle
28938 __cxa_pure_virtual visibility.
28940 2017-03-03 Martin Liska <mliska@suse.cz>
28942 PR tree-optimization/79803
28943 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28945 (pass_loop_prefetch::execute): Disabled optimization if an
28946 assumption about L1 cache size is not met.
28948 2017-03-03 Martin Liska <mliska@suse.cz>
28950 PR rtl-optimization/79574
28951 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28952 (hash_scan_set): Likewise.
28953 (dump_hash_table): Likewise.
28954 (hoist_code): Likewise.
28956 2017-03-03 Richard Biener <rguenther@suse.de>
28958 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28959 in place of uhigh ().
28960 (fixed_convert_from_real): Likewise.
28962 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28965 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28967 2017-03-03 Richard Biener <rguenther@suse.de>
28969 PR middle-end/79818
28970 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28971 TYPE_OVERFLOW_UNDEFINED check.
28973 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28975 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28977 (vector_ae_<mode>_p): Likewise.
28978 (vector_nez_<mode>_p): Likewise.
28979 (vector_ne_v2di_p): Likewise.
28980 (vector_ae_v2di_p): Likewise.
28981 (vector_ne_<mode>_p): Likewise.
28982 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28984 (vsx_tsqrt<mode>2_fe): Likewise.
28986 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
28989 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
28991 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28993 PR rtl-optimization/79780
28994 * cprop.c (one_cprop_pass): When second and further conditional trap
28995 in a single basic block is turned into an unconditional trap, turn it
28996 into a deleted note to avoid RTL verification failures.
28998 2017-03-02 Richard Biener <rguenther@suse.de>
29000 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29002 2017-03-02 Richard Biener <rguenther@suse.de>
29004 PR tree-optimization/79345
29006 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29007 param and abort the walk, returning -1 if it is hit.
29008 (walk_aliased_vdefs): Take a limit param and pass it on.
29009 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29010 defaulting to 0 and return a signed int.
29011 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29012 (check_defs): New helper.
29013 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29014 about uninitialized memory.
29015 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29016 bogus uninitialized warning.
29017 (fixed_convert_from_real): Likewise.
29019 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29021 PR tree-optimization/66768
29022 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29023 iv_use if base object can't be determined.
29025 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29027 PR tree-optimization/79345
29028 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29029 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29030 (get_pattern_stats): Initialize it.
29031 * genemit.c (gen_expand): Verify match_scratch numbers come after
29032 match_operand/match_dup numbers.
29033 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29034 match_scratch numbers.
29035 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29037 * config/s390/s390.md (trunctdsd2): Likewise.
29039 2017-03-02 Richard Biener <rguenther@suse.de>
29041 * wide-int.h (wide_int_storage::operator=): Implement in terms
29044 2017-03-02 Richard Biener <rguenther@suse.de>
29046 PR tree-optimization/79777
29047 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29048 the to insert expression to sth existing.
29050 2017-03-01 Martin Sebor <msebor@redhat.com>
29052 PR middle-end/79692
29053 * gimple-ssa-sprintf.c
29054 (directive::known_width_and_precision): New function.
29055 (format_integer): Use it.
29056 (get_mpfr_format_length): Consider the full range of precision
29057 when computing %g output with the # flag. Set the likely byte
29058 count to 3 rather than 1 when precision is indeterminate.
29059 (format_floating): Correct the lower bound of precision.
29061 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29063 * doc/invoke.texi: Document default code model for 64-bit Linux.
29065 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29068 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29069 udiv rather than div since input pattern is unsigned.
29071 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29073 * config/i386/i386.c (print_reg): Warn for values of
29074 unsupported size in integer register.
29076 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29079 * config/rs6000/predicates.md (current_file_function_operand): Do
29080 not allow self calls to be local if the function is replaceable.
29082 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29085 * config/rs6000/altivec.h (vec_ctz and others): Change the
29086 preprocessor macro that controls conditional compilation from
29087 _ARCH_PWR9 to __POWER9_VECTOR__.
29088 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29089 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29090 control (instead of _ARCH_PWR9 control) so that template
29091 definition uses power9-specific function.
29092 (vec_any_eq): Likewise.
29093 (vec_all_ne): Change macro definition to use a power9-specific
29094 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29095 _ARCH_PWR9 control).
29096 (vec_any_eq) Likewise.
29097 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29098 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29099 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29100 support for xvcmpnedp instruction.
29101 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29102 macro expansion so that Power9 implementation of vec_all_ne does
29103 not use the AltiVec predicate framework.
29104 (VCMPNEH_P): Likewise.
29105 (VCMPNEW_P): Likewise.
29106 (VCMPNED_P): Likewise.
29107 (VCMPNEFP_P): Likewise.
29108 (VCMPNEDP_P): Likewise.
29109 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29110 implementation of vec_any_eq to not use AltiVec predicate
29112 (VCMPAEH_P): Likewise.
29113 (VCMPAEW_P): Likewise.
29114 (VCMPAED_P): Likewise.
29115 (VCMPAEFP_P): Likewise.
29116 (VCMPAEDP_P): Likewise.
29117 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29118 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29119 not use the AltiVec predicate framework.
29120 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29121 of vec_any_eq to not use AltiVec predicate framework.
29122 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29123 support for predefined __POWER9_VECTOR__ macro to indicate that
29124 Power9 instruction selection is enabled.
29125 (altivec_overloaded_builtins): Remove extraneous
29126 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29127 function argument types RS6000_BTI_bool_V16QI and
29128 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29129 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29130 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29131 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29132 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29133 Power9 for implementations of vec_cmpne. Change the signature for
29134 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29135 (representing vec_all_ne) to remove the previously described first
29136 argument of type RS6000_BTI_INTSI, as this was an artifact of
29137 reliance on the AltiVec predicate framework, which is no longer
29138 used in the implementation of these functions. Add
29139 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29140 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29141 since, unlike the AltiVec predicate framework implementation, we
29142 do not share function descriptors between vec_alle and vec_anyeq.
29143 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29144 set of modes that receive special treatment even when
29145 TARGET_P9_VECTOR is true. The special treatment emits code that
29146 does not depend on Power9 instructions.
29147 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29148 define_expand to not rely on AltiVec predicate framework.
29149 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29151 (vector_ne_v2di_p): Change this define_expand to not rely on
29152 AltiVec predicate framework.
29153 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29155 (vector_ne_<mode>_p): Change this define_expand to not rely on
29156 AltiVec predicate framework.
29157 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29159 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29160 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29161 define_insn pattern.
29162 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29163 define_insn pattern because the xvcmpne<VSs>. instruction is not
29165 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29166 instruction is not supported.
29168 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29170 * config/nvptx/nvptx.c: Include intl.h.
29172 2017-03-01 Martin Jambor <mjambor@suse.cz>
29175 * ipa-prop.h (ipa_bits): Removed field known.
29176 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
29177 to pointers. Adjusted their comments to warn about their sharing.
29178 (ipcp_transformation_summary): Change bits to a vector of pointers.
29179 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
29180 (ipa_get_ipa_bits_for_value): Declare.
29181 * tree-vrp.h (value_range): Mark as GTY((for_user)).
29182 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
29183 (ipa_bits_hash_table): Likewise.
29184 (ipa_vr_ggc_hash_traits): Likewise.
29185 (ipa_vr_hash_table): Likewise.
29186 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
29187 being pointers and vr_known being removed.
29188 (ipa_set_jf_unknown): Likewise.
29189 (ipa_get_ipa_bits_for_value): New function.
29190 (ipa_set_jfunc_bits): Likewise.
29191 (ipa_get_value_range): New overloaded functions.
29192 (ipa_set_jfunc_vr): Likewise.
29193 (ipa_compute_jump_functions_for_edge): Use the above functions to
29194 construct bits and vr parts of jump functions.
29195 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
29196 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29198 (ipcp_grow_transformations_if_necessary): Also allocate
29199 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29201 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
29202 them. Fix too long lines.
29203 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
29204 vr_known being removed.
29205 (ipa_read_jump_function): Use new setter functions to construct bits
29206 and vr parts of jump functions or set them to NULL.
29207 (write_ipcp_transformation_info): Adjust for bits being pointers.
29208 (read_ipcp_transformation_info): Likewise.
29209 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
29211 Include gt-ipa-prop.h.
29212 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
29214 (ipcp_store_bits_results): Likewise.
29215 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
29216 Do not write to existing jump functions but use a temporary instead.
29218 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29221 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
29222 attempt to use its first operand as BIT_FIELD_REF base.
29224 2017-03-01 Richard Biener <rguenther@suse.de>
29226 PR middle-end/79721
29227 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
29228 interpolating formula in wrapping arithmetic.
29229 (chrec_apply): Convert chrec_evaluate return value to wanted type.
29231 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29233 PR tree-optimization/79734
29234 * tree-vect-generic.c (expand_vector_condition): Optimize
29235 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29236 Handle VEC_COND_EXPR where comparison has different inner width from
29237 type's inner width.
29239 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29241 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29242 markup, and similar issues. Remove @opindex entries for things
29243 that aren't options. Add missing -mmpy-option entries.
29245 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29247 PR tree-optimization/79737
29248 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29249 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29250 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29251 instead of byte_size. Formatting fix.
29252 (shift_bytes_in_array_right): Formatting fix.
29254 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29257 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29258 condition on optimize for the leaf function test.
29260 2017-02-28 Martin Liska <mliska@suse.cz>
29263 * read-rtl-function.c (function_reader::handle_unknown_directive):
29264 Bail out when one uses -flto.
29266 2017-02-28 Martin Liska <mliska@suse.cz>
29268 * common.opt: Replace space with tabular for options of <number>
29270 * config/i386/i386.opt: Show <number> value for
29271 -mlarge-data-threshold.
29272 * opts.c (print_filtered_help): Do not display number in hexadecimal
29275 2017-02-28 Martin Liska <mliska@suse.cz>
29277 * common.opt: Fix --help=option -Q for options which are of
29280 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29282 * config/i386/i386.c (print_reg): Error out for values
29283 of 8-bit size in invalid integer register.
29285 2017-02-28 Martin Sebor <msebor@redhat.com>
29287 PR tree-optimization/79691
29288 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29290 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29293 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29294 gcc_unreachable with output_operand_lossage.
29296 2017-02-28 Richard Biener <rguenther@suse.de>
29298 PR tree-optimization/79740
29299 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29301 (visit_nary_op): Insert the nary into the hashtable if we
29302 pattern-matched sth.
29303 * tree-ssa-pre.c (eliminate_insert): Robustify.
29305 2017-02-28 Richard Biener <rguenther@suse.de>
29307 PR middle-end/79731
29308 * fold-const.c (decode_field_reference): Reject out-of-bound
29311 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29313 * config/i386/i386.c: Include intl.h.
29314 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29315 instead of just cond ? "..." : "...".
29316 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29317 * coverage.c (read_counts_file): Likewise.
29318 * omp-offload.c: Include intl.h.
29319 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29320 of just cond ? "..." : "...".
29321 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29322 of just cond ? "..." : "...".
29324 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29327 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29329 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29330 'tune for' CPU name.
29331 * config/arm/arm-cpu-data.h: Regenerated.
29333 2017-02-28 Richard Biener <rguenther@suse.de>
29335 PR tree-optimization/79732
29336 * tree-inline.c (expand_call_inline): Do not shadow var.
29338 2017-02-28 Richard Biener <rguenther@suse.de>
29340 PR tree-optimization/79723
29341 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29342 address-space properly.
29344 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29346 * doc/optinfo.texi (Optimization groups): Fix option used for
29348 * doc/invoke.texi (-fopt-info): Document "omp".
29349 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29350 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29351 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29352 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29353 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29355 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29357 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29358 OPTGROUP_OMP with "omp".
29360 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29363 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29364 for arithmetic shift of unsigned V2DI.
29366 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29368 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29369 arc/linux.h headers.
29370 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29371 (LINK_SPEC): Likewise.
29372 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29373 (EXTRA_SPECS): Likewise.
29374 (STARTFILE_SPEC): Likewise.
29375 (ENDFILE_SPEC): Likewise.
29376 (LIB_SPEC): Likewise.
29377 (TARGET_SDATA_DEFAULT): Likewise.
29378 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29379 (MULTILIB_DEFAULTS): Likewise.
29380 (DWARF2_UNWIND_INFO): Likewise.
29381 * config/arc/big.h: New file.
29382 * config/arc/elf.h: Likewise.
29383 * config/arc/linux.h: Likewise.
29384 * config/arc/t-uClibc: Remove.
29386 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29388 PR tree-optimization/77536
29389 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29390 (tree_transform_and_unroll_loop): Use above function to compute the
29391 estimated niter of unrolled loop and use it when scaling profile.
29392 Also use count info rather than frequency if it's non-zero.
29393 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29394 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29395 (vect_transform_loop): Call above function.
29397 2017-02-27 Richard Biener <rguenther@suse.de>
29399 PR tree-optimization/45397
29400 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29401 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29402 (visit_nary_op): Add pattern matching for CSEing sign-changed
29403 or truncated operations with wider ones.
29405 2017-02-27 Richard Biener <rguenther@suse.de>
29407 PR tree-optimization/79690
29408 * tree-vect-stmts.c (vectorizable_store): Use vector type
29409 built from the DR with address-space.
29411 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29413 * doc/invoke.texi (Optimize Options): Refine the description
29414 of asan-use-after-return.
29416 2017-02-25 Alan Modra <amodra@gmail.com>
29418 PR rtl-optimization/79584
29419 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29420 base, not ad->base_term, the reg within base. Remove assertion
29421 that ad->base == ad->base_term. Replace gen_int_mode using
29422 bogus mode with const0_rtx.
29424 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29426 PR middle-end/79396
29427 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29428 FMA_EXPR like tcc_binary or tcc_unary.
29430 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29433 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29435 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29436 (output_loc_operands): Handle DW_OP_call_ref and
29437 DW_OP_GNU_variable_value.
29438 (struct variable_value_struct): New type.
29439 (struct variable_value_hasher): Likewise.
29440 (variable_value_hash): New variable.
29441 (string_types): Remove.
29442 (copy_loc_descr): New function.
29443 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29444 (prepend_loc_descr_to_each): New function.
29445 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29446 instead of add_loc_descr_to_each if the first argument is single
29447 location list and the second has multiple.
29448 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29449 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29450 when looking for variable value which doesn't have other location info.
29451 (loc_list_from_tree): Formatting fix.
29452 (gen_array_type_die): Simplify DW_AT_string_length handling.
29453 (adjust_string_types): Remove.
29454 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29455 string_types. Call resolve_variable_values.
29456 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29457 (resolve_addr_in_expr): Likewise. Add A argument.
29458 (copy_deref_exprloc): Remove deref argument. Adjust for the
29459 original expression being DW_OP_GNU_variable_value with optionally
29460 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29461 optionally after it.
29462 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29463 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29464 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29465 (variable_value_hasher::hash, variable_value_hasher::equal): New
29467 (resolve_variable_value_in_expr, resolve_variable_value,
29468 resolve_variable_values, note_variable_value_in_expr,
29469 note_variable_value): New functions.
29470 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29473 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29476 * opts.h (handle_generated_option): Add GENERATED_P argument.
29477 * opts-common.c (handle_option): Adjust function comment.
29478 (handle_generated_option): Add GENERATED_P argument, pass it to
29480 (control_warning_option): Pass false to handle_generated_option
29482 * opts.c (maybe_default_option): Pass true to handle_generated_option
29484 * optc-gen.awk: Likewise.
29486 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29488 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29489 a REG, look at the REG it is a SUBREG of.
29490 (splitter for cmpeqsi_t): Ditto.
29492 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29494 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29495 the special USEs with the pattern of the insn, not the insn itself.
29497 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29500 * doc/invoke.texi: Document -mload-store-pairs.
29502 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29503 Sandra Loosemore <sandra@codesourcery.com>
29505 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29506 argument isn't a CONST_INT.
29507 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29508 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29509 (nios2_validate_compare): Bypass alternate compare logic if *op2
29510 is not a CONST_INT.
29511 (ldstwm_operation_p): Return false if first_base is not a REG or
29512 if first_offset is not a CONST_INT.
29514 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29516 * config/cris/cris.md: Use correct operand in a define_peephole2.
29518 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29520 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29522 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29524 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29525 this_insn if it is an INSN or JUMP_INSN.
29526 (force_offsettable): Look at base, not at addr.
29527 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29528 on things that aren't necessarily CONST_INTs.
29530 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29532 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29533 -mfpmath=sse is the default also for x86-32 targets with SSE2
29534 instruction set when @option{-ffast-math} is enabled
29536 2017-02-24 Jeff Law <law@redhat.com>
29538 PR rtl-optimizatoin/79286
29539 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29542 2017-02-24 Richard Biener <rguenther@suse.de>
29544 PR tree-optimization/79389
29545 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29548 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29550 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29551 function comment to reflect reality.
29552 (loop_exits_before_overflow): Fix typo in function description.
29554 2017-02-24 Richard Biener <rguenther@suse.de>
29556 PR tree-optimization/79389
29557 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29558 properly that a threading opportunity exists. Detect conditional
29559 copy/constant propagation opportunities.
29561 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29563 * config/visium/visium.md (type): Add trap.
29564 (b): New mode attribute.
29565 (*btst): Rename into...
29566 (*btst<mode>): ...this and adjust.
29567 (*cbranchsi4_btst_insn): Rename into...
29568 (*cbranch<mode>4_btst_insn): ...this and adjust.
29569 (trap): New define_insn.
29571 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29573 PR tree-optimization/79389
29574 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29575 (noce_reversed_cond_code): New function.
29576 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29577 reversed_comparison_code. Formatting fix.
29578 (noce_try_store_flag): Test rev_cond != NULL in addition to
29579 reversed_comparison_code.
29580 (noce_try_store_flag_constants): Likewise.
29581 (noce_try_store_flag_mask): Likewise.
29582 (noce_try_addcc): Use rev_cond if non-NULL instead of
29583 reversed_comparison_code.
29584 (noce_try_cmove_arith): Likewise. Formatting fixes.
29585 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29586 (noce_find_if_block): Initialize rev_cond.
29587 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29588 instead of false as last argument never attempt to reverse it
29591 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29593 PR tree-optimization/79663
29594 * tree-predcom.c (combine_chains): Process refs in reverse order
29595 only for ZERO length chains, and add explaining comment.
29597 2017-02-23 Jeff Law <law@redhat.com>
29599 PR tree-optimization/79578
29600 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29601 in call to operand_equal_p.
29603 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29606 * config/i386/cpuid.h: Fix another undefined behavior.
29608 2017-02-23 Richard Biener <rguenther@suse.de>
29610 PR tree-optimization/79683
29611 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29612 vector types for data-refs.
29614 2017-02-23 Martin Liska <mliska@suse.cz>
29616 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29618 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29620 PR middle-end/79665
29621 * internal-fn.c (get_range_pos_neg): Moved to ...
29622 * tree.c (get_range_pos_neg): ... here. No longer static.
29623 * tree.h (get_range_pos_neg): New prototype.
29624 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29625 are known to be in between 0 and signed maximum inclusive, try to
29626 expand both unsigned and signed divmod and use the cheaper one from
29629 2017-02-22 Jeff Law <law@redhat.com>
29631 PR tree-optimization/79578
29632 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29633 to compare base operands.
29635 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29638 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29639 gpc_reg_operand instead of fpr_reg_operand.
29641 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29643 * config/mips/mips.c (mips_return_in_memory): Force FP
29644 vector types to be returned in memory for o32 ABI.
29646 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29648 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29649 instead of DW_TAG_member for static data member declarations and don't
29650 set no_linkage_name for static inline data members.
29651 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29654 2017-02-22 Martin Liska <mliska@suse.cz>
29656 * doc/invoke.texi: Replace inequality signs with square brackets
29659 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29662 * lra-constraints.c (simplify_operand_subreg): Handle
29663 WORD_REGISTER_OPERATIONS targets.
29665 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29668 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29669 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29670 elimination by swapping fld*.
29672 2017-02-22 Richard Biener <rguenther@suse.de>
29674 PR tree-optimization/79673
29675 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
29676 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
29677 irrelevant address-space qualifiers and avoiding a
29678 ADDR_SPACE_CONVERT_EXPR from fold_convert.
29680 2017-02-22 Richard Biener <rguenther@suse.de>
29682 PR tree-optimization/79666
29683 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
29684 to not symbolically negate if that may introduce undefined
29687 2017-02-22 Martin Liska <mliska@suse.cz>
29690 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29691 * data-streamer-out.c (streamer_write_gcov_count_stream):
29693 * value-prof.c (stream_out_histogram_value): Make assert more
29694 precise based on type of counter.
29696 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29699 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29700 Use nonimmediate_operand instead of memory_operand for operand 1.
29701 (float-extend standard_x87sse_constant_load splitter): Ditto.
29703 2017-02-21 Jeff Law <law@redhat.com>
29705 PR tree-optimization/79621
29706 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29707 blocks with edges to themselves.
29709 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29712 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29713 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29714 Use gimple_call_builtin_p.
29717 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29718 on temporarily removed DEBUG_INSNs.
29720 PR tree-optimization/79649
29721 * tree-loop-distribution.c (classify_partition): Give up on
29722 non-generic address space loads/stores.
29724 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29726 * doc/loop.texi (Loop manipulation): Remove nonexistent
29727 tree_ssa_loop_version from the documentation.
29728 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29730 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29733 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29734 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29735 * config/rs6000/rs6000.c: Include except.h.
29736 (rs6000_expand_split_stack_prologue): Call
29737 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29739 2017-02-21 Martin Jambor <mjambor@suse.cz>
29742 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29743 have been analyzed.
29745 2017-02-21 Martin Jambor <mjambor@suse.cz>
29747 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29748 for backward compatibility only.
29749 * doc/invoke.texi (Option Summary): Remove all references to
29750 -fipa-cp-alignment.
29752 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29756 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29758 * lra-constraints.c (curr_insn_transform): Handle
29759 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29761 2017-02-21 Martin Liska <mliska@suse.cz>
29763 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29765 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29767 PR translation/79638
29768 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29770 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29773 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29774 (arm_function_ok_for_sibcall): Return false for an indirect call by
29775 descriptor if all the argument registers are used.
29776 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29777 alignment of the function.
29779 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29781 PR tree-optimization/61441
29782 * simplify-rtx.c (simplify_const_unary_operation): For
29783 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29784 the sNaN unmodified.
29786 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29788 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29789 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29790 instead of SYSTEM_HEADER_DIR.
29792 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29793 Martin Liška <mliska@suse.cz>
29795 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29796 Fix typos and grammar, use active voice, and clarify.
29798 2017-02-20 Marek Polacek <polacek@redhat.com>
29800 PR middle-end/79537
29801 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29804 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29806 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29809 * config/i386/i386.c (ix86_expand_builtin): Handle
29810 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29811 ix86_builtins_isa[fcode].isa as a requirement of those
29812 flags and any other flag in the bitmask.
29813 (ix86_init_mmx_sse_builtins): Use 0 instead of
29814 ~OPTION_MASK_ISA_64BIT as mask.
29815 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29816 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29817 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29818 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29820 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29823 * lra-constraints.c (split_reg): Check requested split mode
29824 is supported by the register.
29826 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29828 * lra-constraints.c (simplify_operand_subreg): Remove early
29831 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29834 * lra-constraints.c (curr_insn_transform): Tighten condition
29835 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29837 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29840 * lra-constraints.c (curr_insn_transform): Handle
29841 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29843 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29846 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29848 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29850 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29853 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29856 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29858 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29861 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29863 * final.c (last_columnnum, override_columnnum): New variables.
29864 (final_start_function): Set last_columnnum, pass it to begin_prologue
29865 hook and pass 0 to dwarf2out_begin_prologue.
29866 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29867 to source_line debug hook.
29868 (notice_source_line): Compute last_columnnum and for debug_column_info
29869 return true on column changes.
29870 * debug.h (struct gcc_debug_hooks): Add column argument to
29871 source_line and begin_prologue hooks.
29872 (debug_nothing_int_charstar_int_bool): Remove prototype.
29873 (debug_nothing_int_int_charstar,
29874 debug_nothing_int_int_charstar_int_bool): New prototypes.
29875 (dwarf2out_begin_prologue): Add column argument.
29876 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29877 begin_prologue hooks.
29878 (debug_nothing_int_charstar_int_bool): Remove.
29879 (debug_nothing_int_int_charstar,
29880 debug_nothing_int_int_charstar_int_bool): New functions.
29881 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29882 through to dwarf2out_source_line.
29883 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29884 (dwarf2out_source_line): Add column argument, emit it if requested.
29885 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29887 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29888 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29889 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29890 through to dwarf2out_begin_prologue.
29891 (vmsdbgout_source_line): Add column argument, pass it through to
29892 dwarf2out_source_line.
29893 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29894 dbxout_source_line caller.
29895 (dbxout_source_line): Add column argument.
29897 * common.opt (gno-column-info, gcolumn-info): New options.
29898 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29899 (check_die): Also test for multiple DW_AT_decl_column attributes.
29900 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29901 DW_AT_decl_column if requested.
29902 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29904 (gen_variable_die): Likewise.
29905 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29906 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29909 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29910 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29911 (ix86_handle_option): Handle OPT_m3dnowa.
29912 * doc/invoke.texi (-m3dnowa): Document.
29913 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29914 -m3dnowa instead of -m3dnow -march=athlon.
29917 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29918 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29920 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29923 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29924 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29925 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29926 generator for vsx_xxpermdi_<mode>_be.
29927 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29928 force big-endian semantics.
29929 (vsx_xxpermdi_<mode>_be): New define_expand with same
29930 implementation as previous version of vsx_xxpermdi_<mode>.
29932 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29934 PR tree-optimization/79327
29935 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29936 variable, its initialization and use.
29938 2017-02-17 Julia Koval <julia.koval@intel.com>
29940 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29941 (OPTION_MASK_ISA_PKU_UNSET): New.
29942 (ix86_handle_option): Handle -mrdpid.
29943 * config/i386/cpuid.h (bit_RDPID): New.
29944 * config/i386/driver-i386.c (host_detect_local_cpu):
29945 Detect RDPID feature.
29946 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29947 * config/i386/i386-c.c (ix86_target_macros_internal):
29949 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29950 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29951 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29952 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29953 * config/i386/i386.md (define_insn "rdpid"): New.
29954 * config/i386/i386.opt Add -mrdpid.
29955 * config/i386/immintrin.h (_rdpid_u32): New.
29957 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29959 PR rtl-optimization/79541
29960 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29961 instead of transforming it into USE.
29963 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29965 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29966 If HONOR_SNANS (SFmode) force the input to a register.
29967 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29968 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29969 an frsp or similar insn.
29971 2017-02-17 Martin Liska <mliska@suse.cz>
29973 PR rtl-optimization/79577
29974 * params.def (selsched-max-sched-times): Increase minimum to 1.
29976 2017-02-17 Martin Liska <mliska@suse.cz>
29978 PR rtl-optimization/79574
29979 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29981 2017-02-17 Martin Liska <mliska@suse.cz>
29983 PR tree-optimization/79529
29984 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
29985 ssa_defined_default_def_p to handle cases which are implicitly
29987 * tree-ssa.c (ssa_defined_default_def_p): New function.
29988 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
29989 which are implicitly defined.
29990 * tree-ssa.h (ssa_defined_default_def_p): Declare.
29992 2017-02-17 Richard Biener <rguenther@suse.de>
29994 PR middle-end/79576
29995 * params.def (max-ssa-name-query-depth): Limit to 10.
29997 2017-02-17 Richard Biener <rguenther@suse.de>
29999 PR tree-optimization/79552
30000 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30003 2017-02-17 Richard Biener <rguenther@suse.de>
30006 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30008 2017-02-17 Marek Polacek <polacek@redhat.com>
30010 PR middle-end/79536
30011 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30012 (fold_negate_expr): New wrapper.
30014 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30016 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30017 Correct terminology and de-emphasize pre-standard behavior.
30019 2017-02-16 Alan Modra <amodra@gmail.com>
30021 PR rtl-optimization/79286
30022 * ira.c (def_dominates_uses): New function.
30023 (update_equiv_regs): Don't create an equivalence for insns that
30024 may trap where the register def does not dominate the use.
30026 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30028 PR rtl-optimization/78127
30029 * lra.c (lra): Call lra_eliminate before finish the loop after
30032 2017-02-16 Richard Biener <rguenther@suse.de>
30034 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30036 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30037 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30038 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30039 (isl_val_int_from_wi): New function.
30040 (extract_affine_gmp): Rename to ...
30041 (extract_affine_wi): ... this, take a widest_int.
30042 (extract_affine_int): Just wrap extract_affine_wi.
30043 (add_param_constraints): Use isl_val_int_from_wi.
30044 (add_loop_constraints): Likewise, and extract_affine_wi.
30046 2017-02-15 Jeff Law <law@redhat.com>
30048 PR middle-end/79521
30049 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30050 ira_init_register_move_cost_if_necessary.
30052 2017-02-15 Martin Sebor <msebor@redhat.com>
30054 PR middle-end/32003
30055 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30056 removed in a prior commit.
30058 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30060 PR tree-optimization/79347
30061 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30062 counters during peeling.
30064 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30066 * Makefile.in (site.exp): Remove "set ISLVER".
30068 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30071 * real.c (real_from_integer): Call real_convert even for decimal.
30073 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30076 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30078 2017-02-14 Andrew Pinski <apinski@cavium.com>
30080 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30081 cores and change the partno/implementer to be correct.
30082 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30083 the 'B" as the implementer.
30084 * config/aarch64/aarch64-tune.md: Regenerate.
30086 2017-02-14 Carl Love <cel@us.ibm.com>
30088 * config/rs6000/rs6000.c: Add case statement entry to make the
30089 xvcvuxdsp built-in argument unsigned.
30090 * config/rs6000/vsx.md: Fix the source and return operand types so they
30091 match the instruction definitions from the ISA document. Fix typo
30092 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30095 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30098 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30099 member early_clobber_alts.
30100 * lra-lives.c (reg_early_clobber_p): New.
30101 (process_bb_lives): Use it.
30102 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30103 (debug_operand_data): Initialize early_clobber_alts.
30104 (setup_operand_alternative): Set up early_clobber_alts.
30105 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30106 alternatives to new_insn_reg.
30107 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30109 (lra_update_insn_regno_info): Pass the new arg.
30111 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30113 PR middle-end/79505
30114 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30115 (new_oacc_loop_raw): Don't clear already cleared fields.
30118 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30119 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30120 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30122 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30125 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30127 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30130 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30131 the extra instruction to the right place to store 128-bit constant
30134 2017-02-14 Martin Sebor <msebor@redhat.com>
30136 PR middle-end/79448
30137 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30138 warning for strings of unknown length.
30140 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30142 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30144 2017-02-14 Jeff Law <law@redhat.com>
30147 * ira-costs.c (scan_one_insn): Initialize register move costs
30148 for pseudos seen in USE/CLOBBER insns.
30150 PR tree-optimization/79095
30151 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30152 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30153 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30154 if the operands are known to be not equal, then the resulting range
30156 (intersect_ranges): If the new range is ~[0,0] and the old range is
30157 wide, then prefer ~[0,0].
30158 * tree-vrp.c (overflow_comparison_p_1): New function.
30159 (overflow_comparison_p): New function.
30160 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30161 if NAME is used in an overflow test.
30162 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30163 overflow check that can be expressed as an equality test, then adjust
30164 ops to be that equality test.
30166 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30168 * config/s390/s390-builtin-types.def: Remove flags argument.
30169 * config/s390/s390.c (s390_init_builtins): Likewise.
30171 2017-02-14 Martin Liska <mliska@suse.cz>
30173 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
30174 vector. Fix trailing white spaces.
30176 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
30178 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
30181 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30183 PR rtl-optimization/68664
30184 * config/arm/arm.c (arm_sched_can_speculate_insn):
30185 New function. Declare prototype.
30186 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30188 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30190 PR rtl-optimization/68664
30191 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
30193 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30195 2017-02-14 Amit Pawar <amit.pawar@amd.com>
30197 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
30198 max skip bytes for function, loop and jump.
30200 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30202 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
30203 ABS_EXPR for gimple dump.
30205 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30208 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
30210 PR tree-optimization/79408
30211 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
30212 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
30213 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
30214 also if rhs1 is INTEGER_CST.
30216 2017-02-14 Richard Biener <rguenther@suse.de>
30218 PR middle-end/79432
30219 * tree-into-ssa.c (insert_phi_nodes): When the function can
30220 have abnormal edges rewrite SSA names with broken use-def
30221 dominance out of SSA and register them for PHI insertion.
30223 2017-02-13 Martin Sebor <msebor@redhat.com>
30225 PR middle-end/79496
30226 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
30227 clearing info.nowrite flag when snprintf size argument is a range.
30229 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30231 * cprop.c (cprop_jump): Add missing space in string literal.
30232 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
30233 (get_constraint_for_component_ref): Likewise.
30234 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
30235 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30236 * lra-constraints.c (process_alt_operands): Likewise.
30237 * ipa-inline.c (inline_small_functions): Likewise.
30238 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30239 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30240 * trans-mem.c (diagnose_tm_1_op): Likewise.
30241 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30242 (grid_parallel_clauses_gridifiable): Likewise.
30244 * config/nvptx/mkoffload.c (process): Add space in between
30247 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30248 "MOD4_SSE_REGS" and "ALL_REGS".
30250 * spellcheck.c (test_data): Add , in between "foo" and "food".
30252 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30255 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30256 boundary crossing check and subsequent code generation agree.
30258 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30260 * config/aarch64/aarch64.c (has_memory_op): Delete.
30261 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30264 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30266 PR rtl-optimization/79388
30267 PR rtl-optimization/79450
30268 * combine.c (distribute_notes): When removing TEM_INSN for which
30269 corresponding dest has last value recorded, invalidate that last
30272 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30274 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30275 of explicit '@'. Add missing assembly comment marker on branch costs
30278 2017-02-13 Nathan Sidwell <nathan@acm.org>
30280 * gengtype-lex.l (<in_struct>): Add '/'.
30282 2017-02-13 Martin Liska <mliska@suse.cz>
30285 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30287 2017-02-13 Richard Biener <rguenther@suse.de>
30289 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30291 * configure: Re-generate.
30292 * config.in: Likewise.
30293 * graphite-dependences.c: Simplify as if
30294 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30295 * graphite-isl-ast-to-gimple.c: Likewise.
30296 * graphite-optimize-isl.c: Likewise.
30297 * graphite-poly.c: Likewise.
30298 * graphite-sese-to-poly.c: Likewise.
30299 * graphite.h: Likewise.
30300 * toplev.c: Include isl/version.h and use isl_version () for
30301 printing the ISL version.
30302 * doc/install.texi: Update ISL requirement.
30304 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30306 * doc/standards.texi (Standards): Update reference to
30309 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30311 * doc/extend.texi (Named Address Spaces): sourceware.org now
30313 * doc/install.texi (Binaries): Ditto.
30316 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30318 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30319 standard terminology "stringize"/"stringizing" throughout.
30320 * doc/cppinternals.texi: Likewise.
30322 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30324 * doc/extend.texi: Fix some spelling mistakes and typos.
30325 * doc/invoke.texi: Likewise.
30327 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30330 * params.def (inline-min-speedup) Change from 10 to 8.
30332 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30334 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30337 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30340 * ipa-inline-analysis.c (get_minimal_bb): New function.
30341 (record_modified): Use it.
30342 (remap_edge_change_prob): Handle also ancestor functions.
30344 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30346 * doc/contrib.texi (Contributors): Remove broken link into
30347 the Mauve CVS repository.
30349 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30351 PR middle-end/79454
30352 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30353 result computation whenever lhs doesn't have vector mode, not
30354 just when it has BLKmode.
30356 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30358 * doc/makefile.texi (profiledbootstrap): Refer to the
30359 installation instructions only in textual form.
30361 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30364 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30366 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30368 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30369 (Specific): Update mingw-w64 reference.
30371 (Specific): Remove broken link to Renesas RX processor.
30373 2017-02-10 Richard Biener <rguenther@suse.de>
30375 * toplev.c (process_options): Do not mention obsolete graphite
30376 options when printing sorry message about missing graphite support.
30377 Mention -floop-nest-optimize.
30379 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30381 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30382 (vtst_p16): Likewise.
30383 (vtstq_p8): Likewise.
30384 (vtstq_p16): Likewise.
30386 (vtstq_p64): Likewise.
30387 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30388 (vset_lane_p64): New.
30389 (vsetq_lane_p64): New.
30391 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30393 PR tree-optimization/79411
30394 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30395 stmt operands are SSA_NAMEs used in abnormal phis.
30396 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30399 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30402 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30405 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30407 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30409 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30411 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30415 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30416 not arbitrary TREE_CONSTANT.
30419 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30420 "omp declare target link" attribute unless is_global_var.
30421 * omp-offload.c (find_link_var_op): Likewise.
30423 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30424 Chung-Lin Tang <cltang@codesourcery.com>
30426 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30428 (gimplify_adjust_omp_clauses): Don't delete TILE.
30429 (gimplify_omp_for): Deal with TILE.
30430 * internal-fn.c (expand_GOACC_TILE): New function.
30431 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30433 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30434 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30436 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30437 avoid DIV for outermost collapse var.
30438 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30439 Remove out of date comments, fix whitespace.
30440 * omp-general.c (omp_extract_for_data): Deal with tiling.
30441 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30442 adjust OLF_DIM_BASE value.
30443 (struct omp_for_data): Add tiling field.
30444 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30445 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30446 for auto loops. Remove default auto determining, moved to
30447 oacc_loop_fixed_partitions.
30448 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30449 stmts, add e_mask field.
30450 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30451 (oacc_thread_numbers): Use oacc_dim_call.
30452 (oacc_xform_tile): New.
30453 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30454 (finish_oacc_loop): Adjust for ifns vector.
30455 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30456 add case for GOACC_TILE fns.
30457 (oacc_loop_xform_loop): Delete.
30458 (oacc_loop_process): Iterate over call list directly, and add
30459 handling for GOACC_TILE fns.
30460 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30462 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30463 vector partitioning to outer loops. Assign 2 partitions to loops
30464 when available. Add TILE handling.
30465 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30466 (execite_oacc_device_lower): Process GOACC_TILE fns,
30467 ignore unknown specs.
30468 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30469 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30470 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30472 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30474 * configure.ac (ACX_BUGURL): Update.
30475 * configure: Regenerate.
30477 2017-02-09 Richard Biener <rguenther@suse.de>
30479 PR tree-optimization/69823
30480 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30481 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30483 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30485 * config/arc/arc-c.def: Add __NPS400__ definition.
30486 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30487 (TARGET_NPS400): Define.
30489 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30491 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30493 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30494 pointer, arch_info.
30495 (arc_cpu_types): Fill the arch_info field with a pointer into the
30496 arc_arch_types table.
30497 (arc_selected_cpu): Declare.
30498 * config/arc/arc.c (arc_selected_cpu): Make global.
30499 (arc_selected_arch): Delete.
30500 (arc_base_cpu): Delete.
30501 (arc_override_options): Remove references to deleted variables,
30502 update access to arch information.
30503 (ARC_OPT): Update access to arch information.
30504 (ARC_OPTX): Likewise.
30505 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30506 (TARGET_ARC600): Update access to arch information.
30507 (TARGET_ARC601): Likewise.
30508 (TARGET_ARC700): Likewise.
30509 (TARGET_EM): Likewise.
30510 (TARGET_HS): Likewise.
30511 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30514 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30517 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30518 condition/operands for integer GE/LE/GEU/LEU operations.
30520 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30522 PR translation/79397
30523 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30526 2017-02-08 Martin Jambor <mjambor@suse.cz>
30529 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30530 whether allocation happened.
30531 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30532 nothing was allocated.
30534 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30536 PR tree-optimization/79408
30537 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30538 constant, but SSA_NAME with a known integer range, use the minimum
30539 of that range instead of op1 to determine if modulo can be replaced
30540 with its first operand.
30542 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30544 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30546 2017-02-08 Richard Biener <rguenther@suse.de>
30548 PR tree-optimization/71824
30549 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30550 Check all loops contained in the merged region.
30552 2017-02-07 Andrew Pinski <apinski@cavium.com>
30554 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30556 2017-02-07 Andrew Pinski <apinski@cavium.com>
30558 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30559 (thunderxt88): Likewise.
30560 (thunderxt81): Disable LSE and change v8.1 to v8.
30561 (thunderxt83): Likewise.
30563 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30564 Richard Biener <rguenther@suse.de>
30566 PR middle-end/79399
30567 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30568 type from int to size_t.
30569 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30571 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30573 PR rtl-optimization/79386
30574 * cprop.c (bypass_conditional_jumps): Initialize
30575 bypass_last_basic_block already before splitting bbs after
30576 unconditional traps...
30577 (bypass_conditional_jumps): ... rather than here.
30580 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30581 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30582 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30583 fix -masm=intel patterns.
30585 2017-02-07 Richard Biener <rguenther@suse.de>
30587 PR tree-optimization/79256
30588 PR middle-end/79278
30589 * builtins.c (get_object_alignment_2): Use min_align_of_type
30590 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30591 and ADJUST_FIELD_ALIGN.
30593 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30595 * doc/tm.texi: Regenerate.
30596 * stor-layout.c (layout_decl): Adjust.
30597 (update_alignment_for_field): Likewise.
30598 (place_field): Likewise.
30599 (min_align_of_type): Likewise.
30600 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30601 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30602 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30603 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30604 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30605 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30606 * config/i386/i386.c (x86_field_alignment): Likewise.
30607 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30608 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30609 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30610 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30611 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30612 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30616 2017-01-30 Richard Biener <rguenther@suse.de>
30618 PR tree-optimization/79256
30619 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30620 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30623 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30625 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30626 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30627 builtins to SImode and emit a zero-extend, if necessary.
30629 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30631 * docs/invoke.texi (RISC-V Options): Alphabetize.
30633 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30635 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30638 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30640 * config/riscv/riscv.c: New file.
30641 * common/config/riscv/riscv-common.c: Likewise.
30642 * config.gcc: Likewise.
30643 * config/riscv/constraints.md: Likewise.
30644 * config/riscv/elf.h: Likewise.
30645 * config/riscv/generic.md: Likewise.
30646 * config/riscv/linux.h: Likewise.
30647 * config/riscv/multilib-generator: Likewise.
30648 * config/riscv/peephole.md: Likewise.
30649 * config/riscv/pic.md: Likewise.
30650 * config/riscv/predicates.md: Likewise.
30651 * config/riscv/riscv-builtins.c: Likewise.
30652 * config/riscv/riscv-c.c: Likewise.
30653 * config/riscv/riscv-ftypes.def: Likewise.
30654 * config/riscv/riscv-modes.def: Likewise.
30655 * config/riscv/riscv-opts.h: Likewise.
30656 * config/riscv/riscv-protos.h: Likewise.
30657 * config/riscv/riscv.h: Likewise.
30658 * config/riscv/riscv.md: Likewise.
30659 * config/riscv/riscv.opt: Likewise.
30660 * config/riscv/sync.md: Likewise.
30661 * config/riscv/t-elf-multilib: Likewise.
30662 * config/riscv/t-linux: Likewise.
30663 * config/riscv/t-linux-multilib: Likewise.
30664 * config/riscv/t-riscv: Likewise.
30665 * configure.ac: Likewise.
30666 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30667 Waterman as RISC-V maintainers.
30668 * doc/install.texi: Add RISC-V entries.
30669 * doc/invoke.texi: Add RISC-V options section.
30670 * doc/md.texi: Add RISC-V constraints section.
30671 * configure: Regenerated.
30673 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
30676 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
30677 false values to be constant vectors with all 0 or all 1 bits set.
30678 (vcondu<mode><mode>): Likewise.
30679 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
30681 (fpmask_comparison_operator): Update comment.
30682 (vecint_comparison_operator): New predicate.
30683 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
30684 vector conditionals when the true and false values are constant
30685 vectors with all 0 bits or all 1 bits set.
30687 2017-02-06 Martin Sebor <msebor@redhat.com>
30689 PR tree-optimization/79376
30690 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30692 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30694 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30695 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30696 to simplify split condition.
30698 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30700 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30701 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30704 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30706 PR rtl-optimization/68664
30707 * target.def (can_speculate_insn): New hook.
30708 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30709 * doc/tm.texi: Regenerate.
30710 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30711 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30712 (rs6000_sched_can_speculate_insn): New function.
30714 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30716 PR tree-optimization/79284
30717 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30718 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30719 vectorizable_mask_load_store, vectorizable_operation,
30720 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30721 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30722 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30723 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30724 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30725 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30726 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30727 is_gimple_assign (stmt). Replace another such test with
30728 is_gimple_assign (stmt).
30730 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30733 * config/avr/avr.c (rtl-iter.h): Include it.
30734 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30735 (avr_legitimate_combined_insn): ...and implementation.
30737 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30739 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30740 * config/s390/s390.c (s390_const_operand_ok)
30741 (s390_canonicalize_comparison, s390_extract_part)
30742 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30743 (s390_contiguous_bitmask_p, s390_rtx_costs)
30744 (legitimize_pic_address): Likewise.
30745 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30746 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30747 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30748 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30749 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30751 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30753 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30754 REGNO($0) == REGNO($1).
30756 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30758 * config/s390/linux.h(SIZE_TYPE): Add comment.
30760 2017-02-06 Julian Brown <julian@codesourcery.com>
30761 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30762 Virendra Pathak <virendra.pathak@broadcom.com>
30764 * config/aarch64/aarch64-cores.def: Change the scheduler
30766 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30767 * config/aarch64/thunderx2t99.md: New file.
30769 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30771 * doc/standards.texi (Go Language): Update link to language
30774 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30776 * tree-eh.c (lower_resx): Sanitize profile.
30777 (cleanup_empty_eh_move_lp): Likewise.
30779 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30782 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30784 * cfgloopmanip.h (loop_version): Update prototype.
30785 * modulo-sched.c (sms_schedule): Update call of loop_version.
30786 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30787 * tree-parloops.c (gen_parallel_loop): Likewise.
30788 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30789 * tree-ssa-loop-split.c (split_loop): Likewise.
30790 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30791 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30793 2017-02-05 Martin Liska <mliska@suse.cz>
30796 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30798 (print_operand_address): Initialize a struct to zero.
30800 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30802 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30803 garbage collector only in textual form.
30805 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30807 * doc/extend.texi (x86 specific memory model extensions for
30808 transactional memory): Simplify a phrase.
30810 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30813 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30814 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30815 (atomic_storedi_1): Likewise.
30817 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30819 PR tree-optimization/79338
30820 * tree-parloops.c (gather_scalar_reductions): Don't call
30821 vect_analyze_loop_form for loop->inner before destroying loop's
30824 2017-02-03 Martin Sebor <msebor@redhat.com>
30826 PR tree-optimization/79327
30827 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30828 when precision has resulted in leading zeros.
30829 (format_integer): Adjust the likely counter to assume an unknown
30830 argument that may be zero is non-zero.
30832 2017-02-03 Jason Merrill <jason@redhat.com>
30835 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30836 avoid copying non-taken branch.
30838 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30840 PR tree-optimization/79340
30841 * tree-vect-loop.c (vectorizable_reduction): Release
30842 vec_defs elements after safe_splicing them into other vectors.
30845 PR tree-optimization/79327
30846 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30847 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30849 (format_integer): Use wide_int_to_tree instead of build_int_cst
30850 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30851 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30852 of shortest and longest sequence.
30854 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30856 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30857 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30859 2017-02-03 Walter Lee <walt@tilera.com>
30862 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30863 after initial stackframe link reg save.
30864 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30866 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30869 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30870 wu for stxssp alternative.
30872 2017-02-03 Martin Sebor <msebor@redhat.com>
30874 PR tree-optimization/79352
30875 * gimple-fold.c (get_range_strlen): Add argument.
30876 (get_range_strlen): Change return type to bool.
30877 (get_maxval_strlen): Pass in a dummy argument.
30878 * gimple-fold.h (get_range_strlen): Change return type to bool.
30879 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30880 * tree.h (array_at_struct_end_p): Add argument.
30881 * tree.c (array_at_struct_end_p): Handle it.
30883 2017-02-03 Martin Liska <mliska@suse.cz>
30886 * multiple_target.c (create_dispatcher_calls): Redirect edge
30887 from a caller of a dispatcher.
30888 (expand_target_clones): Make the clones local.
30889 (ipa_target_clone): Do both target clones and resolvers.
30890 (ipa_dispatcher_calls): Remove the pass.
30891 (pass_dispatcher_calls::gate): Likewise.
30892 (make_pass_dispatcher_calls): Likewise.
30893 * passes.def (pass_target_clone): Put as very first IPA early
30896 2017-02-03 Martin Liska <mliska@suse.cz>
30898 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30899 in case of a function with ifunc attribute.
30901 2017-02-03 Martin Liska <mliska@suse.cz>
30903 * cgraph.c (cgraph_node::dump): Dump function version info.
30904 * symtab.c (symtab_node::dump_base): Add missing new line.
30906 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30908 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30909 (ifcombine_ifandif): Use it.
30911 2017-02-03 Martin Liska <mliska@suse.cz>
30913 * doc/invoke.texi: Document default value for
30914 use-after-scope-direct-emission-threshold.
30916 2017-02-03 Martin Liska <mliska@suse.cz>
30918 PR tree-optimization/79339
30919 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30920 (format_floating): Likewise.
30922 2017-02-03 Martin Liska <mliska@suse.cz>
30925 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30927 (ipa_node_params_t::remove): Likewise.
30928 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30929 initialization from removed ipa_node_params_t::insert.
30930 (ipa_node_params::~ipa_node_params): Move from removed
30931 ipa_node_params_t::release.
30932 * symbol-summary.h (symbol_summary::m_released): New member.
30933 Do not release a summary twice. Do not allow to call finalizer
30934 for types of a summary that live in GGC memory.
30936 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30938 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30941 2017-02-02 Martin Sebor <msebor@redhat.com>
30943 PR middle-end/79275
30944 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30945 (format_string): Tighten up the range of output for non-constant
30946 strings and correct the expected range for wide non-constant strings.
30948 2017-02-02 Martin Sebor <msebor@redhat.com>
30950 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30952 PR middle-end/32003
30953 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30955 (-fdump-tree-@var): Add to index and document how to come up
30956 with pass-specific option and dump file names.
30957 (-fdump-passes): Clarify where to look for output.
30959 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30961 PR middle-end/77445
30962 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30963 statistics of the analyzed path; allow threading for speed when
30964 any of BBs along the path are optimized for speed.
30966 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30968 PR middle-end/78468
30969 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30970 settings of the virtual registers.
30973 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30975 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30976 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30979 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30981 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30982 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30984 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30986 * config/s390/s390.md: Add missing comments with the expanded
30988 * config/s390/vector.md: Likewise.
30989 * config/s390/vx-builtins.md: Likewise.
30991 2017-02-02 Jakub Jelinek <jakub@redhat.com>
30994 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
30995 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
30996 conditions on a single line.
30998 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31000 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31001 __S390_VX__ to __VX__.
31003 2017-02-01 Andrew Pinski <apinski@cavium.com>
31005 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31006 stmt_info to record_stmt_cost.
31007 (vect_get_known_peeling_cost): Pass stmt_info if known to
31009 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31010 cpu_vector_cost field into
31011 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31012 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31013 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31014 splitting of scalar_stmt_cost and vec_stmt_cost.
31015 (thunderx_vector_cost): Likewise.
31016 (cortexa57_vector_cost): LIkewise.
31017 (exynosm1_vector_cost): Likewise.
31018 (xgene1_vector_cost): Likewise.
31019 (thunderx2t99_vector_cost): Improve after the splitting of the two
31021 (aarch64_builtin_vectorization_cost): Update for the splitting of
31022 scalar_stmt_cost and vec_stmt_cost.
31024 2017-02-01 Torvald Riegel <triegel@redhat.com>
31025 Richard Henderson <rth@redhat.com>
31027 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31028 conditional on existance of a fast atomic load.
31029 * optabs-query.c (can_atomic_load_p): New function.
31030 * optabs-query.h (can_atomic_load_p): Declare it.
31031 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31032 no fast atomic load is available for the particular size of access.
31033 (expand_atomic_compare_and_swap): Likewise.
31034 (expand_atomic_load): Likewise.
31035 (expand_atomic_store): Likewise.
31036 (expand_atomic_fetch_op): Likewise.
31037 * testsuite/lib/target-supports.exp
31038 (check_effective_target_sync_int_128): Remove x86 because it provides
31039 no fast atomic load.
31040 (check_effective_target_sync_int_128_runtime): Likewise.
31042 2017-02-01 Richard Biener <rguenther@suse.de>
31044 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31045 (graphite_transform_loops): Provide opt-info for optimized nests.
31046 * tree-parloop.c (parallelize_loops): Provide opt-info for
31047 parallelized loops.
31049 2017-02-01 Richard Biener <rguenther@suse.de>
31051 PR middle-end/79315
31052 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31053 was not set before.
31055 2017-02-01 Richard Biener <rguenther@suse.de>
31057 PR tree-optimization/71824
31058 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31059 Verify the loops are valid in the merged SESE region.
31060 (scop_detection::can_represent_loop_1): Check analyzing the
31061 evolution of the number of iterations in the region succeeds.
31063 2017-01-31 Ian Lance Taylor <iant@golang.org>
31065 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31066 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31068 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31070 PR preprocessor/79210
31071 * input.c (get_substring_ranges_for_loc): Replace line_width
31072 assertion with error-handling.
31074 2017-01-31 Richard Biener <rguenther@suse.de>
31076 PR tree-optimization/77318
31077 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31078 (create_pw_aff_from_tree): Take loop parameter.
31079 (add_condition_to_pbb): Pass loop of the condition to
31080 create_pw_aff_from_tree.
31082 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31084 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31085 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31087 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31091 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31093 (convert_int_to_float128): Likewise.
31094 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31095 (convert_int_to_float128): Likewise.
31096 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31097 (UNSPEC_IEEE128_CONVERT): Likewise.
31098 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31099 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31100 Use local variables for IBM extended format.
31101 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31102 (fix_trunc<mode>si2_fprs): Likewise.
31103 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31104 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31105 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31106 to know that we can now have integers of all sizes in vector
31108 (fix<uns>_<mode>di2_hw): Likewise.
31109 (float<uns>_<mode>si2_hw): Likewise.
31110 (fix_<mode>si2_hw): Likewise.
31111 (fixuns_<mode>si2_hw): Likewise.
31112 (float<uns>_<mode>di2_hw): Likewise.
31113 (float_<mode>di2_hw): Likewise.
31114 (float_<mode>si2_hw): Likewise.
31115 (floatuns_<mode>di2_hw): Likewise.
31116 (floatuns_<mode>si2_hw): Likewise.
31117 (xscvqp<su>wz_<mode>): Delete, no longer used.
31118 (xscvqp<su>dz_<mode>): Likewise.
31119 (xscv<su>dqp_<mode>): Likewise.
31120 (ieee128_mfvsrd_64bit): Likewise.
31121 (ieee128_mfvsrd_32bit): Likewise.
31122 (ieee128_mfvsrwz): Likewise.
31123 (ieee128_mtvsrw): Likewise.
31124 (ieee128_mtvsrd_64bit): Likewise.
31125 (ieee128_mtvsrd_32bit): Likewise.
31127 2017-01-31 Martin Liska <mliska@suse.cz>
31130 * ipa-prop.c (ipa_free_all_node_params): Call release method
31131 instead of ~sumbol_summary to not to trigger double times
31134 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31136 PR tree-optimization/71691
31137 * bitmap.h (class auto_bitmap): New.
31138 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31139 is_maybe_undefined instead of ssa_undefined_value_p.
31141 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31143 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31144 __S390_ARCH_LEVEL__ to __ARCH__.
31146 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31148 PR tree-optimization/79267
31149 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31150 if should_remove_lhs_p is true.
31152 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31155 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31156 (add_alignment_attribute): New.
31157 (base_type_die): Add alignment attribute.
31158 (subrange_type_die): Likewise.
31159 (modified_type_die): Likewise.
31160 (gen_array_type_die): Likewise.
31161 (gen_descr_array_type_die: Likewise.
31162 (gen_enumeration_type_die): Likewise.
31163 (gen_subprogram_die): Likewise.
31164 (gen_variable_die): Likewise.
31165 (gen_field_die): Likewise.
31166 (gen_ptr_to_mbr_type_die): Likewise.
31167 (gen_struct_or_union_type_die): Likewise.
31168 (gen_subroutine_type_die): Likewise.
31169 (gen_typedef_die): Likewise.
31170 (base_type_cmp): Compare alignment attribute.
31172 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31175 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
31176 (setb_unsigned) New pattern for setb with CCUNS.
31177 * config/rs6000/rs6000.c (expand_block_compare): Use a different
31178 subfc./subfe sequence to avoid overflow problems. Generate a
31179 shorter sequence with cmpld/setb for power9.
31180 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
31181 for generating subfc. instruction.
31182 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
31183 now uses this instruction.
31185 2017-01-30 Ian Lance Taylor <iant@google.com>
31188 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
31189 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
31191 2017-01-30 Martin Sebor <msebor@redhat.com>
31193 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
31194 Move constant to the right of a relational operator.
31195 (get_mpfr_format_length, format_character, format_string): Ditto.
31196 (should_warn_p, maybe_warn): Same.
31198 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
31200 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
31203 * asan.c (get_translation_unit_decl): Remove function.
31204 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
31206 2017-01-30 Martin Liska <mliska@suse.cz>
31208 PR gcov-profile/79259
31209 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
31210 -fprofile-generate.
31212 2017-01-30 Martin Liska <mliska@suse.cz>
31215 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
31216 Initialize variables with NULL value.
31218 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
31221 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
31223 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
31225 2017-01-30 Richard Biener <rguenther@suse.de>
31227 PR tree-optimization/79276
31228 * tree-vrp.c (process_assert_insertions): Properly adjust common
31229 when removing a duplicate.
31231 2017-01-30 Richard Biener <rguenther@suse.de>
31233 PR tree-optimization/79256
31234 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31235 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31237 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31239 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31242 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31243 ("*r<noxa>sbg_<mode>_sll_bitmask")
31244 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31245 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31246 Use contiguous_bitmask_nowrap_operand.
31248 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31251 * config/rs6000/altivec.h (vec_xl): Revise #define.
31252 (vec_xst): Likewise.
31254 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31256 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31258 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31260 PR rtl-optimization/79194
31261 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31262 traps before call to bypass_conditional_jumps.
31264 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31266 PR tree-optimization/71374
31267 * lra-constraints.c (check_conflict_input_operands): New.
31268 (match_reload): Use it.
31270 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31273 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31274 account to calculate conflict_set.
31276 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31278 PR rtl-optimization/78559
31279 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31280 other_insn in combine.
31282 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31284 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31285 uint16_type_node for BT_UINT16.
31287 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31289 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31290 "RTL Tests" to menu.
31291 (GIMPLE Tests): New node.
31292 (RTL Tests): New node.
31294 2017-01-27 Richard Biener <rguenther@suse.de>
31296 PR tree-optimization/79245
31297 * tree-loop-distribution.c (distribute_loop): Apply cost
31298 modeling also to detected patterns.
31300 2017-01-27 Richard Biener <rguenther@suse.de>
31302 PR tree-optimization/71433
31303 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31304 (compare_assert_loc): New function.
31305 (process_assert_insertions): Sort and optimize assert locations
31306 to remove duplicates and push down identical assertions on
31307 edges to their destination block.
31309 2017-01-27 Richard Biener <rguenther@suse.de>
31311 PR tree-optimization/79244
31312 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31313 out SSA names even if abnormal.
31315 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31317 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31318 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31319 instead of MPFR_RNDN.
31321 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31324 * arm.c (arm_option_override): Don't call build_target_option_node
31325 until after doing all option overrides.
31326 (arm_valid_target_attribute_tree): Likewise.
31328 2017-01-27 Martin Liska <mliska@suse.cz>
31330 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31331 for {cd}tors and C++ {cd}tors.
31333 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31335 * config/s390/s390.md ("*setmem_long_and")
31336 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31338 2017-01-26 Martin Sebor <msebor@redhat.com>
31340 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31343 2017-01-26 Martin Sebor <msebor@redhat.com>
31345 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31346 HAVE_DFmode before using XFmode or DFmode.
31347 (parse_directive): Avoid using the z length modifier to avoid
31348 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31350 PR middle-end/78703
31351 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31352 to accept adjustment as an array.
31353 (get_int_range): New function.
31354 (struct directive): Make width and prec arrays.
31355 (directive::set_width, directive::set_precision): Call get_int_range.
31356 (format_integer, format_floating): Handle width and precision ranges.
31357 (format_string, parse_directive): Same.
31359 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31362 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31363 comdat_type_p set, just clone them, but keep the children in the
31367 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31368 which have direct callers with -fvar-tracking-assignments enabled
31370 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31371 inside of type units.
31373 2017-01-26 Martin Sebor <msebor@redhat.com>
31375 PR middle-end/78703
31376 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31378 (struct format_result): Replace number_chars, number_chars_min,
31379 and number_chars_max with a single member of struct result_range.
31381 (format_result::operator+=): Adjust.
31382 (struct fmtresult): Remove bounded. Handle likely and unlikely
31384 (fmtresult::adjust_for_width_or_precision): New function.
31385 (fmtresult:type_max_digits): New function.
31386 (bytes_remaining): Handle likely and unlikely counters.
31387 (min_bytes_remaining): Remove.
31388 (format_percent): Simplify.
31389 (format_integer, format_floating): Set likely and unlikely counters.
31390 (get_string_length, format_character, format_string): Same.
31391 (format_plain, should_warn_p): New function.
31392 (maybe_warn): Call should_warn_p. Update diagnostic messages
31393 and handle those for all directives, including plain strings.
31394 (format_directive): Handle likely and unlikely counters.
31395 Remove unnecessary quoting from diagnostics. Add an informational
31397 (add_bytes): Remove.
31398 (pass_sprintf_length::compute_format_length): Simplify.
31399 (try_substitute_return_value): Handle likely and unlikely counters.
31401 2017-01-26 Carl Love <cel@us.ibm.com>
31403 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31404 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31406 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31409 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31410 endianess for subregs into account.
31411 * lra-constraints.c (lra_constraints): Do risky transformations
31412 always on the first iteration.
31413 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31414 last_call_used_reg_set.
31415 (process_bb_lives): Define and use last_call_used_reg_set.
31416 * lra.c (lra): Always continue after lra_constraints on the first
31419 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31421 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31423 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31425 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31427 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31428 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31429 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31430 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31431 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31432 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31433 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31434 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31435 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31437 2017-01-26 Marek Polacek <polacek@redhat.com>
31440 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31441 for the third operand.
31443 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31445 PR middle-end/79236
31446 * omp-low.c (struct omp_context): Add simt_stmt field.
31447 (scan_omp_for): Return omp_context *.
31448 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31449 context to the _simt_ SIMD stmt.
31450 (lower_omp_for): For combined SIMD with sibling _simt_
31451 SIMD, make sure to use the same decls in _looptemp_
31452 clauses as in the sibling.
31454 2017-01-26 David Sherwood <david.sherwood@arm.com>
31456 PR middle-end/79212
31457 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31460 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31463 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31464 emit fld b; fld a; if possible.
31466 * brig-builtins.def: Update copyright years.
31467 * config/arm/arm_acle_builtins.def: Update copyright years.
31469 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31472 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31473 constraint instead of o for the stxsd instruction.
31475 2017-01-25 Carl Love <cel@us.ibm.com>
31477 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31478 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31480 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31482 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31484 2017-01-25 Richard Biener <rguenther@suse.de>
31486 PR tree-optimization/69264
31487 * target.def (vector_alignment_reachable): Improve documentation.
31488 * doc/tm.texi: Regenerate.
31489 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31491 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31492 earlier changes with respect to TYPE_USER_ALIGN.
31493 (vector_alignment_reachable_p): Likewise. Improve dumping.
31495 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31498 * config/arm/arm.md (xordi3): Force constant operand into a register
31501 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31503 * doc/invoke.texi (-fstore-merging): Correct default optimization
31504 levels at which it is enabled.
31505 (-O): Move -fstore-merging from list to...
31508 2017-01-25 Richard Biener <rguenther@suse.de>
31511 * omp-expand.c: Include debug.h.
31512 (expand_omp_taskreg): Make sure to generate early debug before
31513 outlining anything from a function.
31514 (expand_omp_target): Likewise.
31515 (grid_expand_target_grid_body): Likewise.
31517 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31520 * asan.c (get_translation_unit_decl): New function.
31521 (asan_add_global): Extract modules file name from globals
31522 TRANSLATION_UNIT_DECL name.
31524 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31527 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31528 for long calls with APCS frame and VFP.
31530 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31532 * cfg.c (original_copy_tables_initialized_p): New function.
31533 * cfg.h (original_copy_tables_initialized_p): New decl.
31534 * cfgrtl.c (relink_block_chain): Guard the call to
31535 free_original_copy_tables with a call to
31536 original_copy_tables_initialized_p.
31537 * cgraph.h (symtab_node::native_rtl_p): New decl.
31538 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31539 (symtab_node::needed_p): Don't assert for early assembly output
31540 for __RTL functions.
31541 (cgraph_node::finalize_function): Set "force_output" for __RTL
31543 (cgraph_node::analyze): Bail out early for __RTL functions.
31544 (analyze_functions): Update assertion to support __RTL functions.
31545 (cgraph_node::expand): Bail out early for __RTL functions.
31546 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31548 * function.h (struct function): Update comment for field
31550 * gimple-expr.c: Include "tree-pass.h".
31551 (gimple_has_body_p): Return false for __RTL functions.
31552 * Makefile.in (OBJS): Add run-rtl-passes.o.
31553 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31555 (gcc::pass_manager::get_clean_slate): New accessor.
31556 * passes.c: Include "insn-addr.h".
31557 (should_skip_pass_p): Add logging. Update logic for running
31558 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31559 property-provider override so it is only done for gimple passes.
31561 (skip_pass): New function.
31562 (execute_one_pass): Call skip_pass when skipping passes.
31563 * read-md.c (md_reader::read_char): Support filtering
31564 the input to a subset of line numbers.
31565 (md_reader::md_reader): Initialize fields
31566 m_first_line and m_last_line.
31567 (md_reader::read_file_fragment): New function.
31568 * read-md.h (md_reader::read_file_fragment): New decl.
31569 (md_reader::m_first_line): New field.
31570 (md_reader::m_last_line): New field.
31571 * read-rtl-function.c (function_reader::create_function): Only
31572 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31573 curr_properties. Set DECL_INITIAL to a dummy block.
31574 (read_rtl_function_body_from_file_range): New function.
31575 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31577 * run-rtl-passes.c: New file.
31578 * run-rtl-passes.h: New file.
31580 2017-01-24 Jeff Law <law@redhat.com>
31582 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31585 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31587 PR tree-optimization/79159
31588 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31589 (record_nonwrapping_iv): Improve boundary using above function if no
31590 value range information.
31592 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31593 Martin Jambor <mjambor@suse.cz>
31595 * brig-builtins.def: New file.
31596 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31597 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31598 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31599 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31600 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31601 * builtin-types.def (BT_INT8): New.
31602 (BT_INT16): Likewise.
31603 (BT_UINT8): Likewise.
31604 (BT_UINT16): Likewise.
31605 (BT_FN_ULONG): Likewise.
31606 (BT_FN_UINT_INT): Likewise.
31607 (BT_FN_UINT_ULONG): Likewise.
31608 (BT_FN_UINT_LONG): Likewise.
31609 (BT_FN_UINT_PTR): Likewise.
31610 (BT_FN_ULONG_PTR): Likewise.
31611 (BT_FN_INT8_FLOAT): Likewise.
31612 (BT_FN_INT16_FLOAT): Likewise.
31613 (BT_FN_UINT32_FLOAT): Likewise.
31614 (BT_FN_UINT16_FLOAT): Likewise.
31615 (BT_FN_UINT8_FLOAT): Likewise.
31616 (BT_FN_UINT64_FLOAT): Likewise.
31617 (BT_FN_UINT16_UINT32): Likewise.
31618 (BT_FN_UINT32_UINT16): Likewise.
31619 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31620 (BT_FN_INT_PTR_INT): Likewise.
31621 (BT_FN_UINT_PTR_UINT): Likewise.
31622 (BT_FN_LONG_PTR_LONG): Likewise.
31623 (BT_FN_ULONG_PTR_ULONG): Likewise.
31624 (BT_FN_VOID_UINT64_UINT64): Likewise.
31625 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31626 (BT_FN_INT8_INT8_INT8): Likewise.
31627 (BT_FN_INT16_INT16_INT16): Likewise.
31628 (BT_FN_INT_INT_INT): Likewise.
31629 (BT_FN_UINT_FLOAT_UINT): Likewise.
31630 (BT_FN_FLOAT_UINT_UINT): Likewise.
31631 (BT_FN_ULONG_UINT_UINT): Likewise.
31632 (BT_FN_ULONG_UINT_PTR): Likewise.
31633 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31634 (BT_FN_UINT_UINT_UINT): Likewise.
31635 (BT_FN_VOID_UINT_PTR): Likewise.
31636 (BT_FN_UINT_UINT_PTR: Likewise.
31637 (BT_FN_UINT32_UINT64_PTR): Likewise.
31638 (BT_FN_INT_INT_UINT_UINT): Likewise.
31639 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31640 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31641 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31642 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31643 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31644 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31645 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31646 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31647 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31648 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31649 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31650 * doc/frontends.texi: List BRIG FE.
31651 * doc/install.texi (Testing): Add BRIG tesring requirements.
31652 * doc/invoke.texi (Overall Options): Mention BRIG.
31653 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31655 2017-01-24 Richard Biener <rguenther@suse.de>
31657 PR translation/79208
31658 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31660 2017-01-24 Martin Jambor <mjambor@suse.cz>
31663 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31664 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31665 and known_contexts.
31667 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31669 PR middle-end/79123
31670 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31671 casts from signed to unsigned really don't have a range.
31673 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
31675 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
31676 GMP_RNDx for compatiblity.
31678 2017-01-24 Martin Liska <mliska@suse.cz>
31681 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
31682 that would prevent us to call alloca with -1 as argument.
31684 2017-01-24 Jakub Jelinek <jakub@redhat.com>
31686 * dwarf2out.c (output_compilation_unit_header, output_file_names):
31687 Avoid -Wformat-security warning.
31689 2017-01-23 Andrew Pinski <apinski@cavium.com>
31691 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31694 2017-01-23 Martin Sebor <msebor@redhat.com>
31696 PR middle-end/78703
31697 * gimple-ssa-sprintf.c (warn_level): New global.
31698 (format_integer): Use it here and throughout the rest of the file.
31699 Use the same switch to compute sign as base.
31700 (maybe_warn): New function.
31701 (format_directive): Factor out warnings into maybe_warn.
31702 Add debugging output. Use warn_level.
31703 (add_bytes): Use warn_level.
31704 (pass_sprintf_length::compute_format_length): Add debugging output.
31705 (try_substitute_return_value): Same.
31706 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31708 PR middle-end/78703
31709 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31710 (struct fmtresult, format_integer, format_floating): Adjust.
31711 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31712 (get_string_length, format_string,format_directive): Same.
31713 (pass_sprintf_length::compute_format_length): Same.
31714 (try_substitute_return_value): Simplify slightly.
31716 PR middle-end/78703
31717 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31718 (fmtresult::operator+=): Outlined.
31719 (struct fmtresult): Add ctors.
31720 (struct conversion_spec): Rename...
31721 (struct directive): ...to this. Add and remove data members.
31722 (directive::set_width, directive::set_precision): New functions.
31723 (format_percent): Use fmtresult ctor.
31724 (get_width_and_precision): Remove.
31725 (format_integer): Make naming changes. Avoid computing width and
31727 (format_floating): Same. Adjust indentation.
31728 (format_character, format_none): New functions.
31729 (format_string): Moved character handling to format_character.
31730 (format_directive): Remove arguments, change return type.
31731 (parse_directive): New function.
31732 (pass_sprintf_length::compute_format_length): Move directive
31733 parsing to parse_directive.
31735 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31737 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31738 (assign_assembler_name_if_needed): ... this.
31739 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31740 (assign_assembler_name_if_needed): ... this.
31741 (free_lang_data_in_cgraph): Adjust callers.
31742 * cgraphunit.c (cgraph_node::analyze): Likewise.
31743 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31746 2017-01-23 Richard Biener <rguenther@suse.de>
31748 PR tree-optimization/79088
31749 PR tree-optimization/79188
31750 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31751 resetting loop bounds after last path deletion. Reset loop
31752 bounds of the target loop, make code match the comments.
31753 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31754 Make sure loops need no fixups.
31756 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31758 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31759 exponent support with double type for first argument.
31760 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31761 type returned by __builtin_vec_extract_sig,
31762 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31763 functions from "vector int" to "vector unsigned int" or from
31764 "vector long long int" to "vector unsigned long long int".
31765 Changed type returned by __builtin_vec_extract_exp,
31766 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31767 functions from "vector int" to "vector unsigned int" or from
31768 "vector long long int" to "vector unsigned long long int".
31769 Changed return type of __builtin_vec_test_data_class,
31770 __builtin_vec_test_data_class_sp, and
31771 __builtin_vec_test_data_class_dp from "vector int" to
31772 "vector bool int" or from "vector long long int" to "vector bool
31773 long long int" and changed second argument type from "unsigned
31774 int" to "int". Added new overloaded function forms "vector float
31775 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31776 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31777 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31778 double, vector unsigned long long int)" and "vector double
31779 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31780 long int)". Changed return type of
31781 __builtin_scalar_test_data_class and
31782 __builtin_scalar_test_data_class_sp and
31783 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31784 int" and changed second argument from "unsigned int" to "int".
31785 Changed type returned by __builtin_scalar_test_neg,
31786 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31787 from "int" to "bool int". Added new overloaded function form
31788 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31789 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31790 exponent double-precision with floating point first argument.
31791 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31792 documentation of scalar_test_data_class, scalar_test_neg,
31793 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31794 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31795 vec_test_data_class built-in functions to reflect refinements in
31796 their type signatures.
31798 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31800 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31802 (aarch64_elf_asm_destructor): Likewise.
31804 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31806 PR rtl-optimization/78634
31807 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31808 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31809 * ifcvt.c (noce_try_cmove): Add missing cost check.
31811 PR rtl-optimization/71724
31812 * combine.c (if_then_else_cond): Look for situations where it is
31813 beneficial to undo the work of one of the recursive calls.
31815 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31817 PR tree-optimization/70754
31818 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31819 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31820 combined stmt before it if not NULL.
31821 (combine_chains): Process refs reversely and compute dominance point
31824 2017-01-23 Martin Liska <mliska@suse.cz>
31826 PR tree-optimization/79196
31827 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31828 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31830 (strlen_optimize_stmt): Call the renamed function.
31832 2017-01-23 Michael Matz <matz@suse.de>
31834 PR tree-optimization/78384
31835 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31837 2017-01-23 Richard Biener <rguenther@suse.de>
31839 PR tree-optimization/79186
31840 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31841 both incoming edges before moving an assert.
31843 2017-01-23 Martin Jambor <mjambor@suse.cz>
31845 * ipa-prop.c (load_from_param_1): Removed.
31846 (load_from_unmodified_param): Bits from load_from_param_1 put back
31848 (load_from_param): Removed.
31849 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31850 with stmt. Reverted back to use of load_from_unmodified_param.
31852 2017-01-23 Martin Jambor <mjambor@suse.cz>
31855 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31856 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31857 field a pointer to garbage collected vector, mark lattices and
31858 ipcp_orig_node with GTY((skip)).
31859 (ipa_get_param_count): Adjust to descriptors being a pointer.
31860 (ipa_get_param): Likewise.
31861 (ipa_get_type): Likewise.
31862 (ipa_get_param_move_cost): Likewise.
31863 (ipa_set_param_used): Likewise.
31864 (ipa_get_controlled_uses): Likewise.
31865 (ipa_set_controlled_uses): Likewise.
31866 (ipa_is_param_used): Likewise.
31867 (ipa_node_params_t): Move into garbage collector. New methods insert
31869 (ipa_node_params_sum): Annotate wth GTY(()).
31870 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31872 (ipa_load_from_parm_agg): Adjust declaration.
31873 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31874 * ipa-profile.c (ipa_profile): Likewise.
31875 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31876 (ipa_populate_param_decls): Make descriptors parameter garbage
31878 (ipa_dump_param): Adjust to descriptors being a pointer.
31879 (ipa_alloc_node_params): Likewise.
31880 (ipa_initialize_node_params): Likewise.
31881 (load_from_param_1): Make descriptors parameter garbage collected.
31882 (load_from_unmodified_param): Likewise.
31883 (load_from_param): Likewise.
31884 (ipa_load_from_parm_agg): Likewise.
31885 (ipa_node_params::~ipa_node_params): Removed.
31886 (ipa_free_all_node_params): Remove call to delete operator.
31887 (ipa_node_params_t::insert): New.
31888 (ipa_node_params_t::remove): Likewise.
31889 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31890 copy known_csts and known_contexts vectors.
31891 (ipa_read_node_info): Adjust to descriptors being a pointer.
31892 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31894 (ipcp_transform_function): Make descriptors variable garbage
31897 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31899 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31900 * config/i386/avx512dqintrin.h: Ditto.
31901 * config/i386/avx512fintrin.h: Ditto.
31902 * config/i386/i386.c: Handle new builtins.
31903 * config/i386/i386-builtin.def: Add new builtins.
31904 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31905 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31907 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31908 Martin Liska <mliska@suse.cz>
31910 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31911 * asan.c (asan_expand_poison_ifn): Support stores and use
31912 appropriate ASAN report function.
31913 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31914 * internal-fn.def (ASAN_POISON_USE): Declare.
31915 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31916 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31917 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31918 ASAN_POISON calls w/o LHS.
31919 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31920 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31921 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31922 * gimplify.c (asan_poison_variables): Add attribute
31923 use_after_scope_memory to variables that really needs to live
31925 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31926 having the attribute.
31928 2017-01-23 Martin Liska <mliska@suse.cz>
31930 * asan.c (create_asan_shadow_var): New function.
31931 (asan_expand_poison_ifn): Likewise.
31932 * asan.h (asan_expand_poison_ifn): New declaration.
31933 * internal-fn.c (expand_ASAN_POISON): Likewise.
31934 * internal-fn.def (ASAN_POISON): New builtin.
31935 * sanopt.c (pass_sanopt::execute): Expand
31936 asan_expand_poison_ifn.
31937 * tree-inline.c (copy_decl_for_dup_finish): Make function
31939 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31940 * tree-ssa.c (is_asan_mark_p): New function.
31941 (execute_update_addresses_taken): Rewrite local variables
31942 (identified just by use-after-scope as addressable) into SSA.
31944 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31946 * doc/install.texi (Specific): opensource.apple.com uses https
31947 now. Remove trailing slash.
31949 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31951 * README.Portability: Remove note on an Irix compatibility issue.
31953 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31955 * gcov.c (INCLUDE_ALGORITHM): Define.
31956 (INCLUDE_VECTOR): Define.
31957 No longer include <vector> and <algorithm> directly.
31959 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31961 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31963 * doc/invoke.texi (Code Gen Options): Ditto.
31965 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31968 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31970 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31972 rtl-optimization/79125
31973 * cprop.c (local_cprop_pass): Handle cases where we make an
31974 unconditional trap.
31976 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31980 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31981 read from, for big endian.
31983 2017-01-20 Jiong Wang <jiong.wang@arm.com>
31985 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
31986 register pauth builtins for LP64 only.
31988 2017-01-20 Marek Polacek <polacek@redhat.com>
31991 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
31994 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31996 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
31998 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
31999 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32000 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32002 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32005 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32006 in big-endian mode when they are not a single duplicated value.
32008 2017-01-20 Richard Biener <rguenther@suse.de>
32010 * BASE-VER: Bump to 7.0.1.
32012 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32014 * omp-low.c (omplow_simd_context): New struct. Use it...
32015 (lower_rec_simd_input_clauses): ...here and...
32016 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32017 references to idx, lane, max_vf, is_simt.
32019 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32021 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32024 2017-01-20 Martin Jambor <mjambor@suse.cz>
32026 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32027 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32029 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32030 (GTFILES): Rename hsa.c to hsa-common.c.
32031 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32032 * hsa-dump.c: Likewise.
32033 * hsa-gen.c: Likewise.
32034 * hsa-regalloc.c: Likewise.
32035 * ipa-hsa.c: Likewise.
32036 * omp-expand.c: Likewise.
32037 * omp-low.c: Likewise.
32038 * toplev.c: Likewise.
32040 2017-01-20 Marek Polacek <polacek@redhat.com>
32043 * doc/invoke.texi: Document -Wduplicated-branches.
32044 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32045 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32046 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32047 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32048 return 0 only when not OEP_LEXICOGRAPHIC.
32049 (fold_build_cleanup_point_expr): Use the expression
32050 location when building CLEANUP_POINT_EXPR.
32051 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32052 * tree.c (add_expr): Handle error_mark_node.
32054 2017-01-20 Martin Liska <mliska@suse.cz>
32057 * tree-profile.c (init_ic_make_global_vars): Do not call
32059 (gimple_init_gcov_profiler): Likewise.
32061 2017-01-20 Martin Liska <mliska@suse.cz>
32064 * cgraph.h (maybe_create_reference): Remove argument and
32066 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32068 * ipa-cp.c (create_specialized_node): Likewise.
32069 * symtab.c (symtab_node::maybe_create_reference): Handle
32070 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32072 2017-01-20 Martin Liska <mliska@suse.cz>
32074 * read-rtl-function.c (function_reader::create_function): Use
32075 build_decl instread of build_decl_stat.
32077 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32079 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32080 * config/i386/avx512dqintrin.h: Ditto.
32081 * config/i386/avx512fintrin.h: Ditto.
32082 * config/i386/i386-builtin-types.def: Add new types.
32083 * config/i386/i386.c: Handle new types.
32084 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32085 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32086 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32087 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32088 (__builtin_ia32_kshiftridi): New.
32089 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32091 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32095 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32096 define to rs6000_init_stack_protect_guard.
32097 (rs6000_init_stack_protect_guard): New function.
32099 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32100 Yunqiang Su <yunqiang.su@imgtec.com>
32102 * config.gcc (supported_defaults): Add madd4.
32103 (with_madd4): Add validation.
32104 (all_defaults): Add madd4.
32105 * config/mips/mips.opt (mmadd4): New option.
32106 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32108 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32110 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32111 (ISA_HAS_FUSED_MADD4): Likewise.
32112 * doc/invoke.texi (-mmadd4): Document the new option.
32113 * doc/install.texi (--with-madd4): Document the new option.
32115 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32117 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32118 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32119 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32120 (aarch64_init_pauth_hint_builtins): New.
32121 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32122 (aarch64_expand_builtin): Expand new builtins.
32124 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32126 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32127 * combine-stack-adj.c (no_unhandled_cfa): Handle
32128 REG_CFA_TOGGLE_RA_MANGLE.
32129 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32130 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32131 info for return address signing.
32132 (aarch64_expand_epilogue): Likewise.
32134 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32136 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32137 * config/aarch64/aarch64-protos.h
32138 (aarch64_return_address_signing_enabled): New declaration.
32139 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32141 (aarch64_expand_prologue): Sign return address before it's pushed onto
32143 (aarch64_expand_epilogue): Authenticate return address fetched from
32145 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32146 (aarch64_attributes): New function attributes for "sign-return-address".
32147 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32148 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32149 ("*do_return"): Generate combined instructions according to key index.
32150 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32151 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32153 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32154 * config/aarch64/aarch64.opt (msign-return-address=): New.
32155 * doc/extend.texi (AArch64 Function Attributes): Documents
32156 "sign-return-address=".
32157 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32159 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32161 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32162 overall option summary.
32164 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32166 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32167 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32168 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32169 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32171 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
32173 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
32174 -mpower9-minmax by default for -mcpu=power9.
32175 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
32176 128-bit floating point.
32178 2017-01-20 Alan Modra <amodra@gmail.com>
32180 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
32181 optimizing for size.
32183 2017-01-20 Alan Modra <amodra@gmail.com>
32186 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
32187 for strcmp and strncmp from corresponding builtin decl.
32189 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32191 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
32192 instead of i386/rtems-64.h.
32193 * config/i386/rtems-64.h: Remove.
32195 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32199 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
32201 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
32203 2017-01-19 Tamar Christina <tamar.christina@arm.com>
32205 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
32206 Change int to HOST_WIDE_INT.
32207 * config/aarch64/aarch64-protos.h
32208 (aarch64_simd_gen_const_vector_dup): Likewise.
32209 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
32211 2017-01-19 David Malcolm <dmalcolm@redhat.com>
32213 * langhooks-def.h (lhd_type_for_size): New decl.
32214 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
32215 * langhooks.c (lhd_type_for_size): New function, taken from
32218 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
32220 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
32221 define_bypass for CR latency.
32222 (power9-cracked-alu): Update bypass latency and remove power9-branch.
32223 (power9-alu2): Add define_bypass for CR latency.
32225 (power9-mul): Update insn latency.
32226 (power9-mul-compare): Update insn latency, bypass latency and remove
32229 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32231 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
32233 * config/aarch64/aarch64.md
32234 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
32235 aarch64_nopcrelative_literal_loads.
32236 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32238 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32240 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32241 TARGET_LOONGSON_3A.
32242 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32244 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32247 * config.gcc (supported_defaults): Add lxc1-sxc1.
32248 (with_lxc1_sxc1): Add validation.
32249 (all_defaults): Add lxc1-sxc1.
32250 * config/mips/mips.opt (mlxc1-sxc1): New option.
32251 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32253 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32254 __mips_no_lxc1_sxc1.
32255 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32256 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32257 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32259 2017-01-19 Richard Biener <rguenther@suse.de>
32261 PR tree-optimization/72488
32262 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32263 sure to restore SSA info.
32264 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32266 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32268 PR rtl-optimization/79121
32269 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32270 of the inner type when shifting an extended value.
32272 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32275 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32276 interposable aliases.
32278 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32281 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32282 Use the evmergelohi instruction.
32283 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32284 (mov_sitf_e500_subreg8_2_be): Likewise.
32285 (mov_sitf_e500_subreg12_2_le): Likewise.
32286 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32287 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32288 (mov_sitf_e500_subreg8_2_le): Likewise.
32289 (mov_sitf_e500_subreg12_2_be): Likewise.
32291 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32293 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32294 attribute from vecsimple to vecperm.
32295 (altivec_vbpermq2): Likewise.
32297 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32300 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32302 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32303 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32304 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32305 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32306 case where N arg is SIZE_MAX.
32307 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32308 (cmpstrsi): Add pattern.
32310 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32312 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32313 __builtin_vec_revb builtins.
32314 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32315 built-in functions to support generation of the ISA 3.0 XXBR<x>
32316 vector byte reverse instructions.
32317 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32318 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32319 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32320 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32321 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32322 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32323 (P9V_BUILTIN_VEC_REVB): Likewise.
32324 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32325 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32326 (p9_xxbrq_v16qi): Likewise.
32327 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32328 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32329 (p9_xxbrh_v8hi): Likewise.
32330 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32331 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32332 vec_revb built-in functions.
32334 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32336 PR rtl-optimization/78952
32337 * config/i386/i386.md (any_extract): New code iterator.
32338 (*insvqi_2): Use any_extract for source operand.
32339 (*insvqi_3): Use any_shiftrt for source operand.
32341 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32343 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32345 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32347 2017-01-18 Matthias Klose <doko@ubuntu.com>
32349 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32351 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32353 * config/rs6000/altivec.h (vec_bperm): Change #define.
32354 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32355 (altivec_vbpermq2): New define_insn.
32356 (altivec_vbpermd): Likewise.
32357 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32358 function interface.
32359 (VBPERMD): Likewise.
32360 (VBPERM): New polymorphic function interface.
32361 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32362 Add entries for P9V_BUILTIN_VEC_VBPERM.
32363 * doc/extend.texi: Add interfaces for vec_bperm.
32365 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32367 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32368 first letter of error messages.
32369 (s390_resolve_overloaded_builtin): Likewise.
32370 * config/s390/s390.c (s390_expand_builtin): Likewise.
32371 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32372 (s390_valid_target_attribute_inner_p): Likewise.
32373 * config/s390/s390.md ("tabort"): Likewise.
32375 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32377 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32378 (ISA_AVOID_DIV_HILO): New macro.
32379 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32380 (ISA_HAS_DDIV): Likewise.
32382 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32384 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32386 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32388 * doc/invoke.texi (fabi-version): Spelling fix.
32390 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32393 * doc/invoke.texi (fabi-version): Mention mangling fix for
32396 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32399 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32401 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32404 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32405 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32407 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32409 (stack_protect_test): Ditto.
32410 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32411 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32413 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32414 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32415 -mstack-protector-guard-offset=.
32416 (RS/6000 and PowerPC Options): Ditto.
32418 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32420 * config/i386/i386.h (MASK_CLASS_P): New define.
32421 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32422 there are no registers from different register sets also when
32423 mask registers are used. Update function comment.
32424 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32425 to (*k/*r) and (*k/*km) alternatives.
32427 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32429 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32430 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32431 (EH_RETURN_HANDLER_RTX): New define.
32432 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32433 Force frame pointer in EH return functions.
32434 (aarch64_expand_epilogue): Add barrier for eh_return.
32435 (aarch64_final_eh_return_addr): Remove.
32436 (aarch64_eh_return_handler_rtx): New function.
32437 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32439 (aarch64_eh_return_handler_rtx): New prototype.
32441 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32443 * config/rs6000/altivec.h (vec_rlmi): New #define.
32444 (vec_vrlnm): Likewise.
32445 (vec_rlnm): Likewise.
32446 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32447 (UNSPEC_VRLNM): Likewise.
32448 (VIlong): New mode iterator.
32449 (altivec_vrl<VI_char>mi): New define_insn.
32450 (altivec_vrl<VI_char>nm): Likewise.
32451 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32453 (VRLDNM): Likewise.
32454 (RLNM): New polymorphic function entry.
32455 (VRLWMI): New monomorphic function entry.
32456 (VRLDMI): Likewise.
32457 (RLMI): New polymorphic function entry.
32458 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32459 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32460 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32463 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32466 * dwarf2out.c (field_byte_offset): Restore the
32467 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32468 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32469 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32472 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32475 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32477 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32480 * dwarf2out.c (add_data_member_location_attribute): For constant
32481 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32482 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32483 DW_AT_byte_size attributes.
32485 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32487 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32488 after forcing to constant memory when the code model is medium.
32490 2017-01-17 Julia Koval <julia.koval@intel.com>
32493 * config/i386/avx512fintrin.h
32494 (_mm512_i32gather_ps): Change __addr type to void const*.
32495 (_mm512_mask_i32gather_ps): Ditto.
32496 (_mm512_i32gather_pd): Ditto.
32497 (_mm512_mask_i32gather_pd): Ditto.
32498 (_mm512_i64gather_ps): Ditto.
32499 (_mm512_mask_i64gather_ps): Ditto.
32500 (_mm512_i64gather_pd): Ditto.
32501 (_mm512_mask_i64gather_pd): Ditto.
32502 (_mm512_i32gather_epi32): Ditto.
32503 (_mm512_mask_i32gather_epi32): Ditto.
32504 (_mm512_i32gather_epi64): Ditto.
32505 (_mm512_mask_i32gather_epi64): Ditto.
32506 (_mm512_i64gather_epi32): Ditto.
32507 (_mm512_mask_i64gather_epi32): Ditto.
32508 (_mm512_i64gather_epi64): Ditto.
32509 (_mm512_mask_i64gather_epi64): Ditto.
32510 (_mm512_i32scatter_ps): Change __addr type to void*.
32511 (_mm512_mask_i32scatter_ps): Ditto.
32512 (_mm512_i32scatter_pd): Ditto.
32513 (_mm512_mask_i32scatter_pd): Ditto.
32514 (_mm512_i64scatter_ps): Ditto.
32515 (_mm512_mask_i64scatter_ps): Ditto.
32516 (_mm512_i64scatter_pd): Ditto.
32517 (_mm512_mask_i64scatter_pd): Ditto.
32518 (_mm512_i32scatter_epi32): Ditto.
32519 (_mm512_mask_i32scatter_epi32): Ditto.
32520 (_mm512_i32scatter_epi64): Ditto.
32521 (_mm512_mask_i32scatter_epi64): Ditto.
32522 (_mm512_i64scatter_epi32): Ditto.
32523 (_mm512_mask_i64scatter_epi32): Ditto.
32524 (_mm512_i64scatter_epi64): Ditto.
32525 (_mm512_mask_i64scatter_epi64): Ditto.
32526 * config/i386/avx512pfintrin.h
32527 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32528 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32529 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32530 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32531 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32532 (_mm512_prefetch_i32scatter_ps): Ditto.
32533 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32534 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32535 (_mm512_prefetch_i64scatter_pd): Ditto.
32536 (_mm512_prefetch_i64scatter_ps): Ditto.
32537 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32538 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32539 * config/i386/avx512vlintrin.h
32540 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32541 (_mm_mmask_i32gather_ps): Ditto.
32542 (_mm256_mmask_i32gather_pd): Ditto.
32543 (_mm_mmask_i32gather_pd): Ditto.
32544 (_mm256_mmask_i64gather_ps): Ditto.
32545 (_mm_mmask_i64gather_ps): Ditto.
32546 (_mm256_mmask_i64gather_pd): Ditto.
32547 (_mm_mmask_i64gather_pd): Ditto.
32548 (_mm256_mmask_i32gather_epi32): Ditto.
32549 (_mm_mmask_i32gather_epi32): Ditto.
32550 (_mm256_mmask_i32gather_epi64): Ditto.
32551 (_mm_mmask_i32gather_epi64): Ditto.
32552 (_mm256_mmask_i64gather_epi32): Ditto.
32553 (_mm_mmask_i64gather_epi32): Ditto.
32554 (_mm256_mmask_i64gather_epi64): Ditto.
32555 (_mm_mmask_i64gather_epi64): Ditto.
32556 (_mm256_i32scatter_ps): Change __addr type to void*.
32557 (_mm256_mask_i32scatter_ps): Ditto.
32558 (_mm_i32scatter_ps): Ditto.
32559 (_mm_mask_i32scatter_ps): Ditto.
32560 (_mm256_i32scatter_pd): Ditto.
32561 (_mm256_mask_i32scatter_pd): Ditto.
32562 (_mm_i32scatter_pd): Ditto.
32563 (_mm_mask_i32scatter_pd): Ditto.
32564 (_mm256_i64scatter_ps): Ditto.
32565 (_mm256_mask_i64scatter_ps): Ditto.
32566 (_mm_i64scatter_ps): Ditto.
32567 (_mm_mask_i64scatter_ps): Ditto.
32568 (_mm256_i64scatter_pd): Ditto.
32569 (_mm256_mask_i64scatter_pd): Ditto.
32570 (_mm_i64scatter_pd): Ditto.
32571 (_mm_mask_i64scatter_pd): Ditto.
32572 (_mm256_i32scatter_epi32): Ditto.
32573 (_mm256_mask_i32scatter_epi32): Ditto.
32574 (_mm_i32scatter_epi32): Ditto.
32575 (_mm_mask_i32scatter_epi32): Ditto.
32576 (_mm256_i32scatter_epi64): Ditto.
32577 (_mm256_mask_i32scatter_epi64): Ditto.
32578 (_mm_i32scatter_epi64): Ditto.
32579 (_mm_mask_i32scatter_epi64): Ditto.
32580 (_mm256_i64scatter_epi32): Ditto.
32581 (_mm256_mask_i64scatter_epi32): Ditto.
32582 (_mm_i64scatter_epi32): Ditto.
32583 (_mm_mask_i64scatter_epi32): Ditto.
32584 (_mm256_i64scatter_epi64): Ditto.
32585 (_mm256_mask_i64scatter_epi64): Ditto.
32586 (_mm_i64scatter_epi64): Ditto.
32587 (_mm_mask_i64scatter_epi64): Ditto.
32588 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32589 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32590 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32591 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32592 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32593 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32594 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32595 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32596 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32597 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32598 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32599 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32600 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32601 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32602 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32603 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32604 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32605 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32606 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32607 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32608 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32609 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32610 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32611 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32612 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32613 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32614 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32615 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32616 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32617 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32618 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32619 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32620 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32621 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32622 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32623 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32624 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32625 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32626 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32627 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32628 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32629 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32630 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32631 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32632 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32633 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32634 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32635 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32636 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32637 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32638 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32639 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32640 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32641 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32642 definitions accordingly.
32644 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32645 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32648 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32651 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32654 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32655 hard regs for inner regno.
32657 2017-01-17 Martin Liska <mliska@suse.cz>
32660 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32661 assumption and add comment.
32663 2017-01-17 Nathan Sidwell <nathan@acm.org>
32665 * ipa-visibility.c (localize_node): New function, broken out of ...
32666 (function_and_variable_visibility): ... here. Call it.
32668 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32670 PR middle-end/77445
32671 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32672 correctly set frequency of oudgoing edge.
32673 (duplicate_thread_path): Fix profile updating.
32675 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32678 * configure.ac: Add GCC_BASE_VER.
32679 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
32680 version from BASE-VER file.
32681 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
32682 (gcc.o): Depend on $(BASEVER).
32683 * common.opt (dumpfullversion): New option.
32684 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
32685 * doc/invoke.texi: Document -dumpfullversion.
32686 * doc/install.texi: Document --with-gcc-major-version-only.
32687 * configure: Regenerated.
32689 2017-01-17 Richard Biener <rguenther@suse.de>
32691 PR tree-optimization/71433
32692 * tree-vrp.c (register_new_assert_for): Merge same asserts
32693 on all incoming edges.
32694 (process_assert_insertions_for): Handle insertions at the
32697 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32699 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32700 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32702 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32705 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32708 2017-01-17 Alan Modra <amodra@gmail.com>
32711 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32712 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32713 symbolic stack limit when pic.
32715 2017-01-16 Martin Sebor <msebor@redhat.com>
32717 PR tree-optimization/78608
32718 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32720 2017-01-16 Jeff Law <law@redhat.com>
32723 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32724 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32725 for several include directories that may be relative to sysroot.
32726 * config/i386/x-mingw32 (gplus_includedir): Define.
32727 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32728 (native_system_includedir): Likewise.
32729 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32730 override if TARGET_SYSTEM_ROOT is defined.
32731 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32733 PR tree-optimization/79090
32734 PR tree-optimization/33562
32735 PR tree-optimization/61912
32736 PR tree-optimization/77485
32737 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32738 and computed trims into the dump file.
32740 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32742 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32744 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32747 * gimplify.c (gimplify_init_constructor): If want_value and
32748 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32752 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32753 sequence. Formatting fixes.
32754 (doloop_optimize): Formatting fixes.
32757 * gcc.c (debug_level_greater_than_spec_func): New function.
32758 (static_spec_functions): Add debug-level-gt spec function.
32759 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32761 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32762 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32763 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32764 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32765 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32766 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32768 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32770 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32771 QImode fixups to general and mask registers only.
32773 2017-01-16 Carl Love <cel@us.ibm.com>
32775 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32776 for built-in functions
32777 vector signed char vec_nabs (vector signed char)
32778 vector signed short vec_nabs (vector signed short)
32779 vector signed int vec_nabs (vector signed int)
32780 vector signed long long vec_nabs (vector signed long long)
32781 vector float vec_nabs (vector float)
32782 vector double vec_nabs (vector double)
32783 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32785 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32786 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32787 * doc/extend.texi: Update the documentation file for the new built-in
32790 2017-01-16 Martin Sebor <msebor@redhat.com>
32792 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32795 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32797 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32798 UNSPEC_VSX__XXSPLTD to require special splat handling.
32800 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32803 * system.h: Poison strndup.
32805 2017-01-16 Alan Modra <amodra@gmail.com>
32808 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32811 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32813 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32815 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32817 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32818 call recog here. Assert that INSN_CODE (insn) is non-negative.
32820 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32823 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32825 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32826 in the currently scheduled RTL fragment.
32828 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32830 PR rtl-optimization/78751
32831 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32834 2017-01-14 Jeff Law <law@redhat.com>
32836 PR tree-optimization/79090
32837 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32838 variable length stores.
32839 (compute_trims): Delete dead assignment to *trim_tail.
32840 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32843 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32845 PR rtl-optimization/78626
32846 PR rtl-optimization/78727
32847 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32848 of a block, and split such blocks after everything else is finished.
32850 2017-01-14 Alan Modra <amodra@gmail.com>
32853 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32854 target legitimate_combined_insn.
32855 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32856 (rs6000_legitimate_combined_insn): New function.
32857 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32859 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32860 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32861 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32863 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32865 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32867 2017-01-13 Jeff Law <law@redhat.com>
32869 PR tree-optimization/33562
32870 PR tree-optimization/61912
32871 PR tree-optimization/77485
32872 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32874 (delete_dead_assignment): Likewise.
32875 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32876 statement to delete_dead_call and delete_dead_assignment.
32878 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32881 * substring-locations.c (format_warning_va): Strengthen case 1 so
32882 that both endpoints of the substring must be within the format
32883 range for just the substring to be printed.
32885 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32887 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32888 * config/i386/i386.c (ix86_target_string): Add missing options
32889 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32890 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32891 flags_other and ix86_target_other to flags2_other. Display unknown
32893 (ix86_valid_target_attribute_inner_p): Add missing options and
32894 reorder options by implied ISAs, as in ix86_target_string.
32896 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32898 * hash-table.h (hash_table::too_empty_p): New function.
32899 (hash_table::expand): Use it.
32900 (hash_table::traverse): Likewise.
32901 (hash_table::empty_slot): Use sizeof (value_type) instead of
32902 sizeof (PTR) to convert bytes to elements. Shrink the table
32903 if the current size is excessive for the current number of
32906 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32908 * ira-costs.c (record_reg_classes): Break from the inner loop
32909 early once alt_fail is known to be true. Update outer loop
32910 handling accordingly.
32912 2017-01-13 Jeff Law <law@redhat.com>
32914 * tree-ssa-dse.c (decrement_count): New function.
32915 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32916 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32917 when we know the partially dead statement is a mem* function.
32919 PR tree-optimization/61912
32920 PR tree-optimization/77485
32921 * tree-ssa-dse.c: Include expr.h.
32922 (maybe_trim_constructor_store): New function.
32923 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32925 PR tree-optimization/33562
32926 PR tree-optimization/61912
32927 PR tree-optimization/77485
32928 * doc/invoke.texi: Document new dse-max-object-size param.
32929 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32930 * tree-ssa-dse.c: Include params.h.
32931 (dse_store_status): New enum.
32932 (initialize_ao_ref_for_dse): New, partially extracted from
32934 (valid_ao_ref_for_dse, normalize_ref): New.
32935 (setup_live_bytes_from_ref, compute_trims): Likewise.
32936 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32937 (maybe_trim_partially_dead_store): Likewise.
32938 (maybe_trim_complex_store): Likewise.
32939 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32940 Track what bytes live from the original store. Return tri-state
32941 for dead, partially dead or live.
32942 (dse_dom_walker): Add constructor, destructor and new private members.
32943 (delete_dead_call, delete_dead_assignment): New extracted from
32945 (dse_optimize_stmt): Make a member of dse_dom_walker.
32946 Use initialize_ao_ref_for_dse.
32948 PR tree-optimization/33562
32949 PR tree-optimization/61912
32950 PR tree-optimization/77485
32951 * sbitmap.h (bitmap_count_bits): Prototype.
32952 (bitmap_clear_range, bitmap_set_range): Likewise.
32953 * sbitmap.c (bitmap_clear_range): New function.
32954 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32956 2017-01-13 Martin Liska <mliska@suse.cz>
32959 * function.c (set_cfun): Add new argument force.
32960 * function.h (set_cfun): Likewise.
32961 * ipa-inline-transform.c (inline_call): Use the function when
32962 strict alising from is dropped for function we inline to.
32964 2017-01-13 Richard Biener <rguenther@suse.de>
32966 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32967 for dumping GIMPLE INTEGER_CSTs.
32969 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32971 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32972 to 201112L since C++17.
32974 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32977 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32978 if -fsanitize=kernel-address is present.
32980 2017-01-13 Richard Biener <rguenther@suse.de>
32982 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32983 as _Literal ( type ) number in case usual suffixes do not
32984 preserve all information.
32986 2017-01-13 Richard Biener <rguenther@suse.de>
32988 PR tree-optimization/77283
32989 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
32990 and ssa-iterators.h.
32991 (is_feasible_trace): Implement a cost model based on joiner
32994 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
32997 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
32998 char or short to __float128/_Float128 directly.
33000 2017-01-12 Martin Sebor <msebor@redhat.com>
33002 to -Wformat-overflow.
33003 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33004 (min_bytes_remaining): Same.
33005 (get_string_length): Same.
33006 (format_string): Same.
33007 (format_directive): Same.
33009 (pass_sprintf_length::handle_gimple_call): Same.
33011 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33013 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33014 info.nowrite calls with no lhs that can't throw. Return bool
33015 whether gsi_remove has been called or not.
33016 (pass_sprintf_length::handle_gimple_call): Return bool whether
33017 try_substitute_return_value called gsi_remove. Formatting fix.
33018 (pass_sprintf_length::execute): Don't use gsi_remove if
33019 handle_gimple_call returned true.
33022 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33023 be removed due to side-effects, don't remove following barrier nor
33024 turn the successor edge into fallthru edge.
33026 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33029 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33030 element-reversing loads and stores as not swappable.
33032 2017-01-12 Nathan Sidwell <nathan@acm.org>
33033 Nicolai Stange <nicstange@gmail.com>
33035 * combine.c (try_combine): Don't ignore result of overlap checking
33036 loop. Combine overlap & asm check into single loop.
33038 2017-01-12 Richard Biener <rguenther@suse.de>
33040 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33041 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33043 2017-01-12 Richard Biener <rguenther@suse.de>
33045 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33046 and TS_TARGET_OPTION directly derive from TS_BASE.
33047 * tree-core.h (tree_optimization_option): Derive from tree_base.
33048 (tree_target_option): Likewise.
33050 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33052 * config/i386/i386.c (memory_address_length): Increase len
33053 only when rip_relative_addr_p returns false.
33055 2017-01-11 Julia Koval <julia.koval@intel.com>
33057 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33058 (OPTION_MASK_ISA_SGX_SET): New.
33059 (ix86_handle_option): Handle OPT_msgx.
33060 * config.gcc: Added sgxintrin.h.
33061 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33062 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33063 * config/i386/i386.c (ix86_target_string): Add -msgx.
33065 (ix86_option_override_internal): Handle new options.
33066 (ix86_valid_target_attribute_inner_p): Add sgx.
33067 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33068 * config/i386/i386.opt: Add msgx.
33069 * config/i386/sgxintrin.h: New file.
33070 * config/i386/x86intrin.h: Add sgxintrin.h.
33072 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33075 * fold-const.c (maybe_nonzero_address): Return 1 for function
33077 (tree_single_nonzero_warnv_p): Don't handle function local objects
33081 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33084 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33087 * opts.c: Include "spellcheck.h"
33088 (struct string_fragment): New struct.
33089 (struct edit_distance_traits<const string_fragment &>): New
33091 (get_closest_sanitizer_option): New function.
33092 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33094 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33096 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33098 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33099 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33100 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33101 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33102 for initial die_offset if dwarf_split_debug_info.
33103 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33104 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33106 (output_skeleton_debug_sections): Formatting fix. Use
33107 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33108 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33110 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33112 * config/arm/cortex-a53.md: Add bypasses for
33113 cortex_a53_r2f_cvt.
33114 (cortex_a53_r2f): Only use for transfers.
33115 (cortex_a53_f2r): Likewise.
33116 (cortex_a53_r2f_cvt): Add reservation for conversions.
33117 (cortex_a53_f2r_cvt): Likewise.
33119 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33121 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33122 to all inlined functions, change static to extern.
33124 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33127 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33129 (arm_assemble_integer): Likewise.
33131 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33133 * config.gcc: Use new awk script to check CPU, FPU and architecture
33134 parameters for --with-... options.
33135 * config/arm/parsecpu.awk: New file
33136 * config/arm/arm-cpus.in: New file.
33137 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33139 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33141 * config/arm/t-arm: Update dependency rules.
33142 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33143 of processing .def files.
33144 * config/arm/genopt.sh: Deleted.
33145 * config/arm/gentune.sh: Deleted.
33146 * config/arm/arm-cores.def: Deleted.
33147 * config/arm/arm-arches.def: Deleted.
33148 * config/arm/arm-fpus.def: Deleted.
33149 * config/arm/arm-tune.md: Regenerated.
33150 * config/arm/arm-tables.opt: Regenerated.
33151 * config/arm/arm-cpu.h: New generated file.
33152 * config/arm/arm-cpu-data.h: New generated file.
33153 * config/arm/arm-cpu-cdata.h: New generated file.
33155 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33158 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33160 (input_varpool_node): Unpack dynamically_initialized bit.
33162 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33164 PR rtl-optimization/79032
33165 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33166 the alignment of the adjusted memory reference against that of MODE,
33167 instead of the alignment of the original memory reference.
33169 2017-01-11 Martin Jambor <mjambor@suse.cz>
33171 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
33173 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
33174 decorated functions.
33176 2017-01-11 Richard Biener <rguenther@suse.de>
33178 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
33179 set range/nonnull info for PHI results. Do not set it on
33180 stmts marked for removal.
33182 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
33184 * expr.c (store_field): In the bitfield case, fetch the return value
33185 from the registers before applying a single big-endian adjustment.
33186 Always do a final load for a BLKmode value not larger than a word.
33188 2017-01-10 David Malcolm <dmalcolm@redhat.com>
33191 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33192 that we correctly handle column numbers greater than
33193 LINE_MAP_MAX_COLUMN_NUMBER.
33195 2017-01-10 Martin Sebor <msebor@redhat.com>
33197 PR middle-end/78245
33198 * gimple-ssa-sprintf.c (get_destination_size): Call
33199 {init,fini}object_sizes.
33200 * tree-object-size.c (addr_object_size): Adjust.
33201 (pass_through_call): Adjust.
33202 (pass_object_sizes::execute): Adjust.
33203 * tree-object-size.h (fini_object_sizes): Declare.
33205 2017-01-10 Martin Sebor <msebor@redhat.com>
33207 PR tree-optimization/78775
33208 * builtins.c (get_size_range): Move...
33209 * calls.c: ...to here.
33210 (alloc_max_size): Accept zero argument.
33211 (operand_signed_p): Remove.
33212 (maybe_warn_alloc_args_overflow): Call get_size_range.
33213 * calls.h (get_size_range): Declare.
33215 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
33217 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
33218 from TI's devices.csv file as of September 2016.
33219 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
33221 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
33223 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
33224 * doc/invoke.texi: Likewise.
33225 * doc/md.texi: Likewise.
33226 * doc/objc.texi: Likewise.
33228 2017-01-10 Joshua Conner <joshconner@google.com>
33230 * config/arm/fuchsia-elf.h: New file.
33231 * config/fuchsia.h: New file.
33232 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
33233 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
33235 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33237 2016-01-10 Richard Biener <rguenther@suse.de>
33239 PR tree-optimization/79034
33240 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33241 Propagate out degenerate PHIs in the joiner.
33243 2017-01-10 Martin Liska <mliska@suse.cz>
33245 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33246 (sort_congruence_classes_by_decl_uid): Likewise.
33247 (sort_congruence_class_groups_by_decl_uid): Likewise.
33248 (sem_item_optimizer::merge_classes): Sort class, groups in these
33249 classes and members in the groups by DECL_UID of declarations.
33250 This would make merge operations stable.
33252 2017-01-10 Martin Liska <mliska@suse.cz>
33254 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33255 usage of m_classes_vec.
33256 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33257 (sem_item_optimizer::get_group_by_hash): Likewise.
33258 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33259 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33260 (sem_item_optimizer::verify_classes): Likewise.
33261 (sem_item_optimizer::process_cong_reduction): Likewise.
33262 (sem_item_optimizer::dump_cong_classes): Likewise.
33263 (sem_item_optimizer::merge_classes): Likewise.
33264 * ipa-icf.h (congruence_class_hash): Rename from
33265 congruence_class_group_hash. Remove declaration of m_classes_vec.
33267 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33269 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33270 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33271 * config.gcc: Add avx512vpopcntdqintrin.h.
33272 * config/i386/avx512vpopcntdqintrin.h: New.
33273 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33274 * config/i386/i386-builtin-types.def: Add new types.
33275 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33276 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33277 __builtin_ia32_vpopcountq_v8di_mask): New.
33278 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33279 __AVX512VPOPCNTDQ__.
33280 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33281 (PTA_AVX512VPOPCNTDQ): Define.
33282 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33283 TARGET_AVX512VPOPCNTDQ_P): Define.
33284 * config/i386/i386.opt: Add mavx512vpopcntdq.
33285 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33286 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33288 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33290 PR middle-end/77484
33291 * predict.def (PRED_CALL): Set to 67.
33293 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33295 * expr.c (store_field): In the bitfield case, if the value comes from
33296 a function call and is of an aggregate type returned in registers, do
33297 not modify the field mode; extract the value in all cases if the mode
33298 is BLKmode and the size is not larger than a word.
33300 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33303 * config/i386/cpuid.h: Fix undefined behavior.
33305 2017-01-04 Jeff Law <law@redhat.com>
33307 PR tree-optimization/79007
33308 PR tree-optimization/67955
33309 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33310 conservative for pt.null when flag_non_call_exceptions is on.
33312 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33314 PR translation/79019
33315 PR translation/79020
33316 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33317 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33318 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33320 * config/avr/avr.opt (maccumulate-args): Likewise.
33321 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33322 * common.opt (freport-bug): Likewise.
33323 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33324 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33325 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33326 translatable string.
33327 * config/i386/i386.c (function_value_32): Likewise.
33328 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33329 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33331 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33332 * common/config/msp430/msp430-common.c (msp430_handle_option):
33334 * symtab.c (symtab_node::verify_base): Likewise.
33335 * opts.c (set_debug_level): Likewise.
33336 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33337 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33338 missing whitespace to translatable strings.
33339 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33340 * config/sh/superh.h: Likewise.
33341 * config/i386/xopintrin.h: Likewise.
33342 * config/i386/znver1.md: Likewise.
33343 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33344 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33345 * double-int.h (struct double_int): Likewise.
33346 * double-int.c (div_and_round_double): Likewise.
33347 * wide-int.cc: Likewise.
33348 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33349 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33350 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33351 (crossjumps_occurred): ... this.
33352 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33355 PR tree-optimization/78899
33356 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33357 returning bool return struct loop *, NULL for failure and the new
33359 (versionable_outer_loop_p): Don't version outer loop if it has
33360 dont_vectorized bit set.
33361 (tree_if_conversion): When versioning outer loop, ensure
33362 tree_if_conversion is performed also on the inner loop of the
33363 non-vectorizable outer loop copy.
33364 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33365 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33366 prevent vectorization of it.
33367 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33368 the outer loop vectorization of the non-scalar version is attempted
33369 before vectorization of the inner loop in scalar version. If
33370 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33371 vectorization of its inner loop.
33372 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33373 has 2 inner loops, rename also on edges from bb whose single pred
33374 is outer_loop->header. Fix typo in function comment.
33376 2017-01-09 Martin Sebor <msebor@redhat.com>
33379 * asan.c (asan_emit_stack_protection): Increase local buffer size
33380 to avoid snprintf truncation warning.
33382 2017-01-09 Andrew Pinski <apinski@cavium.com>
33384 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33385 to reference thunderx2t99 for the tuning structure
33386 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33388 (thunderx2t99_extra_costs): This.
33389 * config/aarch64/aarch64-tune.md: Regenerate.
33390 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33391 (vulcan_addrcost_table): This.
33392 (vulcan_regmove_cost): Rename to ...
33393 (thunderx2t99_regmove_cost): This.
33394 (vulcan_vector_cost): Rename to ...
33395 (thunderx2t99_vector_cost): this.
33396 (vulcan_branch_cost): Rename to ...
33397 (thunderx2t99_branch_cost): This.
33398 (vulcan_tunings): Rename to ...
33399 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33400 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33402 2017-01-09 Martin Jambor <mjambor@suse.cz>
33406 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33407 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33408 (propagate_vr_accross_jump_function): Use the above function for all
33409 value range computations for pass-through jump functions and type
33410 converasion from explicit value range values.
33411 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33412 parameters from TYPE_ARG_TYPES.
33413 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33414 (ipa_write_node_info): Stream type of the actual argument.
33415 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33417 2017-01-09 Martin Liska <mliska@suse.cz>
33420 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33421 (lookup_compiler): Do not show error message with have_E.
33423 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33425 PR tree-optimization/78938
33426 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33427 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33428 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33429 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33432 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33434 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33437 2017-01-09 Richard Biener <rguenther@suse.de>
33439 PR tree-optimization/78997
33440 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33441 name condition properly.
33443 2017-01-09 Richard Biener <rguenther@suse.de>
33446 * dwarf2out.c (is_cxx): New overload with context.
33447 (is_naming_typedef_decl): Use it.
33449 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33451 * invoke.texi (Option Summary): Correct spacing in option lists
33452 and add line breaks to fix over-long lines.
33454 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33456 PR middle-end/17660
33458 * extend.texi (Common Variable Attributes): Add xref to GCC
33459 Internals manual to explain mode attribute keywords.
33461 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33464 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33465 and Preprocessor Options.
33466 (Options for Linking): Document -pthread here....
33467 (RS/6000 and PowerPC Options): ...not here.
33468 (Solaris 2 Options): ...or here.
33469 * doc/cppopts.texi: Document -pthread.
33471 2017-01-08 Martin Sebor <msebor@redhat.com>
33473 PR middle-end/77708
33474 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33475 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33476 New member functions.
33477 (format_directive): Used them.
33479 (pass_sprintf_length::handle_gimple_call): Same.
33480 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33481 to avoid truncation for any argument.
33482 (extract_affine_mul): Same.
33483 * tree.c (get_file_function_name): Same.
33485 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33487 PR middle-end/77484
33488 * predict.def (PRED_INDIR_CALL): Set to 86.
33490 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33492 PR preprocessor/54124
33493 * doc/cppopts.texi: Reformat -d subtable to list the full name
33494 of the options. Add cross-reference to the docs for the general
33495 compiler -d options.
33496 * doc/invoke.texi (Developer Options): Add cross-reference to the
33497 preprocessor-specific -d option documentation.
33499 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33501 PR preprocessor/13498
33502 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33503 redudant material, and reflect new command-line options.
33504 (System Headers): Likewise.
33506 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33508 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33509 -isystem, and -idirafter. Copy-edit.
33510 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33511 default for -ftrack-macro-expansion. Delete obsolete and
33512 badly-formatted implementation details about -fdebug-cpp output.
33513 * doc/cppwarnopts.texi: Copy-edit.
33515 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33518 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33519 that the transition from a max line width >= 1<<10 to narrower
33520 lines works correctly.
33522 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33524 * doc/options.texi (PerFunction): New.
33525 * opt-functions.awk (switch_flags): Map both Optimization and
33526 PerFunction to CL_OPTIMIZATION.
33527 * opth-gen.awk: Test for PerFunction flag along with
33529 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33530 it only when the latter is present. Skip those that don't in
33531 the hash function generator.
33532 * common.opt (fvar-tracking): Mark as PerFunction instead of
33534 (fvar-tracking-assignments): Likewise.
33535 (fvar-tracking-assignments-toggle): Likewise.
33536 (fvar-tracking-uninit): Likewise.
33538 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33540 PR translation/79018
33541 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33544 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33547 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33548 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33549 TARGET_LONG_JUMP_TABLE_OFFSETS.
33550 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33551 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33552 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33553 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33554 * config/m68k/m68k.md (tablejump expander): Likewise.
33555 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33556 TARGET_LONG_JUMP_TABLE_OFFSETS.
33557 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33558 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33560 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33561 David Holsgrove <david.holsgrove@xilinx.com>
33563 * common/config/microblaze/microblaze-common.c
33564 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33565 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33567 * config/microblaze/microblaze.c (microblaze_must_save_register)
33568 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33570 (microblaze_eh_return): New function.
33571 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33572 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33573 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33574 * config/microblaze/microblaze.md (eh_return): New pattern.
33576 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33578 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33579 GCC_DIAGNOSTIC_STRINGIFY): Define.
33581 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33583 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33585 * config/arm/arm.md (<mcrr>): New.
33587 * config/arm/arm.c (arm_arch5te): New.
33588 (arm_option_override): Set arm_arch5te.
33589 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33591 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33592 (arm_mcrr_qualifiers): ... this. New.
33593 (MRRC_QUALIFIERS): Define to...
33594 (arm_mrrc_qualifiers): ... this. New.
33595 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33597 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33598 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33599 (MRRCI, mrrc, MRRC): New.
33600 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33601 VUNSPEC_MRRC2): New.
33603 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33605 * config/arm/arm.md (<mcr>): New.
33607 * config/arm/arm.c (arm_coproc_builtin_available): Add
33608 support for mcr, mrc, mcr2 and mrc2.
33609 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33610 (arm_mcr_qualifiers): ... this. New.
33611 (MRC_QUALIFIERS): Define to ...
33612 (arm_mrc_qualifiers): ... this. New.
33613 (MCR_QUALIFIERS): Define to ...
33614 (arm_mcr_qualifiers): ... this. New.
33615 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33617 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33618 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33619 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33620 VUNSPEC_MRC2): New.
33622 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33624 * config/arm/arm.md (*ldc): New.
33628 * config/arm/arm.c (arm_coproc_builtin_available): Add
33629 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33630 (arm_coproc_ldc_stc_legitimate_address): New.
33631 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33632 'qualifier_const_pointer'.
33633 (LDC_QUALIFIERS): Define to...
33634 (arm_ldc_qualifiers): ... this. New.
33635 (STC_QUALIFIERS): Define to...
33636 (arm_stc_qualifiers): ... this. New.
33637 * config/arm/arm-protos.h
33638 (arm_coproc_ldc_stc_legitimate_address): New.
33639 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33640 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33641 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33642 stc2, stcl, stc2l): New.
33643 * config/arm/constraints.md (Uz): New.
33644 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33645 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33646 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33647 VUNSPEC_STC2L): New.
33649 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33651 * config/arm/arm.md (<cdp>): New.
33652 * config/arm/arm.c (neon_const_bounds): Rename this ...
33653 (arm_const_bounds): ... this.
33654 (arm_coproc_builtin_available): New.
33655 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33656 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33657 (CDP_QUALIFIERS): Define to...
33658 (arm_cdp_qualifiers): ... this. New.
33660 (arm_expand_builtin_args): Add case for 6 arguments.
33661 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33662 (arm_const_bounds): ... this.
33663 (arm_coproc_builtin_available): New.
33664 * config/arm/arm_acle.h (__arm_cdp): New.
33666 * config/arm/arm_acle_builtins.def (cdp): New.
33668 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33669 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33670 'arm_const_bounds'.
33671 * config/arm/types.md (coproc): New.
33672 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
33673 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
33674 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
33675 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
33677 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33679 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
33680 (UBINOP_QUALIFIERS): New.
33682 (acle_builtin_data): New. Change comment.
33683 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
33684 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
33685 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
33686 arm_acle_builtins.def.
33687 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
33688 (arm_init_acle_builtins): New.
33689 (CRC32_BUILTIN): Remove.
33690 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33691 crc32cb, crc32ch and crc32cw.
33692 (arm_init_crc32_builtins): Remove.
33693 (arm_init_builtins): Use arm_init_acle_builtins rather
33694 than arm_init_crc32_builtins.
33695 (arm_expand_acle_builtin): New.
33696 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33697 * config/arm/arm_acle_builtins.def: New.
33699 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33701 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33702 (arm_builtin_datum): ... this.
33703 (arm_init_neon_builtin): Rename to ...
33704 (arm_init_builtin): ... this. Add a new parameters PREFIX
33705 and USE_SIG_IN_NAME.
33706 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33707 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33708 'arm_builtin_datum'.
33709 (arm_init_vfp_builtins): Likewise.
33710 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33711 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33712 (arm_expand_neon_args): Rename to ...
33713 (arm_expand_builtin_args): ... this. Rename builtin_arg
33714 enum values and differentiate between ARG_BUILTIN_MEMORY
33715 and ARG_BUILTIN_NEON_MEMORY.
33716 (arm_expand_neon_builtin_1): Rename to ...
33717 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33718 values, arm_expand_builtin_args and add bool parameter NEON.
33719 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33720 (arm_expand_vfp_builtin): Likewise.
33721 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33723 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33725 PR middle-end/77484
33726 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33727 * predict.c (tree_estimate_probability_bb): Reverse direction of
33728 polymorphic call predictor.
33730 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33732 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33733 (determine_pass_name_match): ...this new function and...
33734 (should_skip_pass_p): ...this new function.
33736 2017-01-06 Nathan Sidwell <nathan@acm.org>
33738 * ipa-visibility.c (function_and_variable_visibility): Reformat
33739 comments and long lines. Remove extrneous if.
33740 * symtab.c (symtab_node::make_decl_local): Fix code format.
33741 (symtab_node::set_section_for_node): Fix comment typo.
33743 2017-01-06 Martin Liska <mliska@suse.cz>
33746 * lra-constraints.c: Rename invariant to lra_invariant.
33747 * predict.c (set_even_probabilities): Initialize e to NULL.
33749 2017-01-05 Martin Sebor <msebor@redhat.com>
33751 PR tree-optimization/78910
33752 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33753 (format_integer): Correct off-by-one error in the handling
33754 of precision with negative numbers in signed conversions..
33756 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33758 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33760 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33762 PR tree-optimization/71016
33763 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33764 factor_out_conditional_conversion. Formatting fix.
33765 (factor_out_conditional_conversion): Add cond_stmt argument.
33766 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33767 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33770 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33772 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33773 read-rtl-function.o, and selftest-rtl.o.
33774 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33775 (selftest::aarch64_test_loading_full_dump): New function.
33776 (selftest::aarch64_run_selftests): New function.
33777 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33778 selftest::aarch64_run_selftests.
33779 * config/i386/i386.c
33780 (selftest::ix86_test_loading_dump_fragment_1): New function.
33781 (selftest::ix86_test_loading_call_insn): New function.
33782 (selftest::ix86_test_loading_full_dump): New function.
33783 (selftest::ix86_test_loading_unspec): New function.
33784 (selftest::ix86_run_selftests): Call the new functions.
33785 * emit-rtl.c (maybe_set_max_label_num): New function.
33786 * emit-rtl.h (maybe_set_max_label_num): New decl.
33787 * function.c (instantiate_decls): Guard call to
33788 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33789 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33791 * gensupport.c (gen_reader::gen_reader): Pass "false"
33792 for new "compact" param of rtx_reader.
33793 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33794 rather than an empty string for NULL strings.
33795 * read-md.c: Potentially include config.h rather than bconfig.h.
33796 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33797 (have_error): New global, copied from errors.c.
33798 (md_reader::read_name): Rename to...
33799 (md_reader::read_name_1): ...this, adding "out_loc" param,
33800 and converting "missing name or number" to returning false, rather
33802 (md_reader::read_name): Reimplement in terms of read_name_1.
33803 (md_reader::read_name_or_nil): New function.
33804 (md_reader::read_string): Handle "(nil)" by returning NULL.
33805 (md_reader::md_reader): Add new param "compact".
33806 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33807 (md_reader::read_file): New method.
33808 * read-md.h (md_reader::md_reader): Add new param "compact".
33809 (md_reader::read_file): New method.
33810 (md_reader::is_compact): New accessor.
33811 (md_reader::read_name): Convert return type from void to file_location.
33812 (md_reader::read_name_or_nil): New decl.
33813 (md_reader::read_name_1): New decl.
33814 (md_reader::m_compact): New field.
33815 (noop_reader::noop_reader): Pass "false" for new "compact" param
33817 (rtx_reader::rtx_reader): Add new "compact" param.
33818 (rtx_reader::read_rtx_operand): Make virtual and convert return
33819 type from void to rtx.
33820 (rtx_reader::read_until): New decl.
33821 (rtx_reader::handle_any_trailing_information): New virtual function.
33822 (rtx_reader::postprocess): New virtual function.
33823 (rtx_reader::finalize_string): New virtual function.
33824 (rtx_reader::m_in_call_function_usage): New field.
33825 (rtx_reader::m_reuse_rtx_by_id): New field.
33826 * read-rtl-function.c: New file.
33827 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33828 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33829 (selftest::verify_three_block_rtl_cfg): New decl.
33830 * read-rtl-function.h: New file.
33831 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33832 For host, include function.h, memmodel.h, and emit-rtl.h.
33833 (one_time_initialization): New function.
33834 (struct compact_insn_name): New struct.
33835 (compact_insn_names): New array.
33836 (find_code): Handle insn codes in compact dumps.
33837 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33838 (bind_subst_iter_and_attr): Likewise.
33839 (add_condition_to_string): Likewise.
33840 (add_condition_to_rtx): Likewise.
33841 (apply_attribute_uses): Likewise.
33842 (add_current_iterators): Likewise.
33843 (apply_iterators): Likewise.
33844 (initialize_iterators): Guard usage of apply_subst_iterator with
33845 #ifdef GENERATOR_FILE.
33846 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33847 (md_reader::read_mapping): Likewise.
33848 (add_define_attr_for_define_subst): Likewise.
33849 (add_define_subst_attr): Likewise.
33850 (read_subst_mapping): Likewise.
33851 (check_code_iterator): Likewise.
33852 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33854 (one_time_initialization): New function.
33855 (rtx_reader::read_until): New method.
33856 (read_flags): New function.
33857 (parse_reg_note_name): New function.
33858 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33859 Handle reuse_rtx ids.
33860 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33861 Add parsing support for RTL dumps, mirroring the special-cases in
33862 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33863 values, and calling handle_any_trailing_information.
33864 (rtx_reader::read_rtx_operand): Convert return type from void
33865 to rtx, returning return_rtx. Handle case 'e'. Call
33866 finalize_string on XSTR and XTMPL fields.
33867 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33868 "(nil)" values were omitted. Call the postprocess vfunc on the
33870 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33871 class ctor. Initialize m_in_call_function_usage. Call
33872 one_time_initialization.
33873 * rtl-tests.c (selftest::test_uncond_jump): Call
33874 set_new_first_and_last_insn.
33875 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33876 * selftest-rtl.c: New file.
33877 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33878 (selftest::get_insn_by_uid): New decl.
33879 * selftest-run-tests.c (selftest::run_tests): Call
33880 read_rtl_function_c_tests.
33881 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33882 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33885 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33887 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33888 operands in a special way. Assert that pos+len <= mode precision.
33890 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33892 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33893 3 argument Alias with unlimited for the negative form.
33894 (fno-vect-cost-model): Removed.
33896 2017-01-05 Martin Liska <mliska@suse.cz>
33898 * hsa-gen.c (gen_hsa_divmod): New function.
33899 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33901 2017-01-05 Martin Liska <mliska@suse.cz>
33904 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33907 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33909 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33910 small constant length operands.
33912 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33914 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33915 between loop iterations.
33917 2017-01-05 Martin Liska <mliska@suse.cz>
33920 * gimplify.c (gimplify_decl_expr): Compare to
33921 asan_poisoned_variables instread of checking flags.
33922 (gimplify_target_expr): Likewise.
33923 (gimplify_expr): Likewise.
33924 (gimplify_function_tree): Conditionally initialize
33925 asan_poisoned_variables.
33927 2017-01-04 Jeff Law <law@redhat.com>
33929 PR tree-optimizatin/78812
33930 * rtl.h (contains_mem_rtx_p): Prototype.
33931 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33932 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33933 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33934 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33935 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33937 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33939 * input.c (assert_char_at_range): Default-initialize actual_range.
33941 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33943 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33944 to match the caller.
33946 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33948 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33949 insns after final jump in test to emit dummy move.
33951 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33953 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33954 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33956 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33958 * multiple_target.c (create_dispatcher_calls): Init e_next.
33959 * tree-ssa-loop-split.c (split_loop): Init border.
33960 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33963 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33968 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33969 (altivec_register_operand): Do not return true if the operand
33970 contains a SUBREG mixing SImode and SFmode.
33971 (vsx_register_operand): Likewise.
33972 (vsx_reg_sfsubreg_ok): New predicate.
33973 (vfloat_operand): Do not return true if the operand contains a
33974 SUBREG mixing SImode and SFmode.
33975 (vint_operand): Likewise.
33976 (vlogical_operand): Likewise.
33977 (gpc_reg_operand): Likewise.
33978 (int_reg_operand): Likewise.
33979 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33980 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33981 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33983 (rs6000_emit_move_si_sf_subreg): New helper function.
33984 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
33985 fixup SUBREGs involving SImode and SFmode.
33986 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
33987 numbers for the new peephole2 optimization.
33988 (peephole2 for SFmode unions): New peephole2 to optimize cases in
33989 the GLIBC math library that do AND/IOR/XOR operations on single
33990 precision floating point.
33991 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
33992 target macros to say whether we need to avoid SUBREGs mixing
33994 (TARGET_ALLOW_SF_SUBREG): Likewise.
33995 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
33996 (UNSPEC_SI_FROM_SF): Likewise.
33997 (iorxor): Change spacing.
33998 (and_ior_xor): New iterator for AND, IOR, and XOR.
33999 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34000 (movdi_from_sf_zero_ext): Likewise.
34001 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34002 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34003 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34004 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34005 (fms<mode>4): Likewise.
34006 (fnma<mode>4): Likewise.
34007 (fnms<mode>4): Likewise.
34008 (nfma<mode>4): Likewise.
34009 (nfms<mode>4): Likewise.
34011 2017-01-04 Marek Polacek <polacek@redhat.com>
34014 * doc/invoke.texi: Document -Wpointer-compare.
34016 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34018 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34021 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34022 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34025 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34028 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34029 documentation of the powerpc_popcntb_ok attribute.
34030 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34031 code to issue warning messages if a requested CPU configuration is
34032 not supported by the binary (assembler and loader) toolchain.
34033 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34034 made to define a built-in function that has been disabled.
34035 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34036 made to define a built-in function that has been disabled.
34037 (altivec_init_builtins): Add comment explaining why definition
34038 of the DST built-in functions is not preceded by an assertion
34039 check. Add assertions to prevent ICE if attempts are made to
34040 define an altivec predicate or an abs* built-in function that has
34042 (htm_init_builtins): Add comment explaining why definition of the
34043 htm built-in functions is not preceded by an assertion check.
34045 2017-01-04 Jeff Law <law@redhat.com>
34047 PR tree-optimizatin/67955
34048 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34049 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34050 the points-to solution does not include pt_null. Use DECL_PT_UID
34053 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34055 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34056 Use gen_int_mode instead of gen_lopwart for const_int operands.
34058 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34060 PR tree-optimization/71563
34061 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34062 out of range value - has low bits known to be zero.
34064 2017-01-04 Alan Modra <amodra@gmail.com>
34066 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34067 * configure: Regenerate.
34068 * config.in: Regenerate.
34070 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34073 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34074 a substring of the message, but strcmp with the whole message. Ifdef
34075 ENABLE_NLS, translate the message first using dgettext.
34077 2017-01-03 Jeff Law <law@redhat.com>
34079 PR tree-optimizatin/78856
34080 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34081 (mark_threaded_blocks): Remove code to truncate thread paths that
34082 cross multiple loop headers. Instead invalidate the cached loop
34083 iteration information and handle case of a thread path walking
34084 into an irreducible region.
34086 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34089 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34090 assertions. Add support for doing the signbit if the IEEE 128-bit
34091 floating point value is in a GPR.
34092 * config/rs6000/rs6000.md (Fsignbit): Delete.
34093 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34094 Update the length attribute if the value is in a GPR.
34095 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34096 the sign or zero extension instruction, since the value is always 0/1.
34097 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34100 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34101 extracting SImode to a GPR register so that we can generate a
34102 store, limit the vector to be in a traditional Altivec register
34103 for the vextuwrx instruction.
34105 2017-01-03 Ian Lance Taylor <iant@google.com>
34107 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34109 2017-01-03 Martin Sebor <msebor@redhat.com>
34111 PR tree-optimization/78696
34112 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34113 precision. Use MPFR for %f for greater fidelity. Correct handling
34115 (pass_sprintf_length::compute_format_length): Set width and precision
34116 specified by asrerisk to void_node for vararg functions.
34117 (try_substitute_return_value): Adjust dump output.
34119 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34121 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34123 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34125 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34126 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34127 -mlra/-mno-lra was passed to the compiler.
34129 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34131 PR rtl-optimization/65618
34132 * emit-rtl.c (try_split): Move initialization of "before" and
34133 "after" to just before the call to emit_insn_after_setloc.
34135 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34137 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34139 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34141 * dwarf2out.c (gen_enumeration_type_die): When
34142 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34144 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34146 PR tree-optimization/78965
34147 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34148 Change first argument from const call_info & to call_info &. For %n
34149 set info.nowrite to false.
34151 PR middle-end/78901
34152 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34153 possibly throwing calls.
34155 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34156 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34157 and fns handling, rather than in a separate case SSA_NAME.
34159 2017-01-02 Jeff Law <law@redhat.com>
34161 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34162 fixes for first_period and second_period variables.
34164 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34167 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34168 (*insvqi_1): New insn pattern.
34169 (*insvqi_1_mem_rex64): Ditto.
34170 (*insvqi_2): Ditto.
34171 (*insvqi_3): Rename from *insvqi.
34173 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
34175 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
34177 * doc/cfg.texi (Edges): Remove reference to Java.
34178 (Maintaining the CFG): Ditto.
34180 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34182 PR middle-end/77674
34183 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
34184 transparent aliases.
34186 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34188 PR middle-end/77484
34189 * predict.def (PRED_CALL): Update hitrate.
34190 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
34191 * predict.c (tree_estimate_probability_bb): Split CALL predictor
34192 into direct/indirect/polymorphic variants.
34194 2017-01-01 Jakub Jelinek <jakub@redhat.com>
34196 Update copyright years.
34198 * gcc.c (process_command): Update copyright notice dates.
34199 * gcov-dump.c (print_version): Ditto.
34200 * gcov.c (print_version): Ditto.
34201 * gcov-tool.c (print_version): Ditto.
34202 * gengtype.c (create_file): Ditto.
34203 * doc/cpp.texi: Bump @copying's copyright year.
34204 * doc/cppinternals.texi: Ditto.
34205 * doc/gcc.texi: Ditto.
34206 * doc/gccint.texi: Ditto.
34207 * doc/gcov.texi: Ditto.
34208 * doc/install.texi: Ditto.
34209 * doc/invoke.texi: Ditto.
34211 Copyright (C) 2017 Free Software Foundation, Inc.
34213 Copying and distribution of this file, with or without modification,
34214 are permitted in any medium without royalty provided the copyright
34215 notice and this notice are preserved.