1 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
3 * config/arm/arm.c (output_return_instruction): Add comments to
4 indicate requirement for cmse_nonsecure_entry return to account
5 for the size of clearing instruction output here.
6 (thumb_exit): Likewise.
7 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
8 return in hardfloat mode.
10 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
12 * config/rs6000/rs6000.c (machine_function): Add a bool,
13 "toc_is_wrapped_separately".
14 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
15 if it wasn't explicitly set or unset, we are optimizing for speed, and
16 doing separate shrink-wrapping.
17 (rs6000_get_separate_components): Enable the TOC component if
18 saving the TOC register in the prologue.
19 (rs6000_components_for_bb): Handle the TOC component.
20 (rs6000_emit_prologue_components): Store the TOC register where needed.
21 (rs6000_set_handled_components): Mark TOC as handled, if handled.
22 (rs6000_emit_prologue): Don't save the TOC if that is already done.
24 2017-11-09 Martin Jambor <mjambor@suse.cz>
26 * ipa-param-manipulation.c: New file.
27 * ipa-param-manipulation.h: Likewise.
28 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
29 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
30 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
31 (ipa_parm_adjustment): Likewise.
32 (ipa_parm_adjustment_vec): Likewise.
33 (ipa_get_vector_of_formal_parms): Moved declaration to
34 ipa-param-manipulation.h.
35 (ipa_get_vector_of_formal_parm_types): Likewise.
36 (ipa_modify_formal_parameters): Likewise.
37 (ipa_modify_call_arguments): Likewise.
38 (ipa_combine_adjustments): Likewise.
39 (ipa_dump_param_adjustments): Likewise.
40 (ipa_modify_expr): Likewise.
41 (ipa_get_adjustment_candidate): Likewise.
42 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
43 ipa-param-manipulation.c.
44 (ipa_get_vector_of_formal_parm_types): Likewise.
45 (ipa_modify_formal_parameters): Likewise.
46 (ipa_modify_call_arguments): Likewise.
47 (ipa_modify_expr): Likewise.
48 (get_ssa_base_param): Likewise.
49 (ipa_get_adjustment_candidate): Likewise.
50 (index_in_adjustments_multiple_times_p): Likewise.
51 (ipa_combine_adjustments): Likewise.
52 (ipa_dump_param_adjustments): Likewise.
53 * tree-sra.c: Also include ipa-param-manipulation.h
54 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
57 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
58 Alan Hayward <alan.hayward@arm.com>
59 David Sherwood <david.sherwood@arm.com>
61 * doc/sourcebuild.texi (vect_masked_store): Document.
63 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
64 Alan Hayward <alan.hayward@arm.com>
65 David Sherwood <david.sherwood@arm.com>
67 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
69 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
70 Alan Hayward <alan.hayward@arm.com>
71 David Sherwood <david.sherwood@arm.com>
73 * doc/sourcebuild.texi (vect_variable_length): Document.
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_unaligned_possible): 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_element_align_preferred): 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_perm_short, vect_perm_byte): Document
92 previously undocumented selectors.
93 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
95 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
97 * doc/rtl.texi (const_vector): Say that elements can be
99 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
100 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
101 (gen_vec_duplicate): Use it instead of CONSTANT_P.
102 * optabs.c (expand_vector_broadcast): Likewise.
104 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
105 Alan Hayward <alan.hayward@arm.com>
106 David Sherwood <david.sherwood@arm.com>
108 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
109 scaled index even if the unscaled address was invalid.
110 Don't increase the complexity of using a scale in that case.
112 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
113 Alan Hayward <alan.hayward@arm.com>
114 David Sherwood <david.sherwood@arm.com>
116 * doc/rtl.texi: Rewrite the subreg rules so that they partition
117 the inner register into REGMODE_NATURAL_SIZE bytes rather than
118 UNITS_PER_WORD bytes.
119 * emit-rtl.c (validate_subreg): Divide subregs into blocks
120 based on REGMODE_NATURAL_SIZE of the inner mode.
121 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
122 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
123 * expmed.c (lowpart_bit_field_p): Divide the value up into
124 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
125 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
126 whether something is likely to occupy more than one register.
128 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
131 * ipa-inline-transform.c (update_noncloned_frequencies): Use
132 profile_count::adjust_for_ipa_scaling.
133 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
134 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
136 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
138 2017-11-09 Jakub Jelinek <jakub@redhat.com>
140 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
141 (split_group): Add total_orig and total_new arguments, estimate the
142 number of statements related to the store group without store merging
143 and with store merging.
144 (imm_store_chain_info::output_merged_store): Adjust split_group
145 callers, punt if estimated number of statements with store merging
146 is not smaller than estimated number of statements without it.
148 (handled_load): Remove has_single_use checks.
149 (pass_store_merging::process_store): Likewise.
151 2017-11-09 Richard Biener <rguenther@suse.de>
153 PR tree-optimization/82902
154 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
156 2017-11-09 Martin Liska <mliska@suse.cz>
159 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
162 2017-11-09 Martin Liska <mliska@suse.cz>
164 PR tree-optimization/82669
165 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
167 2017-11-09 Martin Liska <mliska@suse.cz>
169 PR gcov-profile/48463
170 * coverage.c (coverage_begin_function): Output also end locus
171 of a function and information whether the function is
173 * gcov-dump.c (tag_function): Parse and print the information.
174 * gcov.c (INCLUDE_MAP): Add include.
175 (INCLUDE_SET): Likewise.
176 (struct line_info): Move earlier in the source file because
177 of vector<line_info> in function_info structure.
178 (line_info::line_info): Likewise.
179 (line_info::has_block): Likewise.
180 (struct source_info): Add new member index.
181 (source_info::get_functions_at_location): New function.
182 (function_info::group_line_p): New function.
183 (output_intermediate_line): New function.
184 (output_intermediate_file): Use the mentioned function.
185 (struct function_start): New.
186 (struct function_start_pair_hash): Likewise.
187 (process_file): Add code that identifies group functions.
188 Assign lines either to global or function scope.
189 (generate_results): Skip artificial functions.
190 (find_source): Assign index for each source file.
191 (read_graph_file): Read new flag artificial and end_line.
192 (add_line_counts): Assign it either to global of function scope.
193 (accumulate_line_counts): Isolate core of the function to
194 accumulate_line_info and call it for both function and global
196 (accumulate_line_info): New function.
197 (output_line_beginning): Fix GNU coding style.
198 (print_source_line): New function.
199 (output_line_details): Likewise.
200 (output_function_details): Likewise.
201 (output_lines): Iterate both source (global) scope and function
203 (struct function_line_start_cmp): New class.
204 * doc/gcov.texi: Reflect changes in documentation.
206 2017-11-09 Jakub Jelinek <jakub@redhat.com>
209 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
210 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
211 and similarly for not instead of neg.
213 2017-11-08 Andi Kleen <ak@linux.intel.com>
215 * config/i386/i386.opt: Add -mforce-indirect-call.
216 * config/i386/predicates.md: Check for flag_force_indirect_call.
217 * doc/invoke.texi: Document -mforce-indirect-call
219 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
221 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
223 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
224 config/riscv/riscv.c (predict.h): New include.
225 (riscv_slow_unaligned_access_p): No longer static.
226 (riscv_block_move_straight): Add require.
227 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
229 2017-11-08 Jakub Jelinek <jakub@redhat.com>
232 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
233 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
234 nonimmediate_operand predicate for operand 1 instead of
237 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
239 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
241 * config/aarch64/constraints.md (Uml): New constraint.
242 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
245 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
247 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
248 of two vec_duplicates into a vec_concat.
250 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
252 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
253 Simplify vec_merge of vec_duplicate and vec_concat.
254 * config/aarch64/constraints.md (Utq): New constraint.
255 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
258 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
260 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
261 Simplify vec_merge of vec_duplicate and const_vector.
262 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
264 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
265 mode iterator. Update predicate on operand 1 to
266 handle non-const_vec constants. Delete constraints.
267 (*aarch64_combinez_be<mode>): Likewise for operand 2.
269 2017-11-08 Jakub Jelinek <jakub@redhat.com>
271 PR tree-optimization/78821
272 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
274 (store_operand_info::store_operand_info): Initialize it to false.
275 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
276 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
277 store in the group, and if chain_info is non-NULL, to ignore altogether
279 (compatible_load_p): Fail if bit_not_p does not match.
280 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
281 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
282 (pass_store_merging::process_store): Adjust
283 terminate_all_aliasing_chains calls to pass NULL in all current spots,
284 call terminate_all_aliasing_chains newly when adding a store into
285 a chain with non-NULL chain_info.
287 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
289 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
291 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
293 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
295 (aarch64_layout_frame): Initialise emit_frame_chain.
296 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
297 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
299 2017-11-08 Martin Liska <mliska@suse.cz>
301 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
302 of gimple_call_internal_p.
304 2017-11-07 Tom de Vries <tom@codesourcery.com>
306 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
308 2017-11-07 Tom de Vries <tom@codesourcery.com>
310 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
313 2017-11-08 Martin Liska <mliska@suse.cz>
316 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
318 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
320 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
322 2017-11-07 Jakub Jelinek <jakub@redhat.com>
325 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
326 (*cmp<mode>_ccz_1): New insn with $k alternative.
329 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
330 RTX_COMM_COMPARE as commutative as well.
331 (ix86_binary_operator_ok): Formatting fix.
332 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
333 *<code><mode>3<mask_name><round_saeonly_name>,
334 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
335 *<s>mul<mode>3_highpart<mask_name>,
336 *vec_widen_umult_even_v16si<mask_name>,
337 *vec_widen_umult_even_v8si<mask_name>,
338 *vec_widen_umult_even_v4si<mask_name>,
339 *vec_widen_smult_even_v16si<mask_name>,
340 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
341 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
342 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
343 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
344 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
345 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
346 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
347 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
348 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
349 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
350 ix86_binary_operator_ok. Formatting fixes.
351 (*<plusminus_insn><mode>3<mask_name><round_name>,
352 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
355 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
357 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
358 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
359 eq and ne if TARGET_ISEL.
360 (cmp): New code_iterator.
361 (UNS, UNSU_, UNSIK): New code_attrs.
362 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
363 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
365 ("eq<mode>3"): ... this.
366 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
368 ("ne<mode>3"): ... this.
370 2017-11-07 Julia Koval <julia.koval@intel.com>
373 * common/config/i386/i386-common.c
374 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
375 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
376 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
377 * config/i386/i386.opt: Ditto.
378 * config/i386/i386.c (ix86_target_string): Ditto.
379 (ix86_option_override_internal): Ditto.
380 (ix86_init_mpx_builtins): Move MPX to args2.
381 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
382 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
383 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
384 __builtin_ia32_vgf2p8affineinvqb_v32qi,
385 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
386 __builtin_ia32_vgf2p8affineinvqb_v16qi,
387 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
389 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
392 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
393 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
394 (zero-extendsidi peephole2): Remove peephole.
396 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
399 * stor-layout.c: Include attribs.h.
400 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
401 explicit lookup of "aligned" attribute.
403 2017-11-07 Andrew Waterman <andrew@sifive.com>
405 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
406 (riscv_expand_block_move): Likewise.
407 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
409 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
410 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
411 * config/riscv/riscv.c (riscv_block_move_straight): New function.
412 (riscv_adjust_block_mem): Likewise.
413 (riscv_block_move_loop): Likewise.
414 (riscv_expand_block_move): Likewise.
415 gcc/config/riscv/riscv.md (movmemsi): New pattern.
417 2017-11-07 Michael Clark <michaeljclark@mac.com>
419 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
420 (MUSL_DYNAMIC_LINKER): Likewise.
422 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
424 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
425 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
428 2017-11-07 Richard Biener <rguenther@suse.de>
430 * match.pd: Fix build.
432 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
433 Jackson Woodruff <jackson.woodruff@arm.com>
435 PR tree-optimization/71026
436 * match.pd: Canonicalize negate in division.
438 2017-11-07 Sudakshina Das <sudi.das@arm.com>
441 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
443 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
445 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
446 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
448 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
450 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
451 non-scalar integral types.
452 * match.pd (negate_expr_p): Handle MINUS_EXPR.
453 (-(A-B), -(~A)): New transformations.
455 2017-11-07 Tom de Vries <tom@codesourcery.com>
457 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
458 semicolon after "do {} while (0)".
459 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
460 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
461 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
462 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
463 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
465 2017-11-07 Tom de Vries <tom@codesourcery.com>
467 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
468 after "do {} while (0)".
469 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
470 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
471 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
472 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
473 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
475 2017-11-07 Tom de Vries <tom@codesourcery.com>
478 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
480 (arm_rtx_costs_internal): Add missing semicolon after
481 HANDLE_NARROW_SHIFT_ARITH call.
483 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
485 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
486 disable isel if it was not set explicitly.
488 2017-11-06 James Bowman <james.bowman@ftdichip.com>
490 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
492 (add_type_attribute) likewise.
494 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
496 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
497 of struct ix86_frame.
498 (ix86_initial_elimination_offset): Likewise.
499 (ix86_expand_split_stack_prologue): Likewise.
501 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
503 * tree-vrp.h (enum value_range_type): Update stale comment.
505 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
506 Alan Hayward <alan.hayward@arm.com>
507 David Sherwood <david.sherwood@arm.com>
509 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
510 (aarch64_expand_vec_perm_const): Take the number of units too.
511 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
512 (aarch64_expand_vec_perm_const): Likewise.
513 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
514 (vec_perm<mode>): Update accordingly.
516 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
517 Alan Hayward <alan.hayward@arm.com>
518 David Sherwood <david.sherwood@arm.com>
520 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
521 Take the number of units too.
522 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
523 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
524 but check for a vector mode before rather than after the call.
525 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
526 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
527 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
528 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
529 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
530 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
531 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
532 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
533 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
534 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
535 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
536 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
537 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
538 (aarch64_sqdmull2_n<mode>): Update accordingly.
540 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
541 Alan Hayward <alan.hayward@arm.com>
542 David Sherwood <david.sherwood@arm.com>
544 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
545 the number of units too.
546 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
547 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
548 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
549 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
550 (vec_store_lanesxi<mode>): Update accordingly.
552 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
553 Alan Hayward <alan.hayward@arm.com>
554 David Sherwood <david.sherwood@arm.com>
556 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
557 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
558 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
559 of units rather than the mode.
560 * config/aarch64/iterators.md (nunits): New mode attribute.
561 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
562 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
563 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
564 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
565 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
566 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
567 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
568 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
569 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
570 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
571 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
572 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
573 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
574 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
575 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
576 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
577 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
578 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
579 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
580 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
581 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
582 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
583 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
584 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
585 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
586 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
587 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
588 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
589 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
590 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
591 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
592 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
593 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
594 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
595 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
596 (aarch64_simd_vec_setv2di): Likewise.
598 2017-11-06 Carl Love <cel@us.ibm.com>
600 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
602 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
603 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
605 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
607 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
608 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
609 Add power 8 macro expansions.
610 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
611 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
612 power 8 instructions. (VSX_XXBR): Add iterator.
614 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
616 * config/arm/arm.md (predicable_short_it): Change default to "no",
617 improve documentation, remove uses that are identical to the default.
618 (enabled_for_depr_it): Rename to enabled_for_short_it.
619 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
620 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
621 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
622 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
623 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
625 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
628 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
629 float128 helper macros, which are no longer used after deleting
630 the old 'q' built-in functions, and moving the round to odd
631 built-in functions to being special built-in functions.
632 (BU_FLOAT128_2): Likewise.
633 (BU_FLOAT128_1_HW): Likewise.
634 (BU_FLOAT128_2_HW): Likewise.
635 (BU_FLOAT128_3_HW): Likewise.
636 (FABSQ): Delete old 'q' built-in functions.
637 (COPYSIGNQ): Likewise.
638 (SQRTF128_ODD): Move round to odd built-in functions to be
639 special built-in functions, so that we can handle
640 -mabi=ieeelongdouble.
641 (TRUNCF128_ODD): Likewise.
642 (ADDF128_ODD): Likewise.
643 (SUBF128_ODD): Likewise.
644 (MULF128_ODD): Likewise.
645 (DIVF128_ODD): Likewise.
646 (FMAF128_ODD): Likewise.
647 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
648 built-in names to 'f128'.
649 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
650 old 'q' built-in functions, as the machine independent code for
651 'f128' built-in functions handles this.
652 (rs6000_expand_builtin): Add expansion for float128 round to odd
653 functions, keying off on -mabi=ieeelongdouble of whether to use
654 the KFmode or TFmode variant.
655 (rs6000_init_builtins): Initialize the _Float128 round to odd
657 * doc/extend.texi (PowerPC Built-in Functions): Document the old
658 _Float128 'q' built-in functions are now mapped into the new
659 'f128' built-in functions.
661 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
663 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
664 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
666 2017-11-06 David Malcolm <dmalcolm@redhat.com>
669 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
670 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
671 * toplev.c: Include "ipa-fnsummary.h".
672 (toplev::finalize): Call ipa_fnsummary_c_finalize.
674 2017-11-06 Jakub Jelinek <jakub@redhat.com>
676 PR tree-optimization/82838
677 * gimple-ssa-store-merging.c
678 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
679 on a separate gimple_seq which is then appended to seq.
681 2017-11-06 Jeff Law <law@redhat.com>
684 * config/i386/i386.c (PROBE_INTERVAL): Remove.
685 (get_probe_interval): New functions.
686 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
687 (ix86_adjust_stack_and_probe): Likewise.
688 (output_adjust_stack_and_probe): Likewise.
689 (ix86_emit_probe_stack_range): Likewise.
690 (ix86_expand_prologue): Likewise.
692 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
694 PR tree-optimization/82816
695 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
696 if the modes of the two types are the same.
697 (convert_plusminus_to_widen): Likewise.
699 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
701 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
703 (usadv16qi): New define_expand.
704 (usadv8hi): New define_expand.
706 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
709 * ipa-inline-transform.c (update_noncloned_frequencies): Always
711 (inline_transform): Likewise.
712 * predict.c (counts_to_freqs): Remove useless conditional.
713 * profile-count.h (profile_count::apply_scale): Move sanity check.
714 * tree-inline.c (copy_bb): Always scale.
715 (copy_cfg_body): Likewise.
717 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
720 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
722 (*cmp_ite0): Add enabled_for_depr_it attribute.
723 (*cmp_ite1): Likewise.
725 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
727 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
730 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
732 * tree-vrp.c (vrp_int_const_binop): Return true on success and
733 return the value by pointer.
734 (extract_range_from_multiplicative_op_1): Update accordingly.
735 Return as soon as an operation fails.
737 2017-11-05 Tom de Vries <tom@codesourcery.com>
740 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
741 (DEF_SANITIZER_BUILTIN): ... here.
742 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
743 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
745 2017-11-05 Tom de Vries <tom@codesourcery.com>
748 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
750 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
751 ASM_OUTPUT_BEFORE_CASE_LABEL call.
752 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
754 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
755 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
756 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
758 2017-11-05 Tom de Vries <tom@codesourcery.com>
761 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
764 2017-11-04 Michael Clark <michaeljclark@mac.com>
766 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
767 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
769 (*addsi3_extended): Likewise.
770 (*addsi3_extended2): Likewise.
771 (<optab>si3): Likewise.
772 (<optab>di3): Likewise.
773 (<optab><mode>3): Likewise.
774 (<*optabe>si3_internal): Likewise.
775 (zero_extendqi<SUPERQI:mode>2): Likewise.
776 (*add<mode>hi3): Likewise.
777 (*xor<mode>hi3): Likewise.
778 (<optab>di3): Likewise.
779 (*<optab>si3_extend): Likewise.
780 (*sge<u>_<X:mode><GPR:mode>): Likewise.
781 (*slt<u>_<X:mode><GPR:mode>): Likewise.
782 (*sle<u>_<X:mode><GPR:mode>): Likewise.
784 2017-11-04 Andrew Waterman <andrew@sifive.com>
786 * config/riscv/riscv.c (riscv_option_override): Conditionally set
787 TARGET_STRICT_ALIGN based upon -mtune argument.
789 2017-11-04 Andrew Waterman <andrew@sifive.com>
791 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
793 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
795 config/i386/i386.c (choose_basereg): Use optional scratch
796 register and add assertion.
797 (x86_emit_outlined_ms2sysv_save): Use scratch register when
798 needed, and don't allocate stack.
799 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
800 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
801 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
803 2017-11-03 Jeff Law <law@redhat.com>
805 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
806 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
807 to probe at the start of a noreturn function.
809 2017-11-03 Jakub Jelinek <jakub@redhat.com>
811 PR tree-optimization/78821
812 * gimple-ssa-store-merging.c: Update the file comment.
813 (MAX_STORE_ALIAS_CHECKS): Define.
814 (struct store_operand_info): New type.
815 (store_operand_info::store_operand_info): New constructor.
816 (struct store_immediate_info): Add rhs_code and ops data members.
817 (store_immediate_info::store_immediate_info): Add rhscode, op0r
818 and op1r arguments to the ctor, initialize corresponding data members.
819 (struct merged_store_group): Add load_align_base and load_align
821 (merged_store_group::merged_store_group): Initialize them.
822 (merged_store_group::do_merge): Update them.
823 (merged_store_group::apply_stores): Pick the constant for
824 encode_tree_to_bitpos from one of the two operands, or skip
825 encode_tree_to_bitpos if neither operand is a constant.
826 (class pass_store_merging): Add process_store method decl. Remove
827 bool argument from terminate_all_aliasing_chains method decl.
828 (pass_store_merging::terminate_all_aliasing_chains): Remove
829 var_offset_p argument and corresponding handling.
830 (stmts_may_clobber_ref_p): New function.
831 (compatible_load_p): New function.
832 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
833 if there is overlap and rhs_code is not INTEGER_CST. For
834 non-overlapping stores terminate group if rhs is not mergeable.
835 (get_alias_type_for_stmts): Change first argument from
836 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
837 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
838 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
840 (get_location_for_stmts): Change first argument from
841 auto_vec<gimple *> & to vec<gimple *> &.
842 (struct split_store): Remove orig_stmts data member, add orig_stores.
843 (split_store::split_store): Create orig_stores rather than orig_stmts.
844 (find_constituent_stmts): Renamed to ...
845 (find_constituent_stores): ... this. Change second argument from
846 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
847 to info structures rather than the statements.
848 (split_group): Rename ALLOW_UNALIGNED argument to
849 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
850 it. Adjust find_constituent_stores caller.
851 (imm_store_chain_info::output_merged_store): Handle rhs_code other
852 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
853 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
854 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
855 (mem_valid_for_store_merging): New function.
856 (handled_load): New function.
857 (pass_store_merging::process_store): New method.
858 (pass_store_merging::execute): Use process_store method. Adjust
859 terminate_all_aliasing_chains caller.
861 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
863 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
864 Return true for more constants, symbols and label references.
865 (aarch64_valid_floating_const): Remove unused function.
867 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
870 * config/aarch64/aarch64.c (aarch64_layout_frame):
871 Undo forcing of LR at bottom of frame.
873 2017-11-03 Jeff Law <law@redhat.com>
876 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
877 for int_registers_saved.
879 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
880 extracted from tree-ssa-dom.c.
881 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
882 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
883 (record_equivalences_from_incoming_edge): Add additional argument
884 to single_pred_edge_ignoring_loop_edges call.
885 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
886 (uncprop_dom_walker::before_dom_children): Add additional argument
887 to single_pred_edge_ignoring_loop_edges call.
888 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
889 single_pred_edge_ignoring_loop_edges rather than open coding.
890 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
892 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
894 * match.pd (-(-A)): Rewrite.
896 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
898 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
899 (rs6000_emit_int_cmove): New declaration.
900 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
901 (rs6000_emit_sISEL): Delete.
902 (rs6000_emit_int_cmove): Make non-static.
903 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
904 instead of rs6000_emit_sISEL.
906 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
908 * asan.c (create_cond_insert_point): Maintain profile.
909 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
911 * basic-block.h (struct basic_block_def): Remove frequency.
912 (EDGE_FREQUENCY): Use to_frequency
913 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
915 (find_traces): Update to use to_frequency.
916 (find_traces_1_round): Likewise; use only IPA counts.
917 (bb_to_key): Likewise.
918 (connect_traces): Use IPA counts only.
919 (copy_bb_p): Update to use to_frequency.
920 (fix_up_crossing_landing_pad): Likewise.
921 (sanitize_hot_paths): Likewise.
922 * bt-load.c (basic_block_freq): Likewise.
923 * cfg.c (init_flow): Set count_max to uninitialized.
924 (check_bb_profile): Remove frequencies; check counts.
925 (dump_bb_info): Do not dump frequencies.
926 (update_bb_profile_for_threading): Update counts only.
927 (scale_bbs_frequencies_int): Likewise.
928 (MAX_SAFE_MULTIPLIER): Remove.
929 (scale_bbs_frequencies_gcov_type): Update counts only.
930 (scale_bbs_frequencies_profile_count): Update counts only.
931 (scale_bbs_frequencies): Update counts only.
932 * cfg.h (struct control_flow_graph): Add count-max.
933 (update_bb_profile_for_threading): Update prototype.
934 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
935 (find_many_sub_basic_blocks): Likewise.
936 * cfgcleanup.c (try_forward_edges): Likewise.
937 (try_crossjump_to_edge): Likewise.
938 * cfgexpand.c (expand_gimple_cond): Likewise.
939 (expand_gimple_tailcall): Likewise.
940 (construct_init_block): Likewise.
941 (construct_exit_block): Likewise.
942 * cfghooks.c (verify_flow_info): Check consistency of counts.
943 (dump_bb_for_graph): Do not dump frequencies.
944 (split_block_1): Do not update frequencies.
945 (split_edge): Do not update frequencies.
946 (make_forwarder_block): Do not update frequencies.
947 (duplicate_block): Do not update frequencies.
948 (account_profile_record): Do not update frequencies.
949 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
950 for global heuristics.
951 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
952 (expected_loop_iterations_unbounded): Use counts only.
953 * cfgloopmanip.c (scale_loop_profile): Simplify.
954 (create_empty_loop_on_edge): Simplify
956 (duplicate_loop_to_header_edge): Simplify
957 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
958 (update_br_prob_note): Take care of removing note when profile
960 (relink_block_chain): Do not dump frequency.
961 (rtl_account_profile_record): Use to_frequency.
962 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
963 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
964 (cgraph_update_edges_for_call_stmt_node): Likewise.
965 (cgraph_edge::verify_count_and_frequency): Update.
966 (cgraph_node::verify_node): Temporarily disable frequency verification.
967 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
969 (cgraph_edge::rebuild_edges): Convert to ipa counts.
970 * cgraphunit.c (init_lowered_empty_function): Do not initialize
972 (cgraph_node::expand_thunk): Update profile.
973 * except.c (dw2_build_landing_pads): Do not update frequency.
974 * final.c (compute_alignments): Use to_frequency.
975 (dump_basic_block_info): Do not dump frequency.
976 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
977 (dump_gimple_bb_header): Do not dump frequency.
978 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
980 * gimple-streamer-in.c (input_bb): Do not stream frequency.
981 * gimple-streamer-out.c (output_bb): Do not stream frequency.
982 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
983 (init_before_recovery): Do not update frequency.
984 (sched_create_recovery_edges): Do not update frequency.
985 * hsa-gen.c (convert_switch_statements): Do not update frequency.
986 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
987 (ipa_cp_c_finalize): Set max_count to uninitialized.
988 * ipa-fnsummary.c (get_minimal_bb): Use counts.
989 (param_change_prob): Use counts.
990 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
992 * ipa-split.c (consider_split): Use to_frequency.
993 (split_function): Use to_frequency.
994 * ira-build.c (loop_compare_func): Likewise.
995 (mark_loops_for_removal): Likewise.
996 (mark_all_loops_for_removal): Likewise.
997 * loop-doloop.c (doloop_modify): Do not update frequency.
998 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1000 * lto-streamer-in.c (input_function): Update count_max.
1001 * omp-expand.c (expand_omp_taskreg): Update count_max.
1002 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1003 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1004 (maybe_hot_count_p): Use ipa counts only.
1005 (maybe_hot_bb_p): Simplify.
1006 (maybe_hot_edge_p): Simplify.
1007 (probably_never_executed): Do not take frequency argument.
1008 (probably_never_executed_bb_p): Do not pass frequency.
1009 (probably_never_executed_edge_p): Likewise.
1010 (combine_predictions_for_bb): Check that profile is nonzero.
1011 (propagate_freq): Do not set frequency.
1012 (drop_profile): Simplify.
1013 (counts_to_freqs): Simplify.
1014 (expensive_function_p): Use to_frequency.
1015 (propagate_unlikely_bbs_forward): Simplify.
1016 (determine_unlikely_bbs): Simplify.
1017 (estimate_bb_frequencies): Add hack to silence graphite issues.
1018 (compute_function_frequency): Use ipa counts.
1019 (pass_profile::execute): Update.
1020 (rebuild_frequencies): Use counts only.
1021 (force_edge_cold): Use counts only.
1022 * profile-count.c (profile_count::dump): Dump new count types.
1023 (profile_count::differs_from_p): Check compatiblity.
1024 (profile_count::to_frequency): New function.
1025 (profile_count::to_cgraph_frequency): New function.
1026 * profile-count.h (struct function): Declare.
1027 (enum profile_quality): Add profile_guessed_local and
1028 profile_guessed_global0.
1029 (class profile_proability): Decrease number of bits to 29;
1030 update from_reg_br_prob_note and to_reg_br_prob_note.
1031 (class profile_count: Update comment; decrease number of bits
1032 to 61. Check compatibility.
1033 (profile_count::compatible_p): New private member function.
1034 (profile_count::ipa_p): New member function.
1035 (profile_count::operator<): Handle global zero correctly.
1036 (profile_count::operator>): Handle global zero correctly.
1037 (profile_count::operator<=): Handle global zero correctly.
1038 (profile_count::operator>=): Handle global zero correctly.
1039 (profile_count::nonzero_p): New member function.
1040 (profile_count::force_nonzero): New member function.
1041 (profile_count::max): New member function.
1042 (profile_count::apply_scale): Handle IPA scalling.
1043 (profile_count::guessed_local): New member function.
1044 (profile_count::global0): New member function.
1045 (profile_count::ipa): New member function.
1046 (profile_count::to_frequency): Declare.
1047 (profile_count::to_cgraph_frequency): Declare.
1048 * profile.c (OVERLAP_BASE): Delete.
1049 (compute_frequency_overlap): Delete.
1050 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1051 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1052 * sched-ebb.c (rank): Use counts only.
1053 * shrink-wrap.c (handle_simple_exit): Use counts only.
1054 (try_shrink_wrapping): Use counts only.
1055 (place_prologue_for_one_component): Use counts only.
1056 * tracer.c (find_best_predecessor): Use to_frequency.
1057 (find_trace): Use to_frequency.
1058 (tail_duplicate): Use to_frequency.
1059 * trans-mem.c (expand_transaction): Do not update frequency.
1060 * tree-call-cdce.c: Do not update frequency.
1061 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1062 (gimple_merge_blocks): Likewise.
1063 (gimple_split_edge): Likewise.
1064 (gimple_duplicate_sese_region): Likewise.
1065 (gimple_duplicate_sese_tail): Likewise.
1066 (move_sese_region_to_fn): Likewise.
1067 (gimple_account_profile_record): Likewise.
1068 (insert_cond_bb): Likewise.
1069 * tree-complex.c (expand_complex_div_wide): Likewise.
1070 * tree-eh.c (lower_resx): Update profile.
1071 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1073 (initialize_cfun): Do not initialize frequencies
1074 (freqs_to_counts): Delete.
1075 (copy_cfg_body): Ignore count parameter.
1076 (copy_body): Update.
1077 (expand_call_inline): Update count_max.
1078 (optimize_inline_calls): Update count_max.
1079 (tree_function_versioning): Update count_max.
1080 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1081 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1083 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1084 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1085 (try_peel_loop): Likewise.
1086 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1088 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1089 (tree_transform_and_unroll_loop): Do not use frequencies
1090 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1091 Use reliable prediction only.
1092 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1093 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1094 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1095 probability scaling.
1096 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1097 not update frequency
1098 (any_remaining_duplicated_blocks): Likewise.
1099 (update_profile): Likewise.
1100 (estimated_freqs_path): Delete.
1101 (freqs_to_counts_path): Delete.
1102 (clear_counts_path): Delete.
1103 (ssa_fix_duplicate_block_edges): Likewise.
1104 (duplicate_thread_path): Likewise.
1105 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1106 * tree-tailcall.c (decrease_profile): Do not update frequency.
1107 (eliminate_tail_call): Likewise.
1108 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1109 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1110 (optimize_mask_stores): Likewise.
1111 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1112 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1113 (ubsan_expand_ptr_ifn): Update profile.
1114 * value-prof.c (gimple_ic): Simplify.
1115 * value-prof.h (gimple_ic): Update prototype.
1116 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1117 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1119 (want_inline_self_recursive_call_p): Likewise.
1120 (resolve_noninline_speculation): Only cummulate defined counts.
1121 (inline_small_functions): Use nonzero_p.
1122 (ipa_inline): Do not access freed node.
1124 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1126 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1127 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1129 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1131 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1132 non-legitimate address.
1134 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1136 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1137 (*lt0_<mode>di, *lt0_<mode>si): New.
1139 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1141 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1142 TARGET_PAIRED_FLOAT.
1144 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1145 Jim Wilson <jim.wilson@linaro.org>
1147 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1148 * config/aarch64/aarch64-tune.md: Regenerated.
1149 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1150 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1152 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1154 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
1155 (arc_expand_prologue): Restore blink for millicode.
1156 * config/arc/linux.h (LINK_EH_SPEC): Defined.
1158 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
1161 * config/i386/i386.c (ix86_vector_duplicate_value): Use
1162 gen_vec_duplicate after forcing the scalar into a register.
1164 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
1166 * combine (try_combine): Print the insns input to try_combine to the
1169 2017-11-02 Steve Ellcey <sellcey@cavium.com>
1172 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
1173 Remove second argument from aarch64_process_target_attr call.
1174 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
1176 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
1178 (aarch64_handle_attr_arch): Remove second argument.
1179 (aarch64_handle_attr_cpu): Ditto.
1180 (aarch64_handle_attr_tune): Ditto.
1181 (aarch64_handle_attr_isa_flags): Ditto.
1182 (aarch64_process_one_target_attr): Ditto.
1183 (aarch64_process_target_attr): Ditto.
1184 (aarch64_option_valid_attribute_p): Remove second argument.
1185 on aarch64_process_target_attr call.
1187 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1189 * diagnostic.c: Include "selftest-diagnostic.h".
1190 (selftest::assert_location_text): New function.
1191 (selftest::test_diagnostic_get_location_text): New function.
1192 (selftest::diagnostic_c_tests): Call it.
1194 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1196 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
1197 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
1198 (class selftest::test_diagnostic_context): Move to...
1199 * selftest-diagnostic.c: New file.
1200 * selftest-diagnostic.h: New file.
1202 2017-11-02 James Bowman <james.bowman@ftdichip.com>
1204 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
1205 offset range for FT32B.
1206 * config/ft32/ft32.h: option "mcompress" enables relaxation.
1207 * config/ft32/ft32.md: Add TARGET_NOPM.
1208 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
1209 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
1211 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
1213 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
1215 2017-11-02 Jeff Law <law@redhat.com>
1217 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1218 virtual keyword on FINAL OVERRIDE members.
1220 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1221 virtuals together. Add virtual destructor.
1222 (substitute_and_fold_engine): Similarly.
1224 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1226 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1228 2017-11-02 Richard Biener <rguenther@suse.de>
1230 PR tree-optimization/82795
1231 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1233 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1235 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1237 (gcc_GAS_CHECK_FEATURE): Remove.
1238 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1239 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1240 * configure: Regenerate.
1242 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1244 * config/arc/arc.c (hwloop_optimize): Account for empty
1247 2017-11-02 Richard Biener <rguenther@suse.de>
1250 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1251 Truncate ARRAY_REF index and element size.
1253 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1255 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1257 2017-11-01 Jeff Law <law@redhat.com>
1259 * tree-ssa-ccp.c (ccp_folder): New class derived from
1260 substitute_and_fold_engine.
1261 (ccp_folder::get_value): New member function.
1262 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1263 (ccp_fold_stmt): Remove prototype.
1264 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1265 * tree-ssa-copy.c (copy_folder): New class derived from
1266 substitute_and_fold_engine.
1267 (copy_folder::get_value): Renamed from get_value.
1268 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1269 * tree-vrp.c (vrp_folder): New class derived from
1270 substitute_and_fold_engine.
1271 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1272 (vrp_folder::get_value): New member function.
1273 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1274 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1275 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1276 provide a class interface to folder/substitute routines.
1277 (ssa_prop_fold_stmt_fn): Remove typedef.
1278 (ssa_prop_get_value_fn): Likewise.
1279 (subsitute_and_fold): Remove prototype.
1280 (replace_uses_in): Likewise.
1281 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1282 Renamed from replace_uses_in. Call the virtual member function
1283 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1284 (substitute_and_fold_dom_walker): Remove initialization of
1285 data member entries for calbacks. Add substitute_and_fold_engine
1286 member and initialize it.
1287 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1288 member functions for get_value, replace_phi_args_in c
1289 replace_uses_in, and fold_stmt calls.
1290 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1291 substitute_and_fold. Remove assert. Update ctor call.
1293 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1294 (ssa_prop_visit_phi_fn): Likewise.
1295 (class ssa_propagation_engine): New class to provide an interface
1297 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1299 (ssa_prop_visit_phi): Likewise.
1300 (ssa_propagation_engine::simulate_stmt): Moved into class.
1301 Call visit_phi/visit_stmt from the class rather than via
1302 file scoped static variables.
1303 (ssa_propagation_engine::simulate_block): Moved into class.
1304 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1305 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1306 set file scoped statics for the visit_stmt/visit_phi callbacks.
1307 * tree-complex.c (complex_propagate): New class derived from
1308 ssa_propagation_engine.
1309 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1310 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1311 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1313 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1314 ssa_propagation_engine.
1315 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1316 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1317 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1318 * tree-ssa-copy.c (copy_prop): New class derived from
1319 ssa_propagation_engine.
1320 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1321 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1322 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1323 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1324 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1325 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1326 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1328 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1330 PR rtl-optimization/82778
1331 PR rtl-optimization/82597
1332 * compare-elim.c (struct comparison): Add in_a_setter field.
1333 (find_comparison_dom_walker::before_dom_children): Remove killed
1334 bitmap and df_simulate_find_defs call, instead walk the defs.
1335 Compute last_setter and initialize in_a_setter. Merge definitions
1336 with first initialization for a few variables.
1337 (try_validate_parallel): Use insn_invalid_p instead of
1338 recog_memoized. Return insn rather than just the pattern.
1339 (try_merge_compare): Fix up comment. Don't uselessly test if
1340 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1342 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1345 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1346 Alan Hayward <alan.hayward@arm.com>
1347 David Sherwood <david.sherwood@arm.com>
1349 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1350 aarch64_hard_regno_nregs to get the number of registers
1353 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1354 Alan Hayward <alan.hayward@arm.com>
1355 David Sherwood <david.sherwood@arm.com>
1357 * config/aarch64/constraints.md (Upl): Rename to...
1359 * config/aarch64/aarch64.md
1360 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
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_add_constant_internal)
1368 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
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/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1376 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1377 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1379 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1380 of a single element.
1381 * config/aarch64/iterators.md: Add a comment above the permute
1382 unspecs to say that they are generated directly by
1383 aarch64_expand_vec_perm_const.
1384 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1386 2017-11-01 Nathan Sidwell <nathan@acm.org>
1388 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1390 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1392 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1393 and medany code models, and describe what they do.
1395 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1397 Revert accidental duplicate:
1399 * combine.c (can_change_dest_mode): Reject changes in
1400 REGMODE_NATURAL_SIZE.
1402 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1404 PR rtl-optimization/64682
1405 PR rtl-optimization/69567
1406 PR rtl-optimization/69737
1407 PR rtl-optimization/82683
1408 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1409 register mentioned in the note, drop the note, unless it came from I3,
1410 in which case it should go to I3 again.
1412 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1414 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1415 and return false if not.
1416 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1418 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1420 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1421 range is known to be empty.
1423 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1424 Alan Hayward <alan.hayward@arm.com>
1425 David Sherwood <david.sherwood@arm.com>
1427 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1428 and CONST_VECTOR_NUNITS instead of computing the number of units from
1429 the byte sizes of the vector and element.
1430 (simplify_binary_operation_1): Likewise.
1431 (simplify_const_binary_operation): Likewise.
1432 (simplify_ternary_operation): Likewise.
1434 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1435 Alan Hayward <alan.hayward@arm.com>
1436 David Sherwood <david.sherwood@arm.com>
1438 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1439 (int_mem_offset): ...this new function.
1440 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1441 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1442 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1445 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1446 Alan Hayward <alan.hayward@arm.com>
1447 David Sherwood <david.sherwood@arm.com>
1449 * lower-subreg.c (interesting_mode_p): New function.
1450 (compute_costs, find_decomposable_subregs, decompose_register)
1451 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1452 (resolve_clobber, dump_choices): Use it.
1454 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1455 Alan Hayward <alan.hayward@arm.com>
1456 David Sherwood <david.sherwood@arm.com>
1458 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
1460 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1461 Alan Hayward <alan.hayward@arm.com>
1462 David Sherwood <david.sherwood@arm.com>
1464 * alias.c (find_base_value, find_base_term): Only process integer
1465 truncations. Check the precision rather than the size.
1467 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1468 Alan Hayward <alan.hayward@arm.com>
1469 David Sherwood <david.sherwood@arm.com>
1471 * machmode.h (is_narrower_int_mode): New function
1472 * optabs.c (expand_float, expand_fix): Use it.
1473 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1475 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1476 Alan Hayward <alan.hayward@arm.com>
1477 David Sherwood <david.sherwood@arm.com>
1479 * rtl.h (narrower_subreg_mode): New function.
1480 * ira-color.c (update_costs_from_allocno): Use it.
1482 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1483 Alan Hayward <alan.hayward@arm.com>
1484 David Sherwood <david.sherwood@arm.com>
1486 * optabs-query.h (convert_optab_p): New function, split out from...
1487 (convert_optab_handler): ...here.
1488 (widening_optab_handler): Delete.
1489 (find_widening_optab_handler): Remove permit_non_widening parameter.
1490 (find_widening_optab_handler_and_mode): Likewise. Provide an
1491 override that operates on mode class wrappers.
1492 * optabs-query.c (widening_optab_handler): Delete.
1493 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1494 parameter. Assert that the two modes are the same class and that
1495 the "from" mode is narrower than the "to" mode. Use
1496 convert_optab_handler instead of widening_optab_handler.
1497 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1498 instead of widening_optab_handler.
1499 * expr.c (expand_expr_real_2): Update calls to
1500 find_widening_optab_handler.
1501 * optabs.c (expand_widen_pattern_expr): Likewise.
1502 (expand_binop_directly): Take the insn_code as a parameter.
1503 (expand_binop): Only call find_widening_optab_handler for
1504 conversion optabs; use optab_handler otherwise. Update calls
1505 to find_widening_optab_handler and expand_binop_directly.
1506 Use convert_optab_handler instead of widening_optab_handler.
1507 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1508 find_widening_optab_handler and use scalar_mode rather than
1510 (convert_plusminus_to_widen): Likewise.
1512 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1513 Alan Hayward <alan.hayward@arm.com>
1514 David Sherwood <david.sherwood@arm.com>
1516 * machmode.h (fixed_size_mode): New class.
1517 * rtl.h (get_pool_mode): Return fixed_size_mode.
1518 * gengtype.c (main): Add fixed_size_mode.
1519 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1520 (get_raw_arg_mode): Likewise.
1521 * doc/tm.texi: Regenerate.
1522 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1523 * targhooks.c (default_get_reg_raw_mode): Likewise.
1524 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1525 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1526 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1527 (msp430_get_raw_result_mode): Likewise.
1528 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1529 * dbxout.c (dbxout_parms): Require fixed-size modes.
1530 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1531 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1532 * omp-low.c (lower_oacc_reductions): Likewise.
1533 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1534 (simplify_subreg): Update accordingly.
1535 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1536 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1537 that aren't fixed-size.
1538 (get_pool_mode): Return a fixed_size_mode.
1539 (output_constant_pool_2): Take a fixed_size_mode.
1541 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1542 Alan Hayward <alan.hayward@arm.com>
1543 David Sherwood <david.sherwood@arm.com>
1545 * doc/rtl.texi (vec_series): Document.
1546 (const): Say that the operand can be a vec_series.
1547 * rtl.def (VEC_SERIES): New rtx code.
1548 * rtl.h (const_vec_series_p_1): Declare.
1549 (const_vec_series_p): New function.
1550 * emit-rtl.h (gen_const_vec_series): Declare.
1551 (gen_vec_series): Likewise.
1552 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1553 (gen_vec_series): Likewise.
1554 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1555 * simplify-rtx.c (simplify_unary_operation): Handle negations
1557 (simplify_binary_operation_series): New function.
1558 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1559 (test_vector_ops_series): New function.
1560 (test_vector_ops): Call it.
1561 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1562 gen_const_vec_series.
1563 (altivec_lvsr): Likewise.
1564 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1566 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1567 Alan Hayward <alan.hayward@arm.com>
1568 David Sherwood <david.sherwood@arm.com>
1570 * doc/rtl.texi (const): Update description of address constants.
1571 Say that vector constants are allowed too.
1572 * common.md (E, F): Use CONSTANT_P instead of checking for
1574 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1575 checking for CONST_VECTOR.
1576 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1578 * expr.c (expand_expr_real_2): Check for vector modes instead
1579 of checking for CONST_VECTOR.
1580 * rtl.h (const_vec_p): New function.
1581 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1582 (unwrap_const_vec_duplicate): Handle them here too.
1584 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1585 David Malcolm <dmalcolm@redhat.com>
1586 Alan Hayward <alan.hayward@arm.com>
1587 David Sherwood <david.sherwood@arm.com>
1589 * rtl.h (vec_duplicate_p): New function.
1590 * selftest-rtl.c (assert_rtx_eq_at): New function.
1591 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1592 (assert_rtx_eq_at): Declare.
1593 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1594 * selftest-run-tests.c (selftest::run_tests): Call it.
1595 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1596 (simplify_unary_operation_1): Recursively handle vector duplicates.
1597 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1599 (simplify_subreg): Handle subregs of vector duplicates.
1600 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1601 (selftest::simplify_rtx_c_tests): New functions.
1603 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1604 Alan Hayward <alan.hayward@arm.com>
1605 David Sherwood <david.sherwood@arm.com>
1607 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1608 (gen_vec_duplicate): Likewise.
1609 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1611 (gen_const_vector): ...here.
1612 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1613 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1614 whose elements are all equal.
1615 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1616 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1617 (simplify_relational_operation): Likewise.
1618 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1620 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1621 (aarch64_expand_vector_init): Likewise.
1622 * config/arm/arm.c (neon_vdup_constant): Likewise.
1623 (neon_expand_vector_init): Likewise.
1624 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1625 (arm_block_set_unaligned_vect): Likewise.
1626 (arm_block_set_aligned_vect): Likewise.
1627 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1628 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1629 (expand_vec_perm_even_odd_pack): Likewise.
1630 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1631 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1632 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1633 gen_const_vec_duplicate.
1634 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1635 * config/mips/mips.c (mips_gen_const_int_vector): Use
1636 gen_const_vec_duplicate.
1637 (mips_expand_vector_init): Use CONST0_RTX.
1638 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1639 (define_split): Use gen_const_vec_duplicate.
1640 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1641 (define_split): Use gen_const_vec_duplicate.
1642 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1643 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1644 * config/spu/spu.c (spu_const): Likewise.
1646 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1647 Alan Hayward <alan.hayward@arm.com>
1648 David Sherwood <david.sherwood@arm.com>
1650 * combine.c (can_change_dest_mode): Reject changes in
1651 REGMODE_NATURAL_SIZE.
1653 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
1655 * configure.ac (--enable-libssp): New.
1656 (gcc_cv_libc_provides_ssp): Check for explicit setting before
1657 trying to determine target-specific default. Adjust indentation.
1658 * configure: Regenerated.
1659 * doc/install.texi (Configuration): Expand --disable-libssp
1662 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
1664 config/i386/i386.c (ix86_expand_epilogue): Correct stack
1667 2017-10-31 Martin Jambor <mjambor@suse.cz>
1670 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1672 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1674 * auto-profile.c (autofdo_source_profile::read): Use
1675 UNKNOWN_LOCATION rather than 0.
1676 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1677 (warning_at): ...this overload.
1678 (warning_at_rich_loc_n): Rename to...
1679 (warning_n): ...this overload.
1680 (error_at_rich_loc): Rename to...
1681 (error_at): ...this overload.
1682 (pedwarn_at_rich_loc): Rename to...
1683 (pedwarn): ...this overload.
1684 (permerror_at_rich_loc): Rename to...
1685 (permerror): ...this overload.
1686 (inform_at_rich_loc): Rename to...
1687 (inform): ...this overload.
1688 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1689 (diagnostic_n_impl_richloc): Rename to...
1690 (diagnostic_n_impl): ...this rich_location *-based decl.
1691 (inform_at_rich_loc): Rename to...
1692 (inform): ...this, and add an assertion.
1693 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1694 (warning_at_rich_loc): Rename to...
1695 (warning_at): ...this, and add an assertion.
1696 (warning_at_rich_loc_n): Rename to...
1697 (warning_n): ...this, and add an assertion.
1698 (warning_n): Update location_t-based implementation for removal of
1699 location_t-based diagnostic_n_impl.
1700 (pedwarn_at_rich_loc): Rename to...
1701 (pedwarn): ...this, and add an assertion.
1702 (permerror_at_rich_loc): Rename to...
1703 (permerror): ...this, and add an assertion.
1704 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1705 (error_at_rich_loc): Rename to...
1706 (error_at): ...this, and add an assertion.
1707 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1708 (driver::do_spec_on_infiles): Likewise.
1709 * substring-locations.c (format_warning_va): Update for renaming
1710 of inform_at_rich_loc.
1712 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1714 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1715 _Float<N>X built-in functions so that the variant without the
1716 "__builtin_" prefix is only enabled for the GNU C and Objective C
1717 languages when they are in non-strict ANSI/ISO mode.
1718 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1719 * target.def (floatn_builtin_p): Add a target hook to control
1720 whether _Float<N> and _Float<N>X built-in functions without the
1721 "__builtin_" prefix are enabled, and return true for C and
1722 Objective C in the default hook. Include langhooks.h in
1724 * targhooks.h (default_floatn_builtin_p): Likewise.
1725 * targhooks.c (default_floatn_builtin_p): Likewise.
1726 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1727 floatn_builtin_p target hook.
1728 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1730 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1731 Eric Botcazou <ebotcazou@adacore.com>
1733 PR rtl-optimization/81803
1734 * lra-constraints.c (curr_insn_transform): Also reload the whole
1735 register for a strict subreg no wider than a word if this is for
1736 a WORD_REGISTER_OPERATIONS target.
1738 2017-10-31 Jason Merrill <jason@redhat.com>
1740 * gdbinit.in: Skip over inlines from timevar.h.
1742 2017-10-31 Martin Liska <mliska@suse.cz>
1744 * doc/gcov.texi: Document new option.
1745 * gcov.c (print_usage): Likewise print it.
1746 (process_args): Support the argument.
1747 (format_count): New function.
1748 (format_gcov): Use the function.
1750 2017-10-31 Martin Liska <mliska@suse.cz>
1752 * gcov.c (struct name_map): do not use typedef.
1753 Define operator== and operator<.
1754 (name_search): Remove.
1755 (name_sort): Remove.
1756 (main): Do not allocate names.
1757 (process_file): Add vertical space.
1758 (generate_results): Use std::find.
1759 (release_structures): Do not release memory.
1760 (find_source): Use std::find.
1762 2017-10-31 Martin Liska <mliska@suse.cz>
1764 * gcov.c (struct line_info): Remove it's typedef.
1765 (line_info::line_info): Add proper ctor.
1766 (line_info::has_block): Do not use a typedef.
1767 (struct source_info): Do not use typedef.
1768 (circuit): Likewise.
1769 (get_cycles_count): Likewise.
1770 (output_intermediate_file): Iterate via vector iterator.
1771 (add_line_counts): Use std::vector methods.
1772 (accumulate_line_counts): Likewise.
1773 (output_lines): Likewise.
1775 2017-10-31 Martin Liska <mliska@suse.cz>
1777 * gcov.c (struct source_info): Remove typedef.
1778 (source_info::source_info): Add proper ctor.
1779 (accumulate_line_counts): Use struct, not it's typedef.
1780 (output_gcov_file): Likewise.
1781 (output_lines): Likewise.
1782 (main): Do not allocate an array.
1783 (output_intermediate_file): Use size of vector container.
1784 (process_file): Resize the vector.
1785 (generate_results): Do not preallocate, use newly added vector
1787 (release_structures): Do not release sources.
1788 (find_source): Use vector methods.
1789 (add_line_counts): Do not use typedef.
1791 2017-10-31 Martin Liska <mliska@suse.cz>
1793 * doc/gcov.texi: Document that.
1794 * gcov.c (add_line_counts): Mark lines with a non-executed
1796 (output_line_beginning): Handle such lines.
1797 (output_lines): Pass new argument.
1798 (output_intermediate_file): Print it in intermediate format.
1800 2017-10-31 Martin Liska <mliska@suse.cz>
1802 * color-macros.h: New file.
1803 * diagnostic-color.c: Factor out color related to macros to
1805 * doc/gcov.texi: Document -k option.
1806 * gcov.c (INCLUDE_STRING): Include string.h.
1807 (print_usage): Add -k option.
1808 (process_args): Parse it.
1809 (pad_count_string): New function.
1810 (output_line_beginning): Likewise.
1811 (DEFAULT_LINE_START): New macro.
1812 (output_lines): Support color output.
1814 2017-10-31 Martin Liska <mliska@suse.cz>
1816 PR gcov-profile/82633
1817 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1818 their interaction with GCOV infrastructure.
1819 * configure.ac: Add -fkeep-{inline,static}-functions to
1821 * configure: Regenerate.
1823 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1826 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1828 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1831 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1832 into a register if it does not fit into an immediate offset field.
1834 2017-10-31 Olivier Hainque <hainque@adacore.com>
1836 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1838 2017-10-31 Julia Koval <julia.koval@intel.com>
1840 * config.gcc: Add gfniintrin.h.
1841 * config/i386/gfniintrin.h: New.
1842 * config/i386/i386-builtin-types.def
1843 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1844 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1845 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1846 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1847 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1848 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1849 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1850 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1851 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1852 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1853 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1854 * config/i386/immintrin.h: Include gfniintrin.h.
1855 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1857 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1859 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1861 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1863 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1864 (arm_ashldi3_1bit): Remove pattern.
1865 (ashrdi3): Remove shift by 1 expansion.
1866 (arm_ashrdi3_1bit): Remove pattern.
1867 (lshrdi3): Remove shift by 1 expansion.
1868 (arm_lshrdi3_1bit): Remove pattern.
1869 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1870 cost of ashldi3 by 1.
1871 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1872 (<shift>di3_neon): Likewise.
1874 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1876 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1877 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1878 and (*aarch64_simd_mov<VQ:mode>).
1879 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1880 pattern alternative.
1881 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1882 attributes to match pattern alternative.
1884 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1886 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1887 * config/rs6000/emmintrin.h: New file.
1888 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1890 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1892 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1893 * (movdi_vfp_cortexa8): Remove pattern.
1895 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1897 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1899 (Specific, alpha*-dec-osf5.1): Remove.
1900 (Specific, mips-sgi-irix5): Remove.
1901 (Specific, mips-sgi-irix6): Remove.
1903 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1906 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1907 arguments to clear_bit_region_be.
1909 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1911 * gimplify.c: Include memmodel.h.
1913 2017-10-30 Martin Jambor <mjambor@suse.cz>
1915 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1916 condition whether loop should be executed at all.
1918 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1920 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1921 gimple folding of vec_madd() intrinsics.
1922 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1923 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1924 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1926 2017-10-30 Richard Biener <rguenther@suse.de>
1928 PR tree-optimization/82762
1930 2017-10-23 Richard Biener <rguenther@suse.de>
1932 PR tree-optimization/82129
1934 2017-08-01 Richard Biener <rguenther@suse.de>
1936 PR tree-optimization/81181
1937 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1938 (compute_antic): ... end of iteration here.
1940 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1942 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1943 -std=iso9899:2017 and -std=gnu17.
1944 * doc/standards.texi (C Language): Document C17 support.
1945 * doc/cpp.texi (Overview): Mention -std=c17.
1946 (Standard Predefined Macros): Document C11 and C17 values of
1947 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1948 * doc/extend.texi (Inline): Do not list individual options for
1949 standards newer than C99.
1950 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1952 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1955 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1957 * asan.c (asan_finish_file): Align asan globals array by shadow
1960 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1963 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1964 (struct store_immediate_info): Add bitregion_start and bitregion_end
1966 (store_immediate_info::store_immediate_info): Add brs and bre
1967 arguments and initialize bitregion_{start,end} from those.
1968 (struct merged_store_group): Add bitregion_start, bitregion_end,
1969 align_base and mask fields. Drop unnecessary struct keyword from
1970 struct store_immediate_info. Add do_merge method.
1971 (clear_bit_region_be): Use memset instead of loop storing zeros.
1972 (merged_store_group::do_merge): New method.
1973 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
1974 stores as long as the surrounding bitregions have no gaps.
1975 (merged_store_group::merge_overlapping): Use do_merge.
1976 (merged_store_group::apply_stores): Test that bitregion_{start,end}
1977 is byte aligned, rather than requiring that start and width are
1978 byte aligned. Drop unnecessary struct keyword from
1979 struct store_immediate_info. Allocate and populate also mask array.
1980 Make start of the arrays relative to bitregion_start rather than
1981 start and size them according to bitregion_{end,start} difference.
1982 (struct imm_store_chain_info): Drop unnecessary struct keyword from
1983 struct store_immediate_info.
1984 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
1985 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
1986 struct keyword from struct store_immediate_info.
1987 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
1988 between stores as long as the surrounding bitregions have no gaps.
1990 (struct split_store): Add orig non-static data member.
1991 (split_store::split_store): Initialize orig to false.
1992 (find_constituent_stmts): Return store_immediate_info *, non-NULL
1993 if there is exactly a single original stmt. Change stmts argument
1994 to pointer from reference, if NULL, don't push anything to it. Add
1995 first argument, use it to optimize skipping over orig stmts that
1996 are known to be before bitpos already. Simplify.
1997 (split_group): Return unsigned int count how many stores are or
1998 would be needed rather than a bool. Add allow_unaligned argument.
1999 Change split_stores argument from reference to pointer, if NULL,
2000 only do a dry run computing how many stores would be produced.
2001 Rewritten algorithm to use both alignment and misalign if
2002 !allow_unaligned and handle bitfield stores with gaps.
2003 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2004 from bitregion_start instead of start. Compute allow_unaligned
2005 here, if true, do 2 split_group dry runs to compute which one
2006 produces fewer stores and prefer aligned if equal. Punt if
2007 new count is bigger or equal than original before emitting any
2008 statements, rather than during that. Remove no longer needed
2009 new_ssa_names tracking. Replace num_stmts with
2010 split_stores.length (). Use 32-bit stack allocated entries
2011 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2012 if possible. Handle bitfields with gaps.
2013 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2014 Compute bitregion_{start,end} for the stores and construct
2015 store_immediate_info with that. Formatting fixes.
2017 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2020 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2021 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2023 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2025 * gimplify.c: Include tm_p.h.
2027 * common.opt (gcoff): Re-add as ignored option.
2028 (gcoff1, gcoff2, gcoff3): Likewise.
2030 * Makefile.in (OBJS): Delete sdbout.o.
2031 (GTFILES): Delete $(srcdir)/sdbout.c.
2032 * debug.h: Delete sdb_debug_hooks.
2033 * final.c: Delete sdbout.h include.
2034 (final_scan_insn): Delete SDB_DEBUG check.
2035 (rest_of_clean_state): Likewise.
2036 * output.h: Delete sdb_begin_function_line.
2039 * toplev.c: Delete sdbout.h include.
2040 (process_options): Delete SDB_DEBUG check.
2041 * tree-core.h (tree_type_common): Delete pointer field of
2043 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2044 TYPE_SYMTAB_POINTER.
2045 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2046 (TYPE_SYMTAB_IS_POINTER): Delete.
2047 (TYPE_SYMTAB_IS_DIE): Renumber.
2048 * xcoffout.c: Refer to former sdbout.c file.
2049 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2051 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2052 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2053 (Debugging Options): Delete -gcoff.
2054 (-gstabs): Delete SDB reference.
2056 (-gcoff@var{level}): Delete.
2057 * doc/passes.texi (Debugging information output): Delete SDB and
2058 sdbout.c references.
2059 * doc/tm.texi: Regenerate.
2060 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2061 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2063 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2064 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2065 (SDB_DEBUGGING_INFO): Delete.
2066 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2067 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2068 * target.def (output_source_filename): Delete COFF reference.
2070 * common.opt (gcoff): Delete.
2071 (gxcoff+): Update Negative chain.
2072 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2074 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2075 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2076 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2078 (expand_function_start): Change sdb reference to past tense.
2079 (expand_function_end): Change sdb reference to past tense.
2080 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2081 * opts.c (debug_type_names): Delete coff entry.
2082 (common_handle_option): Delete OPT_gcoff case.
2083 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2085 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2086 * config/cris/cris.h: Delete SDB reference in comment.
2087 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2088 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2089 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2090 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2092 (ix86_expand_prologue): Likewise.
2093 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2094 * config/ia64/ia64.h: Likewise.
2095 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2096 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2098 * config/mmix/mmix.h: Likewise.
2099 * config/nds32/nds32.c: Likewise.
2100 * config/stormy/storym16.h: Likewise.
2101 * config/visium/visium.h: Likewise.
2102 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2104 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2106 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2107 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2108 FRAME_POINTER_REGNUM point at high end of local var area.
2110 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2112 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2113 Move comment around. Do not reset best_edge for a copiable
2114 destination if the copy would cause a partition change.
2115 (better_edge_p): Remove redundant check.
2117 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2119 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2121 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2123 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2124 for math functions that have _Float<N> and _Float<N>X variants.
2125 (mathfn_built_in_2): Add support for math functions that have
2126 _Float<N> and _Float<N>X variants.
2127 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2128 (expand_builtin_mathfn_ternary): Add support for fma with
2129 _Float<N> and _Float<N>X variants.
2130 (expand_builtin): Likewise.
2131 (fold_builtin_3): Likewise.
2132 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2133 create math function _Float<N> and _Float<N>X variants as external
2135 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2136 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2137 the __builtin_ prefix and if not strict ansi, without the prefix.
2138 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2139 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2140 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2141 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2142 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2143 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2144 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2145 function signatures for fma _Float<N> and _Float<N>X variants.
2146 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2147 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2148 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2149 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2150 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2151 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2152 * gencfn-macros.c (print_case_cfn): Add support for math functions
2153 that have _Float<N> and _Float<N>X variants.
2154 (print_define_operator_list): Likewise.
2155 (fltfn_suffixes): Likewise.
2157 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
2158 for math functions that have _Float<N> and _Float<N>X variants.
2159 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
2160 and _Float<N>X variants.
2161 (COPYSIGN): Likewise.
2164 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
2165 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
2167 (integer_valued_read_call_p): Likewise.
2168 * fold-const-call.c (fold_const_call_ss): Likewise.
2169 (fold_const_call_sss): Add support for copysign, fmin, and fmax
2170 _Float<N> and _Float<N>X variants.
2171 (fold_const_call_ssss): Add support for fma _Float<N> and
2172 _Float<N>X variants.
2173 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
2174 support for copysign and fma _Float<N> and _Float<N>X variants.
2175 (backprop::process_builtin_call_use): Likewise.
2176 * tree-call-cdce.c (can_test_argument_range); Add support for
2177 sqrt _Float<N> and _Float<N>X variants.
2178 (edom_only_function): Likewise.
2179 (get_no_error_domain): Likewise.
2180 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
2181 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
2182 copysign _Float<N> and _Float<N>X variants.
2183 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
2184 handled by machine independent code.
2185 (FMAF128): Likewise.
2186 * doc/cpp.texi (Common Predefined Macros): Document defining
2187 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
2188 fma _Float<N> and _Float<N>X variants.
2190 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2193 * config/i386/i386-modes.def (CCFPU): Remove definition.
2194 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
2195 (ix86_cc_modes_compatible): Ditto.
2196 (ix86_expand_carry_flag_compare): Ditto.
2197 (ix86_expand_int_movcc): Ditto.
2198 (ix86_expand_int_addcc): Ditto.
2199 (ix86_reverse_condition): Ditto.
2200 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
2201 Return true/false for unordered/ordered fp comparisons.
2202 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
2203 (ix86_prepare_fp_compare_args): Update for rename.
2204 (ix86_expand_fp_compare): Update for rename. Generate unordered
2205 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
2206 (ix86_expand_sse_compare_and_jump): Ditto.
2207 * config/i386/predicates.md (fcmov_comparison_operator):
2208 Remove CCFPU mode handling.
2209 (ix86_comparison_operator): Ditto.
2210 (ix86_carry_flag_operator): Ditto.
2211 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
2212 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
2213 (*cmpu<mode>_cc_i387): Ditto.
2214 (FPCMP): Remove mode iterator.
2215 (unord): Remove mode attribute.
2216 (unord_subst): New define_subst transformation
2217 (unord): New define_subst attribute.
2219 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2220 (*cmpi<unord>xf_i387): Ditto.
2221 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2222 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2223 using unord_subst transformation.
2224 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2225 (round_saeonly): Also handle CCFP mode.
2226 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2227 Remove UNSPEC_SAHF unspec handling.
2229 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2231 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2233 2017-10-27 Jeff Law <law@redhat.com>
2235 * gimple-ssa-sprintf.c: Include domwalk.h.
2236 (class sprintf_dom_walker): New class, derived from dom_walker.
2237 (sprintf_dom_walker::before_dom_children): New function.
2238 (struct call_info): Moved into sprintf_dom_walker class
2239 (compute_formath_length, handle_gimple_call): Likewise.
2240 (sprintf_length::execute): Call the dominator walker rather
2241 than walking the statements.
2243 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2244 gimple statement locations.
2245 (check_array_bounds): Corresponding changes. Get the statement's
2246 location directly from wi->stmt.
2248 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2251 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2253 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2255 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2256 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2258 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2261 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2262 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2263 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2264 maybe_get_pool_constant.
2265 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2268 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2270 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2271 2.26 caveat. Update gas and gld versions.
2272 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2275 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2277 * cgraph.h (set_malloc_flag): Declare.
2278 * cgraph.c (set_malloc_flag_1): New function.
2279 (set_malloc_flag): Likewise.
2280 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2281 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2283 (read_ipa_call_summary): Add support for reading is_return_callee.
2284 (write_ipa_call_summary): Stream is_return_callee.
2285 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2286 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2287 ipa-prop.h, ipa-fnsummary.h.
2288 (pure_const_names): Change to static.
2289 (malloc_state_e): Define.
2290 (malloc_state_names): Define.
2291 (funct_state_d): Add field malloc_state.
2292 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2293 (check_retval_uses): New function.
2294 (malloc_candidate_p): Likewise.
2295 (analyze_function): Add support for malloc attribute.
2296 (pure_const_write_summary): Stream malloc_state.
2297 (pure_const_read_summary): Add support for reading malloc_state.
2298 (dump_malloc_lattice): New function.
2299 (propagate_malloc): New function.
2300 (warn_function_malloc): New function.
2301 (ipa_pure_const::execute): Call propagate_malloc and
2302 ipa_free_fn_summary.
2303 (pass_local_pure_const::execute): Add support for malloc attribute.
2304 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2305 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2307 2017-10-27 Martin Liska <mliska@suse.cz>
2309 PR gcov-profile/82457
2310 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2311 for fork-like functions to be properly instrumented.
2313 2017-10-27 Richard Biener <rguenther@suse.de>
2316 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2317 info when we redirected EH.
2319 2017-10-26 Michael Collison <michael.collison@arm.com>
2321 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2323 (<optab>_trunchf<GPI:mode>2: New pattern.
2324 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2325 * config/aarch64/iterators.md (wv): New mode attribute.
2326 (vf, VF): New mode attributes.
2327 (vgp, VGP): New mode attributes.
2328 (s): Update attribute with SImode and DImode prefixes.
2330 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2332 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2333 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2334 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2335 (nios2_option_overide): Initialize it. Don't allow R0-relative
2336 addressing with PIC.
2337 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2338 (nios2_symbolic_constant_p): Likewise.
2339 (nios2_legitimate_address_p): Likewise.
2340 (nios2_r0rel_section_name_p): New.
2341 (nios2_symbol_ref_in_r0rel_data_p): New.
2342 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2343 (r0rel_constant_p): New.
2344 (nios2_print_operand_address): Handle r0rel_constant_p.
2345 (nios2_cdx_narrow_form_p): Likewise.
2346 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2347 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2348 (Nios II Options): Document -mr0rel-sec.
2350 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2352 * config/nios2/nios2.c: Include xregex.h.
2353 (nios2_gprel_sec_regex): New.
2354 (nios2_option_overide): Initialize it. Don't allow GP-relative
2355 addressing with PIC.
2356 (nios2_small_section_name_p): Check for regex match.
2357 * config/nios2/nios2.opt (mgprel-sec=): New option.
2358 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2359 (Nios II Options): Document -mgprel-sec.
2361 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2363 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2365 2017-10-26 Tom de Vries <tom@codesourcery.com>
2367 PR tree-optimization/82707
2368 * gimple.c (gimple_copy): Fix unsharing of
2369 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2371 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2373 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2374 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2375 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2376 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2377 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2378 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2379 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2380 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2381 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2382 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2383 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2384 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2385 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2386 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2387 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2388 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2389 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2391 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2393 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2395 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2396 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2397 warning to rs6000.c. Remove the Undocumented flag, since it has
2399 (-mabi=ibmlongdouble): Likewise.
2400 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2401 already set, set the default format for long double.
2402 (rs6000_debug_reg_global): Print whether long double is IBM or
2404 (rs6000_option_override_internal): Rework setting long double
2405 format. Only warn if the user is changing the long double default
2406 and they did not use -Wno-psabi.
2407 * doc/invoke.texi (PowerPC options): Update the documentation for
2408 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2410 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2411 Alan Hayward <alan.hayward@arm.com>
2412 David Sherwood <david.sherwood@arm.com>
2414 * rtl.h (wider_subreg_mode): New function.
2415 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2416 rather than an unsigned int *.
2417 * ira-color.c (regno_max_ref_width): Replace with...
2418 (regno_max_ref_mode): ...this new variable.
2419 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2420 Use wider_subreg_mode.
2421 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2422 rather than an unsigned int *.
2423 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2424 (process_alt_operands): Likewise.
2425 (invariant_p): Likewise.
2426 * lra-spills.c (assign_mem_slot): Likewise.
2427 (add_pseudo_to_slot): Likewise.
2428 * lra.c (collect_non_operand_hard_regs): Likewise.
2429 (add_regs_to_insn_regno_info): Likewise.
2430 * reload1.c (regno_max_ref_width): Replace with...
2431 (regno_max_ref_mode): ...this new variable.
2432 (reload): Update accordingly. Update call to
2433 ira_sort_regnos_for_alter_reg.
2434 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2435 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2436 (scan_paradoxical_subregs): Likewise.
2438 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2440 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2441 (aarch64_frame): Add emit_frame_chain boolean.
2442 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2443 Move eh_return case to aarch64_layout_frame.
2444 (aarch64_layout_frame): Initialize emit_frame_chain.
2445 (aarch64_expand_prologue): Use emit_frame_chain.
2447 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2449 * config/aarch64/aarch64.c (aarch64_layout_frame):
2450 Ensure LR is always stored at the bottom of the callee-saves.
2451 Remove rarely used frame layout which saves callee-saves at top of
2452 frame, so the store of LR can be used as a valid probe in all cases.
2454 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2456 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2457 Improve unaligned TImode/TFmode base/offset split.
2459 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2460 Alan Hayward <alan.hayward@arm.com>
2461 David Sherwood <david.sherwood@arm.com>
2463 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2464 * combine.c (find_single_use_1): Likewise.
2465 (expand_field_assignment): Likewise.
2466 (move_deaths): Likewise.
2467 * lra-constraints.c (simplify_operand_subreg): Likewise.
2468 (curr_insn_transform): Likewise.
2469 * lra.c (collect_non_operand_hard_regs): Likewise.
2470 (add_regs_to_insn_regno_info): Likewise.
2471 * rtlanal.c (reg_referenced_p): Likewise.
2472 (covers_regno_no_parallel_p): Likewise.
2474 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2476 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2477 Don't print any bits outside the precision of the value.
2478 * wide-int.cc (test_printing): Add some new tests.
2480 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2482 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2483 supports -xbrace_comment option.
2484 * configure: Regenerate.
2485 * config.in: Regenerate.
2486 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2487 (ASM_CPU_SPEC): Use it.
2489 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2491 * target.def (static_rtx_alignment): New hook.
2492 * targhooks.h (default_static_rtx_alignment): Declare.
2493 * targhooks.c (default_static_rtx_alignment): New function.
2494 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2495 * doc/tm.texi: Regenerate.
2496 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2497 instead of targetm.constant_alignment. Remove call to
2499 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2500 (cris_preferred_mininum_alignment): New function, split out from...
2501 (cris_constant_alignment): ...here.
2502 (cris_static_rtx_alignment): New function.
2503 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2505 (ix86_constant_alignment): ...here.
2506 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2507 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2508 (mmix_static_rtx_alignment): New function.
2509 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2510 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2512 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2515 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2516 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2518 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2520 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2521 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2522 mark as a sign-extending load.
2523 (local_pic_load_u): Define.
2525 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2528 * fold-const.c (operand_equal_for_comparison_p): Also return true
2529 if ARG0 is a simple variant of ARG1 with narrower precision.
2530 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2532 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2534 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2536 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2537 scatter_static, scatter_per_elt.
2538 * x86-tune-costs.h: Add new cost entries.
2540 2017-10-25 Richard Biener <rguenther@suse.de>
2542 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2543 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2544 class pass_fre): Move to ...
2545 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2546 class pass_fre): ... here and adjust for statistics.
2548 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2551 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2552 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2553 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2554 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2557 2017-10-25 Richard Biener <rguenther@suse.de>
2559 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2560 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2561 eliminate_push_avail, eliminate_insert): Move inside...
2562 (class eliminate_dom_walker): ... this class in preparation
2564 (fini_eliminate): Remove by merging with ...
2565 (eliminate): ... this function. Adjust for class changes.
2566 (pass_pre::execute): Remove fini_eliminate call.
2567 (pass_fre::execute): Likewise.
2569 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2572 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2573 (VPERMI2, VPERMI2I): New mode iterators.
2574 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2575 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2577 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2578 mode iterator. Remove 3 old define_insn patterns.
2579 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2580 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2581 VPERMI2 mode iterator, remove the other two expanders.
2582 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2583 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2584 instructions, remove the other two patterns.
2585 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2586 mode iterator, remove the other two patterns.
2587 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2588 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2589 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2590 and adjust argument order accordingly.
2591 (ix86_expand_vec_perm): Adjust caller.
2592 (expand_vec_perm_1): Likewise.
2593 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2594 (expand_vec_perm_vpermt2_vpshub2): ... this.
2595 (ix86_expand_vec_perm_const_1): Adjust caller.
2596 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2599 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2600 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2601 (vec_shl_<mode>): Remove unused expander.
2602 (avx512bw_<shift_insn><mode>3): New define_insn.
2603 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2604 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2606 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2609 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2612 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2614 PR rtl-optimization/82396
2615 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2616 (autopref_multipass_init): Simplify initialization.
2617 (autopref_rank_data): Simplify sort order.
2618 * gcc/sched-int.h (autopref_multipass_data_): Remove
2619 multi_mem_insn_p, min_offset and max_offset.
2621 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2624 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2625 Check special value of flag_omit_frame_pointer.
2626 (aarch64_can_eliminate): Likewise.
2627 (aarch64_override_options_after_change_1): Simplify handling of
2628 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2630 2017-10-24 Richard Biener <rguenther@suse.de>
2632 PR tree-optimization/82697
2633 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2634 zero for conditional load and unconditional store.
2636 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2638 * doc/install.texi: Document bootstrap-cet.
2640 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2643 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2644 ENDBR instruction at function entrance if function is only
2647 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2650 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2651 patterns to better describe from which operation the CF is computed.
2652 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2653 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2654 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
2655 is 0, use _0 suffixed expanders instead of emitting a comparison
2658 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
2660 * config/i386/i386.md(*movsf_internal, *movdf_internal):
2661 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
2663 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
2666 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
2667 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
2668 * loop-iv.c (iv_get_reaching_def): Likewise.
2669 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
2670 variable is promoted and the partition contains undefined values.
2672 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2674 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2676 (nios2_address_cost): Define.
2677 (nios2_legitimize_address): Recognize (exp + constant) directly.
2678 (TARGET_ADDRESS_COST): Define.
2680 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2682 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2683 (nios2_symbolic_memory_operand_p): Declare.
2684 (nios2_split_large_constant): Declare.
2685 (nios2_split_symbolic_memory_operand): Declare.
2686 * config/nios2/nios2.c: Adjust includes.
2687 (nios2_symbolic_constant_allowed): New.
2688 (nios2_symbolic_constant_p): New.
2689 (nios2_plus_symbolic_constant_p): New.
2690 (nios2_valid_addr_expr_p): Recognize addresses involving
2692 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2693 (nios2_symbolic_memory_operand_p): New.
2694 (nios2_large_constant_p): New.
2695 (nios2_split_large_constant): New.
2696 (nios2_split_plus_large_constant): New.
2697 (nios2_split_symbolic_memory_operand): New.
2698 (nios2_legitimize_address): Code refactoring. Handle addresses
2699 involving symbolic constants.
2700 (nios2_emit_move_sequence): Likewise.
2701 (nios2_print_operand): Improve error output.
2702 (nios2_print_operand_address): Handle LO_SUM.
2703 (nios2_cdx_narrow_form_p): Likewise.
2704 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2705 operands involving symbolic constants.
2706 (movhi_internal, movsi_internal): Likewise.
2707 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2708 (extendhisi2, extendqi<mode>2): Likewise.
2710 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2712 * tree-pass.h (PROP_rtl_split_insns): Define.
2713 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2715 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2717 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2719 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2722 * target.def (const_not_ok_for_debug_p): Default to
2723 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2724 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2725 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2726 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2727 which targetm.const_not_ok_for_debug_p returned true.
2728 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2730 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2732 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2733 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2734 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2736 (ix86_const_not_ok_for_debug_p): New function.
2737 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2738 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2740 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2743 * system.h: Conditionally include "unique-ptr.h" if
2744 INCLUDE_UNIQUE_PTR is defined.
2745 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2746 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2748 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2750 * config/rl78/rl78.md: New define_expand "subdi3".
2752 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2755 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2756 DF_REF_INSN if DF_REF_INSN_INFO is false.
2758 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2760 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2761 xmm_move instead of sse_move.
2762 (sse_store_index): New function.
2763 (ix86_register_move_cost): Be more sensible about mismatch stall;
2764 model AVX moves correctly; make difference between sse->integer and
2766 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2767 moves; make difference between SSE and AVX.
2768 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2769 and zmm_move. Increase size of sse load and store tables;
2770 add unaligned load and store tables; add ssemmx_to_integer.
2771 * x86-tune-costs.h: Update all entries according to real
2772 move latencies from Agner Fog's manual and chip documentation.
2774 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2777 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2778 * config/i386/constraints.md (Wf): New constraint.
2779 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2780 (cmp<dwi>_doubleword): Removed.
2781 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2782 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2783 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2784 expand with cmp<dwi>_doubleword. For LTU and GEU use
2785 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2787 * common.opt (gcolumn-info): Enable by default.
2788 * doc/invoke.texi (gcolumn-info): Document new default.
2790 2017-10-23 Richard Biener <rguenther@suse.de>
2792 PR tree-optimization/82672
2793 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2794 Fold the stmt if we propagated into it.
2796 2017-10-23 Richard Biener <rguenther@suse.de>
2798 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2799 (bitmap_remove_expr_from_set): ... this. All callers call this
2800 for non-constant values.
2801 (bitmap_set_subtract): Rename to...
2802 (bitmap_set_subtract_expressions): ... this. Adjust and
2804 (bitmap_set_contains_value): Remove superfluous check.
2805 (bitmap_set_replace_value): Inline into single caller ...
2806 (bitmap_value_replace_in_set): ... here and simplify.
2807 (dependent_clean): Merge into ...
2808 (clean): ... this using an overload. Adjust.
2809 (prune_clobbered_mems): Adjust.
2810 (compute_antic_aux): Likewise.
2811 (compute_partial_antic_aux): Likewise.
2813 2017-10-23 Richard Biener <rguenther@suse.de>
2815 PR tree-optimization/82129
2817 2017-08-01 Richard Biener <rguenther@suse.de>
2819 PR tree-optimization/81181
2820 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2821 (compute_antic): ... end of iteration here.
2823 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2825 * target.def (starting_frame_offset): New hook.
2826 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2827 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2828 * doc/tm.texi.in: Regenerate.
2829 * hooks.h (hook_hwi_void_0): Declare.
2830 * hooks.c (hook_hwi_void_0): New function.
2831 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2832 STARTING_FRAME_OFFSET.
2833 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2834 * reload1.c (reload): Likewise.
2835 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2836 instead of STARTING_FRAME_OFFSET.
2837 * function.c (try_fit_stack_local): Likewise.
2838 (assign_stack_local_1): Likewise
2839 (instantiate_virtual_regs): Likewise.
2840 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2841 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2842 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2843 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2844 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2845 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2846 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2847 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2848 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2849 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2850 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2851 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2852 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2853 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2854 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2855 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2856 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2857 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2858 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2859 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2860 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2861 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2862 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2863 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2864 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2865 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2866 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2867 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2868 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2869 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2870 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2871 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2872 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2873 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2874 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2875 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2876 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2877 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2878 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2879 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2880 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2881 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2882 return a HOST_WIDE_INT.
2883 (avr_builtin_setjmp_frame_value): Use it instead of
2884 STARTING_FRAME_OFFSET.
2885 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2886 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2887 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2889 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2890 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2891 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2892 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2893 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2894 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2895 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2896 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2897 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2898 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2899 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2900 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2902 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2903 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2904 * config/mips/mips.c (mips_compute_frame_info): Refer to
2905 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2906 (mips_starting_frame_offset): New function.
2907 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2908 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2909 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2910 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2911 and return a HOST_WIDE_INT.
2912 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2913 (mmix_initial_elimination_offset): Refer to
2914 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2915 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2916 * config/pa/pa.c (pa_starting_frame_offset): New function.
2917 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2918 (pa_expand_prologue): Likewise.
2919 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2920 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2921 !FRAME_GROWS_DOWNWARD handling to...
2922 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2923 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2924 !FRAME_GROWS_DOWNWARD handling to...
2925 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2926 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2927 !FRAME_GROWS_DOWNWARD handling to...
2928 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2929 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2931 (rs6000_starting_frame_offset): New function.
2932 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2933 !FRAME_GROWS_DOWNWARD handling to...
2934 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2935 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2936 !FRAME_GROWS_DOWNWARD handling to...
2937 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2938 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2939 !FRAME_GROWS_DOWNWARD handling to...
2940 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2941 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2942 (rs6000_starting_frame_offset): New function.
2943 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2944 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2945 * config/vax/vax.c (vax_starting_frame_offset): New function.
2946 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2947 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2948 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2949 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2950 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2951 * system.h (STARTING_FRAME_OFFSET): Poison.
2953 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2955 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2956 SCALAR_TYPE_MODE instead of TYPE_MODE.
2958 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2959 Alan Hayward <alan.hayward@arm.com>
2960 David Sherwood <david.sherwood@arm.com>
2962 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2964 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2965 Alan Hayward <alan.hayward@arm.com>
2966 David Sherwood <david.sherwood@arm.com>
2968 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
2970 2017-10-23 Richard Biener <rguenther@suse.de>
2972 PR tree-optimization/82129
2973 * tree-ssa-pre.c (bitmap_set_and): Remove.
2974 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
2975 canonicalizing expressions in the set to those with lowest
2976 ID rather than taking that from the first edge.
2978 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2980 * combine.c (rtx_equal_for_field_assignment_p): Use
2981 byte_lowpart_offset.
2983 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2984 Alan Hayward <alan.hayward@arm.com>
2985 David Sherwood <david.sherwood@arm.com>
2987 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
2988 to a SUBREG_PROMOTED_VAR.
2990 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2991 Alan Hayward <alan.hayward@arm.com>
2992 David Sherwood <david.sherwood@arm.com>
2994 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
2995 (expand_debug_source_expr): Likewise.
2996 * combine.c (combine_simplify_rtx): Likewise.
2997 * cse.c (fold_rtx): Likewise.
2998 * optabs.c (expand_float): Likewise.
2999 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3000 (simplify_binary_operation_1): Likewise.
3002 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3003 Alan Hayward <alan.hayward@arm.com>
3004 David Sherwood <david.sherwood@arm.com>
3006 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3007 (record_promoted_value): Likewise.
3008 * expr.c (expand_expr_real_2): Likewise.
3009 * ree.c (update_reg_equal_equiv_notes): Likewise.
3010 (combine_set_extension): Likewise.
3011 * rtlanal.c (low_bitmask_len): Likewise.
3012 * simplify-rtx.c (neg_const_int): Likewise.
3013 (simplify_binary_operation_1): Likewise.
3015 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3016 Alan Hayward <alan.hayward@arm.com>
3017 David Sherwood <david.sherwood@arm.com>
3019 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3020 * regcprop.c (maybe_mode_change): Likewise.
3021 * reload1.c (alter_reg): Likewise.
3023 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3025 * inchash.h (inchash::hash::add_wide_int): New function.
3026 * lto-streamer-out.c (hash_tree): Use it.
3028 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3030 * inchash.h (inchash::hash::add_wide_int): Rename to...
3031 (inchash::hash::add_hwi): ...this.
3032 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3033 (polymorphic_call_target_hasher::hash): Likewise.
3034 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3035 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3036 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3037 * lto-streamer-out.c (hash_tree): Likewise.
3038 * optc-save-gen.awk: Likewise.
3039 * tree.c (add_expr): Likewise.
3041 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3044 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3045 for ordered inequality comparisons even with TARGET_IEEE_FP.
3047 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3050 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3051 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3052 Expand with cmp<dwi>_doubleword.
3054 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3056 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3057 List CET intrinsics.
3058 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3059 specific to -fcf-protection option.
3061 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3063 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3064 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3065 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3066 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3067 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3068 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3069 (extra_objs): Add cet.o for Linux/x86 targets.
3070 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3071 * config/i386/cet.c: New file.
3072 * config/i386/cetintrin.h: Likewise.
3073 * config/i386/t-cet: Likewise.
3074 * config/i386/cpuid.h (bit_SHSTK): New.
3075 (bit_IBT): Likewise.
3076 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3077 pass IBT and SHSTK bits.
3078 * config/i386/i386-builtin-types.def
3079 (VOID_FTYPE_UNSIGNED_PVOID): New.
3080 (VOID_FTYPE_UINT64_PVOID): Likewise.
3081 * config/i386/i386-builtin.def: Add CET intrinsics.
3082 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3083 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3084 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3085 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3087 * config/i386/i386.c (rest_of_insert_endbranch): New.
3088 (pass_data_insert_endbranch): Likewise.
3089 (pass_insert_endbranch): Likewise.
3090 (make_pass_insert_endbranch): Likewise.
3091 (ix86_notrack_prefixed_insn_p): Likewise.
3092 (ix86_target_string): Add -mibt, -mshstk flags.
3093 (ix86_option_override_internal): Add flag_cf_protection
3095 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3096 (ix86_print_operand): Add 'notrack' prefix output.
3097 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3098 (ix86_expand_builtin): Expand CET intrinsics.
3099 (x86_output_mi_thunk): Add 'endbranch' instruction.
3100 * config/i386/i386.h (TARGET_IBT): New.
3101 (TARGET_IBT_P): Likewise.
3102 (TARGET_SHSTK): Likewise.
3103 (TARGET_SHSTK_P): Likewise.
3104 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3105 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3106 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3107 (builtin_setjmp_setup): New pattern.
3108 (builtin_longjmp): Likewise.
3109 (rdssp<mode>): Likewise.
3110 (incssp<mode>): Likewise.
3111 (saveprevssp): Likewise.
3112 (rstorssp): Likewise.
3113 (wrss<mode>): Likewise.
3114 (wruss<mode>): Likewise.
3115 (setssbsy): Likewise.
3116 (clrssbsy): Likewise.
3117 (nop_endbr): Likewise.
3118 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3120 * config/i386/immintrin.h: Include <cetintrin.h>.
3121 * config/i386/linux-common.h
3122 (file_end_indicate_exec_stack_and_cet): New prototype.
3123 (TARGET_ASM_FILE_END): New.
3125 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3127 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3128 latencies instead of having separate table; make difference between
3129 integer and float costs.
3130 * i386.h (processor_costs): Remove scalar_stmt_cost,
3131 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3132 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3134 * x86-tune-costs.h: Remove entries which has been removed in
3135 procesor_costs from all tables; make cond_taken_branch_cost
3136 and cond_not_taken_branch_cost COST_N_INSNS based.
3138 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3140 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3142 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3144 * config/i386/i386.md (isa): Remove fma_avx512f.
3145 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3146 <avx512>_fmadd_<mode>_mask3<round_name>,
3147 <avx512>_fmsub_<mode>_mask<round_name>,
3148 <avx512>_fmsub_<mode>_mask3<round_name>,
3149 <avx512>_fnmadd_<mode>_mask<round_name>,
3150 <avx512>_fnmadd_<mode>_mask3<round_name>,
3151 <avx512>_fnmsub_<mode>_mask<round_name>,
3152 <avx512>_fnmsub_<mode>_mask3<round_name>,
3153 <avx512>_fmaddsub_<mode>_mask<round_name>,
3154 <avx512>_fmaddsub_<mode>_mask3<round_name>,
3155 <avx512>_fmsubadd_<mode>_mask<round_name>,
3156 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
3157 (*vec_widen_umult_even_v16si<mask_name>,
3158 *vec_widen_smult_even_v16si<mask_name>): Likewise.
3159 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
3161 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3163 * extend.texi: Add 'nocf_check' documentation.
3164 * gimple.texi: Add second parameter to
3165 gimple_build_call_from_tree.
3166 * invoke.texi: Add -fcf-protection documentation.
3167 * rtl.texi: Add REG_CALL_NOTRACK documenation.
3169 2017-10-20 Richard Biener <rguenther@suse.de>
3171 PR tree-optimization/82473
3172 * tree-vect-loop.c (vectorizable_reduction): Properly get at
3173 the largest input type.
3175 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3177 * c-attribs.c (handle_nocf_check_attribute): New function.
3178 (c_common_attribute_table): Add 'nocf_check' handling.
3179 * gimple-parser.c: Add second argument NULL to
3180 gimple_build_call_from_tree.
3181 * attrib.c (comp_type_attributes): Check nocf_check attribute.
3182 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
3184 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
3185 * common.opt: Add fcf-protection flag.
3186 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
3187 * flag-types.h: Add enum cf_protection_level.
3188 * gimple.c (gimple_build_call_from_tree): Add second parameter.
3189 Add 'nocf_check' attribute propagation to gimple call.
3190 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
3191 (gimple_build_call_from_tree): Update prototype.
3192 (gimple_call_nocf_check_p): New function.
3193 (gimple_call_set_nocf_check): Likewise.
3194 * gimplify.c: Add second argument to gimple_build_call_from_tree.
3195 * ipa-icf.c: Add nocf_check attribute in statement hash.
3196 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
3197 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
3198 * toplev.c (process_options): Add flag_cf_protection handling.
3200 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3202 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
3204 2017-10-20 Richard Biener <rguenther@suse.de>
3206 PR tree-optimization/82603
3207 * tree-if-conv.c (predicate_mem_writes): Make sure to only
3208 remove false predicated stores.
3210 2017-10-20 Richard Biener <rguenther@suse.de>
3212 * graphite-isl-ast-to-gimple.c
3213 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
3214 Remove return value and simplify, dump copied stmt after lhs
3216 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
3217 Reduce dump verbosity.
3218 (gsi_insert_earliest): Likewise.
3219 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3220 * graphite.c (print_global_statistics): Adjust dumping.
3221 (print_graphite_scop_statistics): Likewise.
3222 (print_graphite_statistics): Do not dump loops here.
3223 (graphite_transform_loops): But here.
3225 2017-10-20 Nicolas Roche <roche@adacore.com>
3227 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3228 * configure: Regenerate.
3230 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3233 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3234 functions when optimizing replace GIMPLE_RETURN stmts with
3235 calls to __builtin_unreachable ().
3238 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3239 for -fsanitize=thread link of executables.
3240 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3241 link of executables.
3244 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3246 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3247 define_insns for logical vector shifts to use VI248_AVX512BW
3248 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3249 condition, useless isa and prefix attributes. Change the first
3250 2 of these define_insns to ...
3251 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3252 define_insn for avx512vl.
3253 (<shift_insn><mode>3): ... and this, new define_insn without
3254 masking for non-avx512vl.
3257 * config/i386/sse.md (*andnot<mode>3,
3258 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3259 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3260 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3261 not applied use empty suffix even for TARGET_AVX512VL.
3262 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3263 is applied, supply evex,evex or evex,evex,evex instead of just
3266 2017-10-20 Julia Koval <julia.koval@intel.com>
3268 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3269 (OPTION_MASK_ISA_GFNI_UNSET): New.
3270 (ix86_handle_option): Handle OPT_mgfni.
3271 * config/i386/cpuid.h (bit_GFNI): New.
3272 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3273 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3274 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3275 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3276 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3277 * config/i386/i386.opt: Add mgfni.
3279 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3281 * config/msp430/msp430.c (msp430_option_override): Disable
3282 -fdelete-null-pointer-checks.
3283 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3285 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3287 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3288 of x87 and SSE instructions.
3290 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3292 * asan.c (create_cond_insert_point): Do not update edge count.
3293 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3294 (afdo_propagate_circuit): Likewise.
3295 (afdo_calculate_branch_prob): Likewise.
3296 (afdo_annotate_cfg): Likewise.
3297 * basic-block.h (struct edge_def): Remove count.
3298 (edge_def::count): New accessor.
3299 * bb-reorder.c (rotate_loop): Update.
3300 (find_traces_1_round): Update.
3301 (connect_traces): Update.
3302 (sanitize_hot_paths): Update.
3303 * cfg.c (unchecked_make_edge): Update.
3304 (make_single_succ_edge): Update.
3305 (check_bb_profile): Update.
3306 (dump_edge_info): Update.
3307 (update_bb_profile_for_threading): Update.
3308 (scale_bbs_frequencies_int): Update.
3309 (scale_bbs_frequencies_gcov_type): Update.
3310 (scale_bbs_frequencies_profile_count): Update.
3311 (scale_bbs_frequencies): Update.
3312 * cfganal.c (connect_infinite_loops_to_exit): Update.
3313 * cfgbuild.c (compute_outgoing_frequencies): Update.
3314 (find_many_sub_basic_blocks): Update.
3315 * cfgcleanup.c (try_forward_edges): Update.
3316 (try_crossjump_to_edge): Update
3317 * cfgexpand.c (expand_gimple_cond): Update
3318 (expand_gimple_tailcall): Update
3319 (construct_exit_block): Update
3320 * cfghooks.c (verify_flow_info): Update
3321 (redirect_edge_succ_nodup): Update
3322 (split_edge): Update
3323 (make_forwarder_block): Update
3324 (duplicate_block): Update
3325 (account_profile_record): Update
3326 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3327 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3328 * cfgloopmanip.c (scale_loop_profile): Update.
3330 (lv_adjust_loop_entry_edge): Update.
3331 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3332 (force_nonfallthru_and_redirect): Update.
3333 (purge_dead_edges): Update.
3334 (rtl_flow_call_edges_add): Update.
3335 * cgraphunit.c (init_lowered_empty_function): Update.
3336 (cgraph_node::expand_thunk): Update.
3337 * gimple-pretty-print.c (dump_probability): Update.
3338 (dump_edge_probability): Update.
3339 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3340 * haifa-sched.c (sched_create_recovery_edges): Update.
3341 * hsa-gen.c (convert_switch_statements): Update.
3342 * ifcvt.c (dead_or_predicable): Update.
3343 * ipa-inline-transform.c (inline_transform): Update.
3344 * ipa-split.c (split_function): Update.
3345 * ipa-utils.c (ipa_merge_profiles): Update.
3346 * loop-doloop.c (add_test): Update.
3347 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3348 * lto-streamer-in.c (input_cfg): Update.
3349 (input_function): Update.
3350 * lto-streamer-out.c (output_cfg): Update.
3351 * modulo-sched.c (sms_schedule): Update.
3352 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3353 * predict.c (maybe_hot_edge_p): Update.
3354 (unlikely_executed_edge_p): Update.
3355 (probably_never_executed_edge_p): Update.
3356 (dump_prediction): Update.
3357 (drop_profile): Update.
3358 (propagate_unlikely_bbs_forward): Update.
3359 (determine_unlikely_bbs): Update.
3360 (force_edge_cold): Update.
3361 * profile.c (compute_branch_probabilities): Update.
3362 * reg-stack.c (better_edge): Update.
3363 * shrink-wrap.c (handle_simple_exit): Update.
3364 * tracer.c (better_p): Update.
3365 * trans-mem.c (expand_transaction): Update.
3366 (split_bb_make_tm_edge): Update.
3367 * tree-call-cdce.c: Update.
3368 * tree-cfg.c (gimple_find_sub_bbs): Update.
3369 (gimple_split_edge): Update.
3370 (gimple_duplicate_sese_region): Update.
3371 (gimple_duplicate_sese_tail): Update.
3372 (gimple_flow_call_edges_add): Update.
3373 (insert_cond_bb): Update.
3374 (execute_fixup_cfg): Update.
3375 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3376 * tree-complex.c (expand_complex_div_wide): Update.
3377 * tree-eh.c (lower_resx): Update.
3378 (unsplit_eh): Update.
3379 (cleanup_empty_eh_move_lp): Update.
3380 * tree-inline.c (copy_edges_for_bb): Update.
3381 (freqs_to_counts): Update.
3382 (copy_cfg_body): Update.
3383 * tree-ssa-dce.c (remove_dead_stmt): Update.
3384 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3385 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3386 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3387 (unloop_loops): Update.
3388 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3389 * tree-ssa-loop-split.c (connect_loops): Update.
3390 (split_loop): Update.
3391 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3392 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3393 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3394 * tree-ssa-reassoc.c (branch_fixup): Update.
3395 * tree-ssa-tail-merge.c (replace_block_by): Update.
3396 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3397 (compute_path_counts): Update.
3398 (update_profile): Update.
3399 (recompute_probabilities): Update.
3400 (update_joiner_offpath_counts): Update.
3401 (estimated_freqs_path): Update.
3402 (freqs_to_counts_path): Update.
3403 (clear_counts_path): Update.
3404 (ssa_fix_duplicate_block_edges): Update.
3405 (duplicate_thread_path): Update.
3406 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3407 (case_bit_test_cmp): Update.
3408 (collect_switch_conv_info): Update.
3409 (gen_inbound_check): Update.
3410 (do_jump_if_equal): Update.
3411 (emit_cmp_and_jump_insns): Update.
3412 * tree-tailcall.c (decrease_profile): Update.
3413 (eliminate_tail_call): Update.
3414 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3415 (vect_do_peeling): Update.
3416 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3417 * ubsan.c (ubsan_expand_null_ifn): Update.
3418 (ubsan_expand_ptr_ifn): Update.
3419 * value-prof.c (gimple_divmod_fixed_value): Update.
3420 (gimple_mod_pow2): Update.
3421 (gimple_mod_subtract): Update.
3422 (gimple_ic): Update.
3423 (gimple_stringop_fixed_value): Update.
3425 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3428 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3430 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3432 PR rtl-optimization/82395
3433 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3434 based on non_spilled_static_chain_regno_p.
3436 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3438 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3439 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3440 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3442 2017-10-19 Martin Sebor <msebor@redhat.com>
3444 PR tree-optimization/82596
3445 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3447 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3449 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3450 (get_mem_refs_of_builtin_call): Likewise.
3451 * builtins.c (expand_builtin_apply): Adjust call to
3452 allocate_dynamic_stack_space.
3453 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3454 the third argument to allocate_dynamic_stack_space, otherwise -1.
3455 (expand_builtin): Deal with all alloca variants.
3456 (is_inexpensive_builtin): Likewise.
3457 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3458 * calls.c (special_function_p): Deal with all alloca variants.
3459 (initialize_argument_information): Adjust call to
3460 allocate_dynamic_stack_space.
3461 (expand_call): Likewise.
3462 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3463 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3464 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3465 use it for the stack usage computation.
3466 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3467 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3468 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3469 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3470 (in_loop_p): Remove first argument and useless check.
3471 (pass_walloca::execute): Remove useless test and adjust call to above.
3472 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3473 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3474 (gimplify_call_expr): Deal with all alloca variants.
3475 * hsa-gen.c (gen_hsa_alloca): Likewise.
3476 (gen_hsa_insns_for_call): Likewise.
3477 * ipa-pure-const.c (special_builtin_state): Likewise.
3478 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3479 * tree-object-size.c (alloc_object_size): Likewise.
3480 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3481 (call_may_clobber_ref_p_1): Likewise.
3482 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3483 (ccp_fold_stmt): Likewise.
3484 (optimize_stack_restore): Likewise.
3485 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3486 (mark_all_reaching_defs_necessary_1): Likewise.
3487 (propagate_necessity): Likewise.
3488 (eliminate_unnecessary_stmts): Likewise.
3489 * tree.c (build_common_builtin_nodes): Build
3490 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3491 (build_alloca_call_expr): New function.
3492 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3493 (CASE_BUILT_IN_ALLOCA): Likewise.
3494 (build_alloca_call_expr): Declare.
3495 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3497 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3500 * dwarf2out.c (new_die_raw): New static inline function.
3501 (new_die): Use it to create the DIE.
3502 (add_AT_external_die_ref): Likewise.
3503 (clone_die): Likewise.
3504 (clone_as_declaration): Likewise.
3505 (dwarf2out_vms_debug_main_pointer): Likewise.
3506 (base_type_die): Likewise. Remove early return for corner cases.
3507 Do not call add_pubtype on the DIE here.
3508 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3509 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3510 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3511 native order exists for base types, attach the DIE manually and call
3512 add_pubtype on it. Do not equate a reverse order DIE to the type.
3514 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3516 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3517 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3519 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3521 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3523 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3524 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3525 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3526 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3527 * opts.c (sanitizer_opts): Add builtin.
3528 * ubsan.c (instrument_builtin): New function.
3529 (pass_ubsan::execute): Call it.
3530 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3531 * doc/invoke.texi: Document -fsanitize=builtin.
3533 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3534 builtins, store max (log2 (align), 0) into uchar field instead of
3535 align into uptr field.
3536 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3537 store uchar 0 field instead of uptr 0 field.
3538 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3539 instead of passing one address of struct with 2 locations pass
3540 two addresses of structs with 1 location each.
3541 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3542 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3543 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3544 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3545 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3546 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3547 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3548 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3550 2017-10-19 Martin Liska <mliska@suse.cz>
3553 * file-find.c (remove_prefix): Remove.
3554 * file-find.h (remove_prefix): Likewise.
3555 * gcc-ar.c: Remove smartness of lookup.
3557 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3559 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3560 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3561 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3563 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3566 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3567 (setcc + and to xor + setcc): New peephole2.
3569 2017-10-19 Tom de Vries <tom@codesourcery.com>
3571 * doc/sourcebuild.texi (Test Directives, Variants of
3572 dg-require-support): Add dg-require-stack-size.
3574 2017-10-19 Martin Liska <mliska@suse.cz>
3577 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3578 that have a large alignment.
3579 (gimplify_target_expr): Likewise.
3581 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3583 PR rtl-optimization/82602
3584 * ira.c (rtx_moveable_p): Return false for volatile asm.
3586 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3589 * config/i386/i386-modes.def (CCGZ): New CC mode.
3590 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3591 * config/i386/predicates.md (ix86_comparison_operator):
3593 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3594 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3595 with double-word subtraction.
3596 (put_condition_code): Handle CCGZmode.
3598 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3600 * wide-int.cc (debug (const wide_int &)): New.
3601 (debug (const wide_int *)): New.
3602 (debug (const widest_int &)): New.
3603 (debug (const widest_int *)): New.
3605 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3608 * lra-constraints.c (curr_insn_transform): Use non-input operand
3609 instead of output one for matched reload.
3611 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3613 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3614 (tree-ssa-loop-ivopts.h): New header file.
3615 (struct builtin_info): New fields.
3616 (classify_builtin_1): Compute and record base and offset parts for
3617 memset builtin partition by calling strip_offset.
3618 (offset_cmp, fuse_memset_builtins): New functions.
3619 (finalize_partitions): Fuse adjacent memset partitions by calling
3621 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3622 Expose the interface.
3623 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3625 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3627 PR tree-optimization/82574
3628 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3629 that data reference must be executed exactly once per iteration
3630 against the outermost loop in nest.
3631 (classify_partition): Update call to above function.
3633 2017-10-18 Richard Biener <rguenther@suse.de>
3635 PR tree-optimization/82591
3636 * graphite.c (graphite_transform_loops): Move code gen message
3638 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3639 Here. Handle scop_to_isl_ast failing.
3640 (scop_to_isl_ast): Limit the number of ISL operations.
3642 2017-10-18 Richard Biener <rguenther@suse.de>
3644 * graphite-isl-ast-to-gimple.c
3645 (translate_isl_ast_to_gimple::set_rename): Simplify.
3646 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3647 (graphite_copy_stmts_from_block): ... here.
3648 (copy_bb_and_scalar_dependences): Simplify.
3649 (add_parameters_to_ivs_params): Canonicalize.
3650 (generate_entry_out_of_ssa_copies): Simplify.
3651 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3652 by passing in ISL dimension.
3653 (parameter_index_in_region_1): Rename to ...
3654 (parameter_index_in_region): ... this.
3655 (extract_affine): Adjust assert, pass down parameter index.
3656 (add_param_constraints): Use range-info when available.
3657 (build_scop_context): Adjust.
3658 * sese.c (new_sese_info): Adjust.
3659 (free_sese_info): Likewise.
3660 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
3661 Remove unused typedefs.
3662 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
3664 2017-10-18 Martin Liska <mliska@suse.cz>
3666 * combine.c (simplify_compare_const): Add gcc_fallthrough.
3668 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3670 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3671 (s390_sched_init): Do not reset s390_sched_state if we entered the
3672 current basic block via a fallthru edge and all others are unlikely.
3674 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3676 * config/s390/s390.c (NUM_SIDES): New variable.
3677 (LONGRUNNING_THRESHOLD): New variable.
3678 (LATENCY_FACTOR): New variable.
3679 (s390_sched_score): Decrease score for long-running instructions on
3681 (s390_sched_variable_issue): Perform bookkeeping for long-running
3684 2017-10-18 Richard Biener <rguenther@suse.de>
3686 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3687 Simplify with removal of the parameter rename map.
3688 (set_rename): Likewise.
3689 (should_copy_to_new_region): Likewise.
3690 (graphite_copy_stmts_from_block): Likewise.
3691 (copy_bb_and_scalar_dependences): Remove initialization of
3692 unused copied_bb_map.
3694 (copy_internal_parameters): Likewise.
3695 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3696 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3697 Use INTEGRAL_TYPE_P.
3698 (parameter_index_in_region_1): Rename to ...
3699 (assign_parameter_index_in_region): ... this. Assert we have
3700 a parameter we handle.
3701 (scan_tree_for_params): Adjust.
3702 * sese.h (parameter_rename_map_t): Remove.
3703 (struct sese_info_t): Remove unused parameter_rename_map and
3704 copied_bb_map members.
3705 * sese.c (new_sese_info): Adjust.
3706 (free_sese_info): Likewise.
3708 2017-10-18 Martin Liska <mliska@suse.cz>
3711 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3712 on an abnormal edge.
3714 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3716 * doc/invoke.texi (ffunction-sections and fdata-sections):
3719 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3721 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3722 the use statement can throw internally.
3724 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3726 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3727 any RTX present on the RHS of a SET.
3728 * compare-elim.c (try_eliminate_compare): Restore comment.
3730 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3732 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3733 may be also called on tcc_type nodes.
3734 * langhooks.c (lhd_tree_size): Likewise.
3736 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3738 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3739 format_warning_at_substring.
3740 (maybe_warn): Convert source_range * param to a location_t. Pass
3741 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3742 (format_directive): Remove code to extract source_ranges and
3743 source_range * in favor of just a location_t.
3744 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3746 * substring-locations.c (format_warning_va): Convert
3747 source_range * param to a location_t.
3748 (format_warning_at_substring): Likewise.
3749 * substring-locations.h (format_warning_va): Likewise.
3750 (format_warning_at_substring): Likewise.
3752 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3754 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3756 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3757 and scatter/gather ops.
3759 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3760 vec_gather_load and vec_scatter_store.
3761 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3762 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3763 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3764 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3765 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3766 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3768 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3770 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3771 Detect FTST instruction and handle its register pops. Only pop
3772 second operand if can_pop_second_op is true.
3773 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3774 set can_pop_second_op to false in the compare_for_stack_reg call.
3776 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3777 output_fp_compare for stack register operands.
3778 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3779 instructions here. Do not emit stack register pops here. Assert
3780 that FCOMPP pops next to top stack register. Rewrite function.
3782 2017-10-17 Nathan Sidwell <nathan@acm.org>
3785 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3786 use DECL_ASSEMBLER_NAME_RAW.
3789 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3792 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3793 Wilco Dijkstra <wilco.dijkstra@arm.com>
3795 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3796 converstion to Pmode from the buf_addr.
3798 2017-10-17 Richard Biener <rguenther@suse.de>
3800 * graphite-dependences.c (scop_get_reads_and_writes): Change
3801 output parameters to references.
3803 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3805 PR 71026/tree-optimization
3806 * fold-const.c (distribute_real_division): Removed.
3807 (fold_binary_loc): Remove calls to distribute_real_divison.
3809 2017-10-17 Richard Biener <rguenther@suse.de>
3811 * graphite-scop-detection.c
3812 (scop_detection::stmt_has_simple_data_refs_p): Always use
3813 the full nest as region.
3814 (try_generate_gimple_bb): Likewise.
3815 * sese.c (scalar_evolution_in_region): Simplify now that
3816 SCEV can handle instantiation in regions.
3817 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3818 in the non-loop part of a function if requested.
3820 2017-10-17 Richard Biener <rguenther@suse.de>
3822 PR tree-optimization/82563
3823 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3825 (graphite_regenerate_ast_isl): Call it.
3826 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3828 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3830 PR tree-optimization/82549
3831 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3832 Formatting fixes. Instead of calling make_bit_field_ref with negative
3835 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3837 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3838 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3839 _mm_maskz_reduce_ss): New.
3840 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3841 __builtin_ia32_reducess_mask): Ditto..
3842 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3843 * config/i386/sse.md (reduces<mode>): Renamed to ...
3844 (reduces<mode><mask_scalar_name>): ... this.
3845 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3847 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3848 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3850 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3852 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3853 * selftest-run-tests.c (selftest::run_tests): Call
3854 selftest::unique_ptr_tests_cc_tests.
3855 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3856 * unique-ptr-tests.cc: New file.
3858 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3861 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3863 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3864 (make_hard_regno_born, make_hard_regno_dead): Update
3865 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3867 2017-10-16 Jeff Law <law@redhat.com>
3869 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3871 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3873 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3875 2017-10-16 Olivier Hainque <hainque@adacore.com>
3877 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3878 with_cpu if we were configured for an e500v2 target cpu name.
3880 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3882 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3883 * doc/invoke.texi: Document +nodsp as a valid extension for
3886 2017-10-16 Martin Liska <mliska@suse.cz>
3888 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3889 (test_set_range): Likewise.
3890 (test_range_functions): Rename to ...
3891 (test_bit_in_range): ... this.
3892 (sbitmap_c_tests): Add new test.
3894 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3896 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3898 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3899 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3901 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3903 * config/aarch64/aarch64-builtins.c
3904 (aarch64_types_quadopu_lane_qualifiers): New.
3905 (TYPES_QUADOPU_LANE): New.
3906 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3907 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3908 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3909 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3910 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3911 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3912 (Vdottype, DOTPROD): New.
3913 (sur): Add SDOT and UDOT.
3915 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3917 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3918 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3919 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3921 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3922 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3923 Enable TARGET_DOTPROD.
3924 (cortex-a75.cortex-a55): Likewise.
3925 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3927 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3929 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3930 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3932 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3934 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3935 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3936 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3937 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3938 * config/arm/types.md (neon_dot, neon_dot_q): New.
3939 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3941 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3943 * config/arm/arm.h (TARGET_DOTPROD): New.
3944 * config/arm/arm.c (arm_arch_dotprod): New.
3945 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3946 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3947 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3948 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3949 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3950 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3951 * doc/invoke.texi (armv8.2-a): Document dotprod
3953 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3955 * i386.c (ix86_vec_cost): New function.
3956 (ix86_rtx_costs): Handle vector operations better.
3957 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3958 * x86-tune-costs.h: Add new costs to all tables.
3960 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3962 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3964 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3965 divsd, sqrtss and sqrtsd
3966 * x86-tune-costs.h: Add new entries to all costs.
3967 (znver1_cost): Fix to match real instruction latencies.
3969 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3970 Michael Collison <michael.collison@arm.com>
3972 * compare-elim.c: Include emit-rtl.h.
3973 (can_merge_compare_into_arith): New function.
3974 (try_validate_parallel): Likewise.
3975 (try_merge_compare): Likewise.
3976 (try_eliminate_compare): Call the above when no previous clobber
3978 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
3981 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3985 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
3986 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
3990 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
3991 to be any operand_equal_p operands. For & (B - 1) require
3992 B to be power of 2. Recognize
3993 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
3995 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3998 * optabs.c (expand_memory_blockage): Fix call of
3999 targetm.have_memory_blockage.
4001 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4004 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4005 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4006 objects to extra_objs instead of overwriting it.
4008 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4010 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4011 Use any_fp_register_operand as operand[3] predicate. Simplify
4012 equality test for operands[2] and operands[4] memory location.
4013 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4014 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4015 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4016 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4017 any_fp_register_operand as operand[1] predicate. Simplify
4018 equality test for operands[0] and operands[3] memory location.
4019 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4020 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4021 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4023 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4025 * target-insns.def: Add memory_blockage.
4026 * optabs.c (expand_memory_blockage): New function.
4027 (expand_asm_memory_barrier): Rename ...
4028 (expand_asm_memory_blockage): ... to this.
4029 (expand_mem_thread_fence): Call expand_memory_blockage
4030 instead of expand_asm_memory_barrier.
4031 (expand_mem_singnal_fence): Ditto.
4032 (expand_atomic_load): Ditto.
4033 (expand_atomic_store): Ditto.
4034 * doc/md.texi (Standard Pattern Names For Generation):
4035 Document memory_blockage instruction pattern.
4037 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4039 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4040 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4041 * config/rl78/rl78.md: New define_expand "adddi3".
4043 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4045 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4048 2017-10-13 Jeff Law <law@redhat.com>
4050 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4052 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4055 * internal-fn.c (expand_mul_overflow): If both operands have
4056 the same highpart of -1 or 0 and the topmost bit of lowpart
4057 is different, overflow is if res <= 0 rather than res < 0.
4059 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4061 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4062 TARGET_P9_VECTOR code for unaligned_load case.
4064 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4066 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4068 2017-10-13 Nathan Sidwell <nathan@acm.org>
4070 * tree-core.h (tree_contains_struct): Make bool.
4071 * tree.c (tree_contains_struct): Likewise.
4072 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4073 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4074 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4075 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4076 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4078 2017-10-13 Richard Biener <rguenther@suse.de>
4080 * graphite-isl-ast-to-gimple.c
4081 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4082 parameters and dominance check.
4083 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4084 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4085 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4086 Do not update SSA form here or do intermediate IL verification.
4087 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4088 (graphite_initialize): Remove check on the number of loops in
4089 the function and inline into graphite_transform_loops.
4090 (graphite_finalize): Inline into graphite_transform_loops.
4091 (graphite_transform_loops): Perform SSA update and IL verification
4093 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4095 2017-10-13 Richard Biener <rguenther@suse.de>
4097 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4098 graphite_expression_type_precision): Avoid global constructor
4100 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4101 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4102 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4103 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4104 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4105 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4107 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4110 * config/i386/i386.h (ix86_red_zone_size): New.
4111 * config/i386/i386.md (push peephole2s): Replace
4112 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4114 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4115 Alan Hayward <alan.hayward@arm.com>
4116 David Sherwood <david.sherwood@arm.com>
4118 * combine.c (can_change_dest_mode): Reject changes in
4119 REGMODE_NATURAL_SIZE.
4121 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4122 Alan Hayward <alan.hayward@arm.com>
4123 David Sherwood <david.sherwood@arm.com>
4125 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4126 (expand_debug_source_expr): Likewise.
4127 * combine.c (combine_simplify_rtx): Likewise.
4128 * cse.c (fold_rtx): Likewise.
4129 * fwprop.c (canonicalize_address): Likewise.
4130 * targhooks.c (default_shift_truncation_mask): Likewise.
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 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4137 (widened_mode): Likewise.
4138 (expand_unop): Likewise.
4139 * ree.c (transform_ifelse): Likewise.
4140 (merge_def_and_ext): Likewise.
4141 (combine_reaching_defs): Likewise.
4142 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4144 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4145 Alan Hayward <alan.hayward@arm.com>
4146 David Sherwood <david.sherwood@arm.com>
4148 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4149 * combine.c (gen_lowpart_for_combine): Likewise.
4150 * dwarf2out.c (rtl_for_decl_location): Likewise.
4151 * final.c (alter_subreg): Likewise.
4152 * rtlhooks.c (gen_lowpart_general): Likewise.
4153 (gen_lowpart_if_possible): Likewise.
4155 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4156 Alan Hayward <alan.hayward@arm.com>
4157 David Sherwood <david.sherwood@arm.com>
4159 * calls.c (expand_call): Use subreg_lowpart_offset.
4160 * cse.c (cse_insn): Likewise.
4161 * regcprop.c (copy_value): Likewise.
4162 (copyprop_hardreg_forward_1): Likewise.
4164 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4167 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
4168 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
4169 =Q constraints to +Q and into insn condition add check
4170 that operands[0] and operands[1] are equal.
4171 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
4172 =Q constraints to +Q and into insn condition add check
4173 that operands[0] is equal to either operands[1] or operands[2].
4176 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
4177 instead of handling MINUS_EXPR twice (once for each argument),
4178 canonicalize operand order and handle just once, use rtype where
4179 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
4182 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
4183 any values of __C while still being pattern recognizable as a simple
4186 2017-10-13 Richard Biener <rguenther@suse.de>
4188 PR tree-optimization/82451
4190 2017-10-02 Richard Biener <rguenther@suse.de>
4192 PR tree-optimization/82355
4193 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4194 a mapping for the enclosing loop but avoid generating one for
4196 (copy_bb_and_scalar_dependences): Remove premature codegen
4197 error on PHIs in blocks duplicated into multiple places.
4198 * graphite-scop-detection.c
4199 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4200 in the region use it as loop and nest to analyze the DR in.
4201 (try_generate_gimple_bb): Likewise.
4202 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4203 (add_loop_constraints): For blocks in a loop not in the region
4204 create a dimension with a single iteration.
4205 * sese.h (gbb_loop_at_index): Remove assert.
4207 * cfgloop.c (loop_preheader_edge): For the loop tree root
4208 return the single successor of the entry block.
4209 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4210 Reset the SCEV hashtable and niters.
4211 * graphite-scop-detection.c
4212 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
4213 assert that we only have POLYNOMIAL_CHREC that vary in loops
4214 contained in the region.
4215 (scop_detection::graphite_can_represent_expr): Adjust.
4216 (scop_detection::stmt_has_simple_data_refs_p): For loops
4217 not in the region set loop to NULL. The nest is now the
4218 entry edge to the region.
4219 (try_generate_gimple_bb): Likewise.
4220 * sese.c (scalar_evolution_in_region): Adjust for
4221 instantiate_scev change.
4222 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4223 Make nest parameter the edge into the region.
4224 (create_data_ref): Likewise.
4225 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4226 entry edge into a region and adjust instantiate_scev calls.
4227 (create_data_ref): Likewise.
4228 (graphite_find_data_references_in_stmt): Likewise.
4229 (find_data_references_in_stmt): Pass the loop preheader edge
4230 from the nest argument.
4231 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4232 parameter the edge into the region.
4233 (instantiate_parameters): Use the loop preheader edge as entry.
4234 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4236 (get_instantiated_value_entry): Make instantiate_below parameter
4237 the edge into the region.
4238 (instantiate_scev_name): Likewise. Adjust dominance checks,
4239 when we cannot use loop-based instantiation instantiate by
4240 walking use-def chains.
4241 (instantiate_scev_poly): Adjust.
4242 (instantiate_scev_binary): Likewise.
4243 (instantiate_scev_convert): Likewise.
4244 (instantiate_scev_not): Likewise.
4245 (instantiate_array_ref): Remove.
4246 (instantiate_scev_3): Likewise.
4247 (instantiate_scev_2): Likewise.
4248 (instantiate_scev_1): Likewise.
4249 (instantiate_scev_r): Do not blindly handle N-operand trees.
4250 Do not instantiate array-refs. Handle all constants and invariants.
4251 (instantiate_scev): Make instantiate_below parameter
4252 the edge into the region.
4253 (resolve_mixers): Use the loop preheader edge for the region
4254 parameter to instantiate_scev_r.
4255 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4257 2017-10-13 Richard Biener <rguenther@suse.de>
4259 PR tree-optimization/82525
4260 * graphite-isl-ast-to-gimple.c
4261 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4263 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4264 Fail code generation when we cannot represent the isl integer.
4265 (binary_op_to_tree): Elide modulo operations that are no-ops
4266 in the type we code generate. Remove now superfluous code
4269 2017-10-13 Richard Biener <rguenther@suse.de>
4271 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4272 (scop_detection::harmful_loop_in_region): Remove premature
4273 IV type restriction.
4274 (scop_detection::graphite_can_represent_scev): We can handle
4275 pointer IVs just fine.
4277 2017-10-13 Alan Modra <amodra@gmail.com>
4279 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4280 "Clobbers and Scratch Registers". Add paragraph on
4281 alternative to clobbers for scratch registers and OpenBLAS
4284 2017-10-13 Alan Modra <amodra@gmail.com>
4286 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4287 example of a memory input for a string of known length. Move
4288 commentary out of table. Add a number of new examples
4289 covering array memory inputs.
4291 2017-10-12 Martin Liska <mliska@suse.cz>
4293 PR tree-optimization/82493
4294 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4295 (test_range_functions): New function.
4296 (sbitmap_c_tests): Likewise.
4297 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4298 * selftest.h (sbitmap_c_tests): New function.
4300 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4302 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4304 * config/rs6000/amo.h: Fix spacing issue.
4306 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4309 * config/i386/i386.md (*ashl<mode>3_mask_1,
4310 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4311 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4314 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4316 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4317 (profile_probability): Set max_probability
4318 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4320 * profile-count.c (profile_probability::differs_from_p): Do not
4321 rely on max_probaiblity == 10000
4323 2017-10-12 Jeff Law <law@redhat.com>
4325 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4328 2017-10-12 Martin Sebor <msebor@redhat.com>
4332 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4333 (handle_alias_pairs): Call it.
4334 * common.opt (-Wattribute-alias): New option.
4335 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4336 * doc/invoke.texi (-Wattribute-alias): Document.
4338 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4341 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4343 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4345 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4346 (make_hard_regno_born, make_hard_regno_dead): Update
4347 bb_killed_pseudos and bb_gen_pseudos.
4349 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4351 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4353 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4355 * config/alpha/alpha.c (alpha_split_conditional_move):
4356 Use std::swap instead of manually swapping.
4357 (alpha_stdarg_optimize_hook): Ditto.
4358 (alpha_canonicalize_comparison): Ditto.
4360 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4362 * tree-loop-distribution.c (struct builtin_info): New struct.
4363 (struct partition): Refactor fields into struct builtin_info.
4364 (partition_free): Free struct builtin_info.
4365 (build_size_arg_loc, build_addr_arg_loc): Delete.
4366 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4367 information from struct builtin_info.
4368 (find_single_drs): New function refactored from classify_partition.
4369 Also moved builtin validity checks to this function.
4370 (compute_access_range, alloc_builtin): New functions.
4371 (classify_builtin_st, classify_builtin_ldst): New functions.
4372 (classify_partition): Refactor code into functions find_single_drs,
4373 classify_builtin_st and classify_builtin_ldst.
4374 (distribute_loop): Don't do runtime alias check when distributing
4376 (find_seed_stmts_for_distribution): New function.
4377 (pass_loop_distribution::execute): Refactor code finding seed
4378 stmts into above function. Support distribution for the innermost
4379 two-level loop nest. Adjust dump information.
4381 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4383 * tree-loop-distribution.c: Adjust the general comment.
4384 (NUM_PARTITION_THRESHOLD): New macro.
4385 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4386 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4387 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4388 in call to build_partition_graph.
4389 (finalize_partitions): New parameter. Make loop distribution more
4390 conservative by fusing more partitions.
4391 (distribute_loop): Don't do runtime alias check in case of loop nest
4393 (find_seed_stmts_for_distribution): New function.
4394 (prepare_perfect_loop_nest): New function.
4395 (pass_loop_distribution::execute): Refactor code finding seed stmts
4396 and loop nest into above functions. Support loop nest distribution.
4397 Adjust dump information accordingly.
4399 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4401 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4402 and set PTYPE_SEQUENTIAL for merged partition.
4404 2017-10-12 Richard Biener <rguenther@suse.de>
4406 PR tree-optimization/69728
4408 2017-09-19 Richard Biener <rguenther@suse.de>
4410 PR tree-optimization/69728
4411 * graphite-sese-to-poly.c (schedule_error): New global.
4412 (add_loop_schedule): Handle empty domain by failing the
4414 (build_original_schedule): Handle schedule_error.
4416 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4417 domain by returning an unchanged schedule.
4419 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4421 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4422 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4424 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4426 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4429 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4432 * expr.c (store_field): Don't optimize away bitsize == 0 store
4433 from CALL_EXPR with addressable return type.
4435 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4437 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4438 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4439 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4440 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4441 TARGET_ISEL instead of TARGET_ISEL<sel>.
4443 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4445 * config/rs6000/rs6000.c
4446 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4448 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4450 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4452 (reg_or_cint_operand): Fix comment.
4453 (reg_or_zero_operand): New predicate.
4454 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4455 * config/rs6000/rs6000.c (output_isel): Delete.
4456 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4457 instead of reg_or_cint_operand. Output instruction directly (not via
4459 (isel_unsigned_<mode>): Ditto.
4460 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4461 gpc_reg_operand. Add an instruction alternative for this. Output
4462 instruction directly.
4463 (*isel_reversed_unsigned_<mode>): Ditto.
4465 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4467 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4468 (TARGET_CANONICALIZE_COMPARISON): Define.
4470 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4473 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4474 Check whether the dest is REG before adding REG_EQUIV note.
4476 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4479 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4481 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4482 (make_hard_regno_born, make_hard_regno_dead): Update
4483 bb_killed_pseudos and bb_gen_pseudos.
4485 2017-10-11 Nathan Sidwell <nathan@acm.org>
4487 * incpath.h (enum incpath_kind): Name enum, prefix values.
4488 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4489 * incpath.c (heads, tails): Use INC_MAX.
4490 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4491 (merge_include_chains, split_quote_chain,
4492 register_include_chains): Update incpath_kind names.
4493 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4494 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4496 (add_framework_path, darwin_register_objc_includes): Likewise.
4497 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4499 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4501 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4502 Do not use float_operator operator predicate.
4503 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4504 * config/i386/predicates.md (float_operator): Remove predicate.
4506 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4508 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4509 (*jccxf_i387): Ditto.
4510 (*jcc<mode>_i387): Ditto.
4511 (*jccu<mode>_i387): Ditto.
4512 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4513 (*jcc_*_i387 splitters): Remove.
4514 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4515 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4516 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4519 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4521 * profile-count.h (slow_safe_scale_64bit): New function.
4522 (safe_scale_64bit): New inline.
4523 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4524 * profile-count.c: Include wide-int.h
4525 (slow_safe_scale_64bit): New.
4527 2017-10-11 Nathan Sidwell <nathan@acm.org>
4529 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4530 HAS_DECL_ASSEMBLER_NAME_P.
4531 * gimple-expr.c (gimple_decl_printable_name: Check
4532 HAS_DECL_ASSEMBLER_NAME_P too.
4533 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4534 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4535 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4536 * passes.c (rest_of_decl_compilation): Check
4537 HAS_DECL_ASSEMBLER_NAME_P too.
4538 * recog.c (verify_changes): Likewise.
4539 * tree-pretty-print.c (dump_decl_name): Likewise.
4540 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4542 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4543 (SET_DECL_ASSEMBLER_NAME): Use it.
4544 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4545 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4546 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4548 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4550 * config.gcc (i386, x86_64): Add extra objects.
4551 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4552 (ix86_min_insn_size): Declare.
4553 (ix86_issue_rate): Declare.
4554 (ix86_adjust_cost): Declare.
4555 (ia32_multipass_dfa_lookahead): Declare.
4556 (ix86_macro_fusion_p): Declare.
4557 (ix86_macro_fusion_pair_p): Declare.
4558 (ix86_bd_has_dispatch): Declare.
4559 (ix86_bd_do_dispatch): Declare.
4560 (ix86_core2i7_init_hooks): Declare.
4561 (ix86_atom_sched_reorder): Declare.
4562 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4563 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4564 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4565 (rip_relative_addr_p): Rename to ...
4566 (ix86_rip_relative_addr_p): ... this one; export.
4567 (memory_address_length): Update.
4568 (ix86_issue_rate): Move to x86-tune-sched.c.
4569 (ix86_flags_dependent): Move to x86-tune-sched.c.
4570 (ix86_agi_dependent): Move to x86-tune-sched.c.
4571 (exact_dependency_1): Move to x86-tune-sched.c.
4572 (exact_store_load_dependency): Move to x86-tune-sched.c.
4573 (ix86_adjust_cost): Move to x86-tune-sched.c.
4574 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4575 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4576 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4577 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4578 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4579 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4580 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4581 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4582 (min_insn_size): Rename to ...
4583 (ix86_min_insn_size): ... this one; export.
4584 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4585 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4586 (core2i7_first_cycle_multipass_backtrack): Move to
4587 x86-tune-sched-core.c.
4588 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4589 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4590 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4591 (ix86_avoid_jump_mispredicts): Update.
4592 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4593 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4594 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4595 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4596 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4597 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4598 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4599 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4600 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4601 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4602 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4603 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4604 (BIG): Move to ix86-tune-sched-bd.c.
4605 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4606 (enum insn_path): Move to ix86-tune-sched-bd.c.
4607 (get_mem_group): Move to ix86-tune-sched-bd.c.
4608 (is_cmp): Move to ix86-tune-sched-bd.c.
4609 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4610 (is_branch): Move to ix86-tune-sched-bd.c.
4611 (is_prefetch): Move to ix86-tune-sched-bd.c.
4612 (init_window): Move to ix86-tune-sched-bd.c.
4613 (allocate_window): Move to ix86-tune-sched-bd.c.
4614 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4615 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4616 (process_end_window): Move to ix86-tune-sched-bd.c.
4617 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4618 (find_constant): Move to ix86-tune-sched-bd.c.
4619 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4620 (has_immediate): Move to ix86-tune-sched-bd.c.
4621 (get_insn_path): Move to ix86-tune-sched-bd.c.
4622 (get_insn_group): Move to ix86-tune-sched-bd.c.
4623 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4624 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4625 (add_insn_window): Move to ix86-tune-sched-bd.c.
4626 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4627 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4628 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4629 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4630 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4631 (do_dispatch): Move to ix86-tune-sched-bd.c.
4632 (has_dispatch): Move to ix86-tune-sched-bd.c.
4633 * config/i386/t-i386: Add new object files.
4634 * config/i386/x86-tune-costs.h: New file.
4635 * config/i386/x86-tune-sched-atom.c: New file.
4636 * config/i386/x86-tune-sched-bd.c: New file.
4637 * config/i386/x86-tune-sched-core.c: New file.
4638 * config/i386/x86-tune-sched.c: New file.
4640 2017-10-11 Liu Hao <lh_mouse@126.com>
4642 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4643 the generic version below instead.
4644 (should_colorize): Recognize Windows consoles as terminals
4646 * pretty-print.c [__MINGW32__] (write_all): New function.
4647 [__MINGW32__] (find_esc_head): Likewise.
4648 [__MINGW32__] (find_esc_terminator): Likewise.
4649 [__MINGW32__] (eat_esc_sequence): Likewise.
4650 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4652 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
4655 2017-10-11 Richard Biener <rguenther@suse.de>
4657 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
4658 Properly call analyze_scalar_evolution with the loop of the stmt.
4660 2017-10-11 Richard Biener <rguenther@suse.de>
4662 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
4663 * tree-core.h (tree_base): Add chrec_var union member.
4664 * tree.h (CHREC_VAR): Remove.
4665 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
4666 * tree-chrec.h (build_polynomial_chrec): Adjust.
4667 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
4668 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
4670 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4672 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4673 * match.pd: ... here.
4674 ((T) X == (T) Y): Relax condition.
4676 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4678 PR tree-optimization/82472
4679 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4681 (break_alias_scc_partitions): Update postorder number.
4683 2017-10-11 Martin Liska <mliska@suse.cz>
4686 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4688 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4690 2017-10-11 Martin Liska <mliska@suse.cz>
4692 * print-rtl.c (print_insn): Move declaration of idbuf
4693 to same scope as name.
4695 2017-10-11 Martin Liska <mliska@suse.cz>
4700 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4703 2017-10-11 Martin Liska <mliska@suse.cz>
4706 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4709 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4711 * config.gcc (default_gnu_indirect_function): Default to yes for
4712 arm*-*-linux* with glibc.
4714 2017-10-11 Richard Biener <rguenther@suse.de>
4716 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4717 default-defs and types we do not want to analyze.
4718 (interpret_loop_phi): Replace unreachable code with an assert.
4719 (compute_scalar_evolution_in_loop): Remove and inline ...
4720 (analyze_scalar_evolution_1): ... here, replacing condition with
4721 what makes the intent clearer. Remove handling of cases
4722 get_scalar_evolution now handles.
4724 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4726 PR rtl-optimization/81434
4727 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4728 comment for main loop. In sched_group_found if, also add checks for
4729 pass and min_cost_group.
4731 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4733 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4734 (rs6000_insn_cost): New function.
4735 * config/rs6000/rs6000.md (cost): New attribute.
4737 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4738 H.J. Lu <hongjiu.lu@intel.com>
4742 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4743 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4744 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4745 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4746 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4747 OPTION_MASK_ISA_MMX and some other bit set require both
4748 mmx and the ISAs without the mmx bit.
4749 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4750 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4751 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4752 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4753 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4754 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4755 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4756 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4757 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4758 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4759 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4760 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4761 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4762 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4763 Add OPTION_MASK_ISA_MMX.
4765 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4767 * config.gcc (armv7*-*-freebsd*): New target.
4768 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4770 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4772 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4773 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4776 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4778 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4779 a template parameter.
4780 (WIDE_INT_REF_FOR): Update accordingly.
4781 * tree.h (wi::int_traits <const_tree>): Delete.
4782 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4783 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4784 (wi::tree_to_wide_ref): New typedef.
4785 (wi::to_wide): New function.
4786 * calls.c (get_size_range): Use wi::to_wide when operating on
4788 * cgraph.c (cgraph_node::create_thunk): Likewise.
4789 * config/i386/i386.c (ix86_data_alignment): Likewise.
4790 (ix86_local_alignment): Likewise.
4791 * dbxout.c (stabstr_O): Likewise.
4792 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4793 * expr.c (const_vector_from_tree): Likewise.
4794 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4795 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4796 (fold_negate_expr_1, int_const_binop_1, const_binop)
4797 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4798 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4799 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4800 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4801 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4802 (fold_not_const, round_up_loc): Likewise.
4803 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4804 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4805 (alloca_call_type): Likewise.
4806 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4807 * godump.c (go_output_typedef): Likewise.
4808 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4809 * internal-fn.c (get_min_precision): Likewise.
4810 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4811 * ipa-polymorphic-call.c
4812 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4813 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4814 (ipa_modify_call_arguments): Likewise.
4815 * match.pd: Likewise.
4816 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4817 * print-tree.c (print_node_brief, print_node): Likewise.
4818 * stmt.c (expand_case): Likewise.
4819 * stor-layout.c (layout_type): Likewise.
4820 * tree-affine.c (tree_to_aff_combination): Likewise.
4821 * tree-cfg.c (group_case_labels_stmt): Likewise.
4822 * tree-data-ref.c (dr_analyze_indices): Likewise.
4823 (prune_runtime_alias_test_list): Likewise.
4824 * tree-dump.c (dequeue_and_dump): Likewise.
4825 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4826 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4827 * tree-pretty-print.c (dump_generic_node): Likewise.
4828 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4829 (simple_iv_with_niters): Likewise.
4830 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4831 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4832 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4833 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4834 (refine_value_range_using_guard, number_of_iterations_ne_max)
4835 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4836 (get_cst_init_from_scev, record_nonwrapping_iv)
4837 (scev_var_range_cant_overflow): Likewise.
4838 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4839 * tree-ssa-pre.c (compute_avail): Likewise.
4840 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4841 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4842 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4843 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4844 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4845 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4846 (dump_case_nodes, try_switch_expansion): Likewise.
4847 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4848 (vect_do_peeling): Likewise.
4849 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4850 * tree-vect-stmts.c (vectorizable_load): Likewise.
4851 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4852 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4853 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4854 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4855 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4856 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4857 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4858 (evrp_dom_walker::before_dom_children): Likewise.
4859 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4860 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4861 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4862 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4863 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4864 (get_range_pos_neg): Likewise.
4865 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4866 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4867 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4868 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4869 * config/avr/avr.c (avr_fold_builtin): Likewise.
4870 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4871 * config/msp430/msp430.c (msp430_attr): Likewise.
4872 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4873 * config/powerpcspe/powerpcspe-c.c
4874 (altivec_resolve_overloaded_builtin): Likewise.
4875 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4876 (rs6000_expand_ternop_builtin): Likewise.
4877 * config/rs6000/rs6000-c.c
4878 (altivec_resolve_overloaded_builtin): Likewise.
4879 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4880 (rs6000_expand_ternop_builtin): Likewise.
4881 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4883 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4885 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4886 when copying loop nest with only one inner loop.
4888 2017-10-10 Richard Biener <rguenther@suse.de>
4890 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4891 blocks if SCEV is active.
4892 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4894 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4895 (scev_initialize): Assert we are not yet initialized.
4897 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4899 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4900 inner loop's exit stmt by making it always exit the loop, otherwise
4901 we would generate an infinite empty loop.
4903 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4905 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4906 renaming variables in new preheader if it's deleted.
4908 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4910 * tree-loop-distribution.c (struct partition): Remove unused field
4911 loops of the structure.
4912 (partition_alloc, partition_free): Ditto.
4913 (build_rdg_partition_for_vertex): Ditto.
4915 2017-10-09 Jeff Law <law@redhat.com>
4917 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4918 return type to match prototype and documentation.
4920 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4922 * config/rs6000/rs6000.c (processor_costs): Move to ...
4923 * config/rs6000/rs6000.h: ... here.
4924 (rs6000_cost): Declare.
4926 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4928 * except.c (setjmp_fn): New global variable.
4929 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4930 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4931 if DONT_USE_BUILTIN_SETJMP is defined.
4933 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4935 * target.def (insn_cost): New hook.
4936 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4937 * doc/tm.texi: Regenerate.
4938 * rtlanal.c (insn_cost): Use the new hook.
4940 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4942 * combine.c (combine_validate_cost): Compute the new insn_cost,
4943 not just pattern_cost.
4944 (try_combine): Adjust comment.
4946 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4948 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4950 * combine.c (uid_insn_cost): Adjust comment.
4951 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4953 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4954 * dse.c (find_shift_sequence): Ditto.
4955 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4956 (bb_valid_for_noce_process_p): Use pattern_cost.
4957 * rtl.h (insn_rtx_cost): Delete.
4958 (pattern_cost): New prototype.
4959 (insn_cost): New prototype.
4960 * rtlanal.c (insn_rtx_cost): Rename to...
4961 (pattern_cost): ... this.
4964 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4966 * config/i386/i386.md (*jcc_2): Remove insn pattern.
4967 (*jcc<mode>_0_r_i387): Ditto.
4968 (*jccxf_r_i387): Ditto.
4969 (*jcc<mode>_r_i387): Ditto.
4970 (*jccu<mode>_r_i387): Ditto.
4971 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
4972 (*jcc): Rename from *jcc_1.
4974 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4976 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
4977 deferred rescans after the lvx/stvx recombination pre-pass.
4979 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
4981 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
4982 memory operation instruction support.
4983 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
4984 (rs6000-ibm-aix[789]*): Likewise.
4985 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
4986 Document new functions.
4988 2017-10-09 Richard Biener <rguenther@suse.de>
4990 PR tree-optimization/82397
4991 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
4992 equality only for semantically equal trees.
4994 2017-10-09 Richard Biener <rguenther@suse.de>
4996 PR tree-optimization/82449
4997 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
4998 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
4999 allow constant addresses.
5000 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5003 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5005 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5008 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5011 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5014 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5017 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5019 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5022 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5023 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5025 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5027 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5028 (WI_BINARY_PREDICATE_RESULT): Likewise.
5029 (wi::binary_traits::operator_result): New type.
5030 (wi::binary_traits::predicate_result): Likewise.
5031 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5032 (generic_wide_int::operator==, generic_wide_int::operator!=)
5033 (generic_wide_int::operator&, generic_wide_int::and_not)
5034 (generic_wide_int::operator|, generic_wide_int::or_not)
5035 (generic_wide_int::operator^, generic_wide_int::operator+
5036 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5038 (operator~, unary operator-, operator==, operator!=, operator&)
5039 (operator|, operator^, operator+, binary operator-, operator*): New
5041 * expr.c (get_inner_reference): Use wi::bit_and_not.
5042 * fold-const.c (fold_binary_loc): Likewise.
5043 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5044 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5045 (bit_value_binop): Likewise.
5046 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5047 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5048 (extract_range_from_binary_expr_1): Likewise.
5049 (masked_increment): Likewise.
5050 (simplify_bit_ops_using_ranges): Likewise.
5052 2017-10-09 Martin Jambor <mjambor@suse.cz>
5055 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5056 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5057 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5059 (hsa_fixup_mov_insn_type): New function.
5060 (hsa_op_with_type::get_in_type): Use it.
5061 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5062 immediates in an assert.
5063 (hsa_op_with_type::extend_int_to_32bit): New method.
5064 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5065 types. Convert to dest type if necessary.
5066 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5067 (reg_for_gimple_ssa): Pass false as min32int to
5068 hsa_type_for_scalar_tree_type.
5069 (gen_hsa_addr): Fixup type when creating addresable temporary.
5070 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5071 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5072 necessary. Call hsa_fixup_mov_insn_type.
5073 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5074 extend operands and convert to dest type if necessary.
5075 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5076 to dest type if necessary.
5077 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5078 if conversion nt necessary and size matches.
5079 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5080 to dest type if necessary.
5081 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5082 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5084 (gen_hsa_clrsb): Likewise.
5085 (gen_hsa_ffs): Likewise.
5086 (gen_hsa_divmod): Extend operands and convert to dest type if
5088 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5090 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5092 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5093 Remove empty default arguments. Use a brace block as output
5095 (conditional return): Ditto.
5097 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5098 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5100 (group_ending_nop): Ditto.
5101 (doloop_end): Ditto.
5102 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5103 (splitters for those): Ditto.
5105 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5107 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5108 a conditional jump (and the compare for it) so that pc_rtx is the
5110 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5111 for the deleted and renamed ctr<mode>_internal[234] patterns.
5112 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5113 Delete second conditional return pattern.
5114 (ctr<mode>_internal2): Delete this second bdnz pattern.
5115 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5116 (ctr<mode>_internal4): Delete this second bdz pattern.
5118 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5120 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5121 (always_initialized_rtx_for_ssa_name_p): New predicate.
5122 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5123 (finish_out_of_ssa): Free new field of SA.
5124 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5125 * tree-ssa-coalesce.c: Include tree-ssa.h.
5126 (get_parm_default_def_partitions): Remove extern keyword.
5127 (get_undefined_value_partitions): New function.
5128 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5129 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5131 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5133 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5135 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5137 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5139 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5140 for targets that preffer 128bit.
5142 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5144 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5146 2017-10-08 Olivier Hainque <hainque@adacore.com>
5148 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5149 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5151 (thumb_set_return_address): Likewise.
5153 2017-10-08 Olivier Hainque <hainque@adacore.com>
5155 * common/config/arm/arm-common.c (arm_except_unwind_info):
5156 Handle DWARF2_UNWIND_INFO.
5158 2017-10-07 Michael Collison <michael.collison@arm.com>
5160 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
5163 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
5165 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
5166 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
5167 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
5168 defined, force the creation of a new block for a dispatch label.
5170 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
5172 * invoke.texi (Wsuggest-attribute=cold): Document.
5173 * common.opt (Wsuggest-attribute=cold): New
5174 * ipa-pure-const.c (warn_function_cold): New function.
5175 * predict.c (compute_function_frequency): Use it.
5176 * predict.h (warn_function_cold): Declare.
5178 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
5180 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
5183 2017-10-06 Martin Liska <mliska@suse.cz>
5185 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
5186 keyword for member functions.
5187 (struct sanopt_tree_couple): New struct.
5188 (struct sanopt_tree_couple_hash): New function.
5189 (struct sanopt_ctx): Add new hash_map.
5190 (has_dominating_ubsan_ptr_check): New function.
5191 (record_ubsan_ptr_check_stmt): Likewise.
5192 (maybe_optimize_ubsan_ptr_ifn): Likewise.
5193 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
5194 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
5196 2017-10-06 Sudakshina Das <sudi.das@arm.com>
5199 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
5200 aarch64_simd_valid_immediate on CONST_VECTORs.
5201 (aarch64_reg_or_bic_imm): Likewise.
5203 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
5205 PR rtl-optimization/82396
5206 * haifa-sched.c (ready_sort_real): Disable qsort checking.
5208 2017-10-06 Sebastian Pop <sebpop@gmail.com>
5210 * graphite-dependences.c (scop_get_reads): Move code to...
5211 (scop_get_must_writes): Move code to...
5212 (scop_get_may_writes): Move code to...
5213 (scop_get_reads_and_writes): ... here.
5214 (scop_get_dependences): Call scop_get_reads_and_writes.
5216 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5218 PR tree-optimization/82434
5219 * fold-const.h (can_native_encode_type_p,
5220 can_native_encode_string_p): Remove.
5221 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5222 don't encode anything, just return what would be otherwise returned.
5223 (native_encode_fixed, native_encode_complex, native_encode_vector):
5225 (native_encode_string): Likewise. Inline by hand
5226 can_native_encode_string_p.
5227 (can_native_encode_type_p): Remove.
5228 (can_native_encode_string_p): Remove.
5229 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5230 STRING_CSTs using can_native_encode_string_p, test all
5231 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5232 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5233 argument from native_encode_expr.
5234 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5235 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5238 2017-10-06 Richard Biener <rguenther@suse.de>
5240 PR tree-optimization/82397
5241 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5242 operand_equal_p but rely on data_ref_compare_tree for detecting
5244 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5245 to match up with dr_group_sort_cmp.
5247 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5250 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5252 * config/s390/s390-builtin-types.def: Regenerate.
5254 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5257 * config/s390/s390-builtin-types.def: Regenerate.
5258 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5259 Change flag from B_VXE to B_VX.
5260 (s390_vec_min_dbl): Remove B_VXE flag.
5262 2017-10-06 Richard Biener <rguenther@suse.de>
5264 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5265 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5266 translate_isl_ast_to_gimple::is_valid_rename,
5267 translate_isl_ast_to_gimple::get_rename,
5268 translate_isl_ast_to_gimple::get_def_bb_for_const,
5269 translate_isl_ast_to_gimple::get_new_name,
5270 translate_isl_ast_to_gimple::collect_all_ssa_names,
5271 translate_isl_ast_to_gimple::copy_loop_phi_args,
5272 translate_isl_ast_to_gimple::collect_all_ssa_names,
5273 translate_isl_ast_to_gimple::copy_loop_phi_args,
5274 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5275 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5276 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5277 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5278 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5279 translate_isl_ast_to_gimple::copy_cond_phi_args,
5280 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5281 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5282 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5283 translate_isl_ast_to_gimple::rename_uses,
5284 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5285 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5286 (set_rename_for_each_def): Likewise.
5287 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5288 here. Handle rewriting SCEV analyzable uses here.
5289 (copy_bb_and_scalar_dependences): Generate code for PHI
5291 (graphite_regenerate_ast_isl): Adjust.
5292 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5293 (add_write, add_read): New functions.
5294 (build_cross_bb_scalars_def): Use it and simplify.
5295 (build_cross_bb_scalars_use): Likewise.
5296 (graphite_find_cross_bb_scalar_vars): Inline into...
5297 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5298 simulating out-of-SSA. Compute liveout and add dependencies.
5299 (build_scops): Force an empty entry block.
5300 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5302 (sese_build_liveouts): Declare.
5303 (sese_trivially_empty_bb_p): Likewise.
5304 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5305 compute liveout and debug_liveout.
5306 (sese_bad_liveouts_use): Remove.
5307 (sese_reset_debug_liveouts_bb): Likewise.
5308 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5309 (sese_build_liveouts): Build liveout and debug_liveout and store
5311 (new_sese_info): Adjust.
5312 (free_sese_info): Likewise.
5313 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5314 do not build liveout here.
5315 (move_sese_in_condition): Adjust region entry.
5316 (scev_analyzable_p): Match up with chrec_apply requirements.
5317 (sese_trivially_empty_bb_p): New.
5318 * tree-into-ssa.c (get_reaching_def): Properly support generating
5319 default-defs for incremental rewrite of anonymous names.
5321 2017-10-06 Richard Biener <rguenther@suse.de>
5323 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5324 precision do not perform modulo reduction.
5326 2017-10-06 Richard Biener <rguenther@suse.de>
5328 PR tree-optimization/82436
5329 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5330 conservatively choose the vectorization factor when checking
5331 whether we can perform the required load permutation.
5332 (vect_transform_slp_perm_load): Assert when we may not fail.
5334 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5336 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5337 message for incompatible -msdata=* and -mcall-* options.
5339 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5341 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5342 rate for post-reload scheduling.
5344 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5346 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5348 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5350 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5351 to improve monte carlo in scimark.
5353 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5355 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5356 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5357 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5358 pentium4_cost, nocona_cost): Set reassociation width to 1.
5359 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5360 width to 2 for fp operations and 1 otherwise.
5361 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5363 (atom_cost): Set reassociation width to 2.
5364 (slm_cost, generic_cost): Set fp reassociation width
5365 to 2 and 1 otherwise.
5366 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5367 (core_cost): Set fp reassociation width to 4 and vector to 2.
5368 (ix86_reassociation_width): Rewrite using cost table; special case
5369 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5370 and TARGET_AVX128_OPTIMAL.
5371 * config/i386/i386.h (processor_costs): Add
5372 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5373 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5374 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5375 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5376 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5377 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5379 2017-10-05 Nathan Sidwell <nathan@acm.org>
5381 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5383 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5385 * config/arm/arm.c (arm_test_fpu_data): New.
5386 (arm_run_selftests): Call arm_test_fpu_data.
5388 2017-10-04 Nathan Sidwell <nathan@acm.org>
5390 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5392 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5394 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5396 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5397 check type for aarch64_simd_valid_immediate.
5398 (aarch64_output_simd_mov_immediate): Update prototype.
5399 (aarch64_simd_valid_immediate): Update prototype.
5400 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5401 support for ORR-immediate.
5402 (and<mode>3): modified pattern to add support for BIC-immediate.
5403 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5404 now checks for valid immediate for BIC and ORR based on new enum
5406 (aarch64_output_simd_mov_immediate): Function now used to output
5407 BIC/ORR imm as well based on new enum argument.
5408 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5410 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5411 (aarch64_reg_or_bic_imm): Likewise.
5413 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5415 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5416 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5418 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5422 PR rtl-optimization/82396
5423 * haifa-sched.c (autopref_multipass_init): Simplify
5425 (autopref_rank_data): Simplify sort order.
5426 * sched-int.h (autopref_multipass_data_): Remove
5427 multi_mem_insn_p, min_offset and max_offset.
5429 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5431 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5433 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5435 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5438 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5440 * doc/sourcebuild.texi: Document vect_long_mult.
5442 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5444 PR tree-optimization/82413
5445 * fold-const.c (build_range_check): Use widest_int when comparing
5446 the maximum ETYPE value with HIGH.
5448 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5450 PR rtl-optimization/82396
5451 * haifa-sched.c (autopref_multipass_init): Simplify
5453 (autopref_rank_data): Simplify sort order.
5454 * sched-int.h (autopref_multipass_data_): Remove
5455 multi_mem_insn_p, min_offset and max_offset.
5457 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5459 PR tree-optimization/82381
5460 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5461 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5464 PR tree-optimization/82374
5465 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5466 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5467 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5468 current_function_decl to the new decl.
5470 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5472 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5473 helper macro for IEEE float128 hardware built-in functions.
5474 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5476 (TRUNCF128_ODD): Likewise.
5477 (ADDF128_ODD): Likewise.
5478 (SUBF128_ODD): Likewise.
5479 (MULF128_ODD): Likewise.
5480 (DIVF128_ODD): Likewise.
5481 (FMAF128_ODD): Likewise.
5482 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5483 UNSPEC_TRUNC_ROUND_TO_ODD.
5484 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5485 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5486 floating point round to odd instructions.
5487 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5488 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5489 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5490 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5491 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5492 (trunc<mode>sf2_hw): Change the truncate with round to odd
5493 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5494 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5495 to odd hardware instructions.
5496 (sub<mode>3_odd): Likewise.
5497 (mul<mode>3_odd): Likewise.
5498 (div<mode>3_odd): Likewise.
5499 (sqrt<mode>2_odd): Likewise.
5500 (fma<mode>4_odd): Likewise.
5501 (fms<mode>4_odd): Likewise.
5502 (nfma<mode>4_odd): Likewise.
5503 (nfms<mode>4_odd): Likewise.
5504 (trunc<mode>df2_odd): Change the truncate with round to odd
5505 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5507 * doc/extend.texi (PowerPC built-in functions): Update documentation
5508 for existing IEEE float128-bit built-in functions. Add built-in
5509 functions that generate the IEEE 128-bit floating point round to
5512 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5514 PR rtl-optimization/77729
5515 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5516 to (X&(C1&~C2))|C2 transformations.
5518 2017-10-03 Martin Jambor <mjambor@suse.cz>
5520 PR tree-optimization/82363
5521 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5522 mismatch, mark lacc written regardless of racc.
5524 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5526 PR tree-optimization/82381
5527 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5528 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5529 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5530 fallthrough into reassoc_stmt_dominates_stmt_p.
5533 * combine.c (combine_instructions): Don't combine in unreachable
5536 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5539 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5540 function to not use the have_cpu variable. Do not set cpu_index,
5541 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5543 (rs6000_valid_attribute_p): Remove duplicate initializations of
5544 old_optimize and func_optimize.
5545 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5546 (rs6000_activate_target_options): Make global.
5547 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5550 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5552 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5553 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5554 Set *poffset to 0 and *psize and *pmax_size to -1 if
5555 *poffset + *psize overflows in HOST_WIDE_INT.
5557 PR tree-optimization/82387
5558 PR tree-optimization/82388
5559 PR tree-optimization/82389
5560 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5561 instead of live_bytes non-NULL.
5563 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5566 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5569 2017-10-02 Richard Biener <rguenther@suse.de>
5571 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5572 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5573 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5575 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5577 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5578 requested precision matches the type's.
5579 * calls.c (alloc_max_size): Calculate the new candidate size as
5580 a widest_int and use wi::to_widest when comparing it with the
5581 current candidate size.
5582 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5583 zero rather than integer_zero_node.
5584 * match.pd: Check for a no-op conversion before using wi::add
5585 rather than after. Use tree_to_uhwi when summing small shift
5586 counts into an unsigned int.
5588 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5589 Alan Hayward <alan.hayward@arm.com>
5590 David Sherwood <david.sherwood@arm.com>
5593 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5594 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5595 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5596 POINTER_AND_FP_REGS.
5598 2017-10-02 Richard Biener <rguenther@suse.de>
5600 PR tree-optimization/82355
5601 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5602 a mapping for the enclosing loop but avoid generating one for
5604 (copy_bb_and_scalar_dependences): Remove premature codegen
5605 error on PHIs in blocks duplicated into multiple places.
5606 * graphite-scop-detection.c
5607 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5608 in the region use it as loop and nest to analyze the DR in.
5609 (try_generate_gimple_bb): Likewise.
5610 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5611 (add_loop_constraints): For blocks in a loop not in the region
5612 create a dimension with a single iteration.
5613 * sese.h (gbb_loop_at_index): Remove assert.
5615 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5617 * omp-expand.c (adjust_context_scope): New function.
5618 (expand_parallel_call): Call adjust_context_scope.
5620 2017-10-01 Jeff Law <law@redhat.com>
5622 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5623 dom_opt_dom_walker class with direct access to private members.
5624 Add comments. Call test_for_singularity.
5625 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5626 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5627 m_dummy_cond anymore.
5628 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5630 (pass_dominator:execute): Build the dummy_cond here and pass it
5631 to the dom_opt_dom_walker ctor.
5632 (test_for_singularity): New function.
5634 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5635 Maya Rashish <coypu@sdf.org>
5637 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5638 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5639 (alpha*-*-netbsd*) Use nbsd_tm_file.
5640 (arm*-*-netbsdelf*) Likewise.
5641 (i[34567]86-*-netbsdelf*) Likewise.
5642 (x86_64-*-netbsd*) Likewise.
5643 (mips*-*-netbsd*) Likewise.
5644 (powerpc-*-netbsd*) Likewise.
5645 (sh*-*-netbsd*) Likewise.
5646 (sparc-*-netbsdelf*) Likewise.
5647 (sparc64-*-netbsd*) Likewise.
5648 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5650 (vax-*-netbsdelf*) Likewise.
5651 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5652 (UINT_FAST8_TYPE) Likewise.
5653 (INT_FAST16_TYPE) Check CHAR_FAST16.
5654 (UINT_FAST16_TYPE) Likewise.
5656 2017-09-30 Jakub Jelinek <jakub@redhat.com>
5659 * config/i386/i386.md
5660 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
5661 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
5662 *divmodsi4_zext_2): New define_insn_and_split.
5663 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
5664 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
5665 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
5666 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
5667 New define_insn_and_split.
5668 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
5669 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
5670 operands[1] having DImode when mode is SImode.
5672 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5673 always SImode for DIV and MOD in REG_EQUAL notes.
5675 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5678 * match.pd: Fix handling of NaNs in pattern.
5680 2017-09-29 Jeff Law <law@redhat.com>
5682 * sbitmap.c (bitmap_bit_in_range_p): New function.
5683 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5684 * tree-ssa-dse.c (live_bytes_read): New function.
5685 (dse_classify_store): Ignore reads of dead bytes.
5687 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5688 typos and whitespace errors.
5689 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5690 * config/s390/s390.c (s390_emit_prologue): Likewise.
5692 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5695 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5696 with a symbol for LRA.
5698 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5700 PR rtl-optimization/82338
5701 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5703 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5705 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5706 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5707 (qsort_chk): Declare.
5708 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5709 (qsort_chk): New function.
5711 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5713 PR tree-optimization/82337
5714 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5715 phi definition if the PHI result appears in an abnormal PHI.
5716 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5717 basis appears in an abnormal PHI.
5719 2017-09-29 Richard Biener <rguenther@suse.de>
5721 * graphite-isl-ast-to-gimple.c
5722 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5723 (binary_op_to_tree): Use it.
5724 (get_rename_from_scev): Likewise.
5725 (copy_loop_phi_nodes): Likewise.
5726 (copy_bb_and_scalar_dependences): Likewise.
5727 (translate_pending_phi_nodes): Likewise.
5729 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5732 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5733 for movabsq $(i32 << shift), r64.
5735 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5737 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5738 index when encoding %esp as %rsp to avoid 0x67 prefix.
5740 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5742 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5743 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5745 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5747 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5748 Extensions with more than 16 double VFP registers.
5749 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5750 to_clear_mask and all code related to it. Replace the remaining
5751 entry by a sbitmap and adapt code accordingly.
5753 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5755 * brig-builtins.def: Change pure attributes to const.
5757 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5759 * config.gcc (default_gnu_indirect_function): Default to yes for
5760 sparc*-*-linux* with glibc.
5762 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5764 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5765 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5766 when creating .init_array and .fini_array sections with priority
5769 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5772 * config/aarch64/aarch64.c
5773 (aarch64_builtin_support_vector_misalignment): Always return false
5774 when misalignment is unknown.
5776 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5778 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5779 this function to return false if the definition used by the swap
5780 instruction is artificial, or if the memory address from which the
5781 constant value is loaded is not represented by a base address held
5782 in a register or if the base address register is a frame or stack
5783 pointer. Additionally, return false if the base address of the
5784 loaded constant is a SYMBOL_REF but is not considered to be a
5786 (replace_swapped_load_constant): New function.
5787 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5788 loaded constant vector with a load of a swapped constant vector.
5790 2017-09-27 Carl Love <cel@us.ibm.com>
5792 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5793 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5794 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5797 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5799 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5800 first, always call autopref_rank_data otherwise.
5802 2017-09-27 Richard Biener <rguenther@suse.de>
5804 * graphite-scop-detection.c (find_scop_parameters): Move
5805 loop bound handling ...
5806 (gather_bbs::before_dom_children): ... here, avoiding the need
5807 to build scop_info->loop_nest.
5808 (record_loop_in_sese): Remove.
5809 * sese.h (sese_info_t::loop_nest): Remove.
5810 * sese.c (new_sese_info): Do not allocate loop_nest.
5811 (free_sese_info): Do not free loop_nest.
5813 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5816 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5817 lhs from calls if the lhs has addressable type.
5819 2017-09-27 Richard Biener <rguenther@suse.de>
5821 * graphite.h (scop::max_alias_set): New member.
5822 * graphite-scop-detection.c: Remove references to non-existing
5823 --param in comments.
5824 (build_alias_sets): Record the maximum alias set used for drs.
5825 (build_scops): Support zero as unlimited for
5826 --param graphite-max-arrays-per-scop.
5827 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5829 (build_poly_sr_1): ... here. Compute alias set based on the
5830 maximum alias set used for drs rather than
5831 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5833 2017-09-27 Richard Biener <rguenther@suse.de>
5835 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5836 --param loop-block-tile-size=0 to disable tiling.
5838 2017-09-27 Richard Biener <rguenther@suse.de>
5840 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5841 (graphite-max-nb-scop-params): Document special value zero.
5842 * domwalk.h (dom_walker::STOP): New symbolical constant.
5843 (dom_walker::dom_walker): Add optional parameter for bb to
5845 (dom_walker::~dom_walker): Declare.
5846 (dom_walker::before_dom_children): Document STOP return value.
5847 (dom_walker::m_user_bb_to_rpo): New member.
5848 (dom_walker::m_bb_to_rpo): Likewise.
5849 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5850 mapping here if not provided by the user.
5851 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5852 provided by the user.
5853 (dom_walker::STOP): Define.
5854 (dom_walker::walk): Do not compute bb to RPO mapping here.
5855 Support STOP return value from before_dom_children to stop
5857 * graphite-optimize-isl.c (optimize_isl): If the schedule
5858 is the same still generate code if -fgraphite-identity
5859 or -floop-parallelize-all are given.
5860 * graphite-scop-detection.c: Include cfganal.h.
5861 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5863 (gather_bbs::before_dom_children): Return STOP for BBs
5865 (build_scops): Compute bb to RPO mapping and pass it to
5866 the domwalk. Treat --param graphite-max-nb-scop-params=0
5867 as not limiting the number of params.
5868 * graphite.c (graphite_initialize): Remove limit on the
5869 number of basic-blocks in a function.
5870 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5871 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5872 default value of 10.
5874 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5876 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5877 Adjust code to eliminate needing to do the shift right 32-bits
5878 operation after XSCVDPSPN.
5880 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5882 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5883 ((X / Y) != 0 -> X >= Y): Likewise.
5885 2017-09-26 Carl Love <cel@us.ibm.com>
5887 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5888 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5889 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5890 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5891 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5892 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5893 definitions and overloading.
5894 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5895 statement for P9V_BUILTIN_XST_LEN_R.
5896 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5897 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5898 define_expand and define_insn for the instructions and builtins.
5899 * doc/extend.texi: Update the built-in documentation file for the new
5901 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5902 define_insn for the instructions
5904 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5907 * gcc/config/netbsd-protos.h: New file.
5908 * gcc/config/netbsd.c: New file.
5909 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5910 * gcc/config/t-netbsd: New file.
5911 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5912 (tmake_file) Add t-netbsd.
5913 (extra_objs) Add netbsd.o.
5915 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5919 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5921 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5923 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5924 sign extension from a vector register to a GPR by doing a 32-bit
5925 direct move and then an EXTSW.
5926 (extendsi<mode>2 splitter): Likewise.
5927 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5928 right or vector extract after doing XSCVDPSPN. Use
5929 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5931 (movdi_from_sf_zero_ext): Likewise.
5932 (reload_gpr_from_vsxsf): Likewise.
5933 (p8_mfvsrd_4_disf): Delete, no longer used.
5934 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5935 then needing to move the SFmode to a GPR to use the XSCVDPSP
5936 instruction instead of FRSP and XSCVDPSPN.
5937 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5938 it is adjacent to the other XSCVSPDP insns.
5939 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5940 SFmode to be in traditional Altivec registers.
5941 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5942 (vsx_xscvspdpn): Likewise.
5943 (vsx_xscvspdpn_scalar): Likewise.
5945 2017-09-26 Martin Jambor <mjambor@suse.cz>
5947 * tree-sra.c (compare_access_positions): Put integral types first,
5948 stabilize sorting of integral types, remove conditions putting
5949 non-full-precision integers last.
5950 (sort_and_splice_var_accesses): Disable scalarization if a
5951 non-integert would be represented by a non-full-precision integer.
5953 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5955 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5956 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5957 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5958 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5959 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5960 conditionals inside the function instead of around it. Call
5961 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5962 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5964 2017-09-26 Richard Biener <rguenther@suse.de>
5966 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5968 (scop_detection::build_scop_breadth): ... this. Removed.
5969 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
5970 (scop_detection::harmful_stmt_in_bb): Likewise.
5971 (scop_detection::graphite_can_represent_stmt): Likewise.
5972 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
5973 (scop_detection::can_represent_loop): Remove recursion, fold in ...
5974 (scop_detection::can_represent_loop_1): ... this. Removed.
5975 (scop_detection::harmful_loop_in_region): Simplify after inlining
5976 the above and remove more quadraticness.
5977 (build_scops): Adjust.
5978 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
5981 2017-09-26 Jakub Jelinek <jakub@redhat.com>
5984 * config/i386/i386.c (ix86_print_operand_address_as): Only test
5985 REGNO (base) == SP_REG if base is a REG.
5988 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
5989 if it is different SSA_NAME.
5990 (optimize_range_tests_cmp_bitwise): New function.
5991 (optimize_range_tests): Call it.
5993 2017-09-26 Richard Biener <rguenther@suse.de>
5995 PR tree-optimization/82321
5996 * graphite.c (canonicalize_loop_closed_ssa): Properly check
5997 for the def being inside the loop.
5999 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6001 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6003 * config/s390/s390-builtins.def: Fix constraint on op4.
6005 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6007 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6008 independent expanders.
6009 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6010 ("vec_ordered", "vec_unordered"): New expanders.
6012 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6014 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6017 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6019 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6020 vec_unpacks_lo_v16qi.
6021 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6023 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6025 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6026 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6027 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6029 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6031 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6033 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6035 ("*vec_slb<mode>"): New insn pattern.
6036 ("vec_shr_<mode>"): New expander.
6037 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6038 and force the shift count operand to V16QImode.
6039 ("vec_srb<mode>"): Set shift count mode to V16QI.
6041 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6043 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6044 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6045 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6047 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6050 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6052 2017-09-26 Richard Biener <rguenther@suse.de>
6054 PR tree-optimization/82320
6055 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6058 2017-09-25 Jeff Law <law@redhat.com>
6060 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6061 prototype for new argument.
6062 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6063 mostly extracted from rs6000_emit_allocate_stack.
6064 (rs6000_emit_probe_stack_range_stack_clash): New function.
6065 (rs6000_emit_allocate_stack): Call
6066 rs6000_emit_probe_stack_range_stack_clash as needed.
6067 (rs6000_emit_probe_stack_range): Add additional argument
6068 to call to gen_probe_stack_range{si,di}.
6069 (output_probe_stack_range): New.
6070 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6071 (output_probe_stack_range_stack_clash): New.
6072 (rs6000_emit_prologue): Emit notes into dump file as requested.
6073 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6074 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6075 Add additional operand and pass it to output_probe_stack_range.
6077 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6079 PR tree-optimization/82163
6080 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6081 (checking_verify_loop_closed_ssa): New parameter.
6082 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6083 (check_loop_closed_ssa_stmt): Delete.
6084 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6085 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6086 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6089 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6091 * brig-builtins.def: Treat HSAIL barrier builtins as
6092 setjmp/longjump style functions.
6094 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6096 * target.def (constant_alignment): New hook.
6097 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6098 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6099 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6100 * doc/tm.texi: Regenerate.
6101 * targhooks.h (default_constant_alignment): Declare.
6102 (constant_alignment_word_strings): Likewise.
6103 * targhooks.c (default_constant_alignment): New function.
6104 (constant_alignment_word_strings): Likewise.
6105 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6106 instead of CONSTANT_ALIGNMENT.
6107 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6108 (force_const_mem): Likewise.
6109 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6110 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6111 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6112 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6113 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6115 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6116 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6117 constant_alignment_word_strings.
6118 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6119 (CONSTANT_ALIGNMENT): Likewise.
6120 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6121 (arm_constant_alignment): New function.
6122 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6123 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6124 constant_alignment_word_strings.
6125 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6126 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6127 constant_alignment_word_strings.
6128 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6129 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6130 (cris_constant_alignment): New function.
6131 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6132 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6133 (epiphany_constant_alignment): New function.
6134 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6135 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6136 constant_alignment_word_strings.
6137 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6138 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6139 constant_alignment_word_strings.
6140 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6141 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6142 constant_alignment_word_strings.
6143 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6144 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6145 * config/i386/i386.c (ix86_constant_alignment): Make static.
6146 Use the same interface as the target hook.
6147 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6148 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6149 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6150 constant_alignment_word_strings.
6151 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6152 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6153 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6154 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
6155 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6156 constant_alignment_word_strings.
6157 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
6158 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6159 constant_alignment_word_strings.
6160 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
6161 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6162 constant_alignment_word_strings.
6163 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
6164 * config/microblaze/microblaze.c (microblaze_constant_alignment):
6166 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6167 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
6168 * config/mips/mips.c (mips_constant_alignment): New function.
6169 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6170 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
6171 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
6172 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6173 (mmix_constant_alignment): Make static. Use the same interface
6175 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
6176 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6177 constant_alignment_word_strings.
6178 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
6179 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6180 constant_alignment_word_strings.
6181 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
6182 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6183 constant_alignment_word_strings.
6184 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
6185 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6186 (rs6000_constant_alignment): New function.
6187 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
6188 * config/riscv/riscv.c (riscv_constant_alignment): New function.
6189 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6190 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
6191 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6192 (rs6000_constant_alignment): New function.
6193 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
6194 * config/s390/s390.c (s390_constant_alignment): New function.
6195 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6196 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
6197 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6198 constant_alignment_word_strings.
6199 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
6200 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6201 (sparc_constant_alignment): New function.
6202 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
6203 * config/spu/spu.c (spu_constant_alignment): New function.
6204 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6205 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
6206 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6207 constant_alignment_word_strings.
6208 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
6209 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6210 constant_alignment_word_strings.
6211 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
6212 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6213 constant_alignment_word_strings.
6214 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
6215 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6216 (visium_constant_alignment): New function.
6217 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6218 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6219 (xtensa_constant_alignment): New function.
6220 * system.h (CONSTANT_ALIGNMENT): Poison.
6222 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6224 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6225 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6226 (rs6000_builtin_valid_without_lhs): New helper function.
6227 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6228 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6230 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6232 * target.h (vec_perm_indices): Use unsigned short rather than
6234 (auto_vec_perm_indices): Likewise.
6235 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6236 Use unsigned int rather than unsigned char.
6237 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6239 2017-09-25 Richard Biener <rguenther@suse.de>
6241 * cfgloop.h (sort_sibling_loops): Declare.
6242 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6243 (sort_sibling_loops): New function sorting the sibling loop list
6245 * graphite.c (graphite_transform_loops): Sort sibling loops.
6247 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6249 * target.def (vec_perm_const_ok): Change sel parameter to
6251 * optabs-query.c (can_vec_perm_p): Update accordingly.
6252 * doc/tm.texi: Regenerate.
6253 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6254 auto_vec_perm_indices and remove separate nelt field.
6255 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6256 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6257 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6258 (aarch64_expand_vec_perm_const): Update accordingly.
6259 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6260 to vec_perm_indices.
6261 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6262 auto_vec_perm_indices and remove separate nelt field.
6263 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6264 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6265 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6267 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6268 to vec_perm_indices.
6269 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6270 sel to vec_perm_indices.
6271 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6272 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6273 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6275 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6278 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6281 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6282 on the FUNCTION_DECL function context if it has a DIE that is a
6285 2017-09-25 Richard Biener <rguenther@suse.de>
6287 PR tree-optimization/82285
6288 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6291 2017-09-25 Tom de Vries <tom@codesourcery.com>
6295 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6298 2017-09-25 Richard Biener <rguenther@suse.de>
6300 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6301 ISL errors other than isl_error_quota happen. Dump if the
6302 schedule is the same.
6303 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6304 errors instead of aborting inside ISL.
6306 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6309 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6310 of libgcc_eh for m64.
6311 * config/i386/darwin64.h: Likewise.
6313 2017-09-25 Richard Biener <rguenther@suse.de>
6316 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6317 attribute for incomplete types nor twice for complete ones.
6319 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6322 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6323 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6326 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6329 * config/i386/i386.opt (mprefer-avx256): Use
6330 ix86_target_flags variable.
6331 * config/i386/i386.c (ix86_target_string): Move
6332 -mprefer-avx256 to flag2_opts.
6334 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6337 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6338 and x != -1 | y != -1 into (x & y) != -1.
6340 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6342 * config.gcc: Add new case statement to set
6343 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6344 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6345 s390x-*-linux* case statements. Added aarch64 to the list of
6346 supported architectures.
6348 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6350 PR tree-optimization/82289
6351 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6352 STMT_VINFO_RELEVANT_P.
6354 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6355 Alan Hayward <alan.hayward@arm.com>
6356 David Sherwood <david.sherwood@arm.com>
6358 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6359 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6360 (extract_range_from_binary_expr_1): Don't call
6361 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6363 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6364 Alan Hayward <alan.hayward@arm.com>
6365 David Sherwood <david.sherwood@arm.com>
6367 * target.def (preferred_vector_alignment): New hook.
6368 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6370 * doc/tm.texi: Regenerate.
6371 * targhooks.h (default_preferred_vector_alignment): Declare.
6372 * targhooks.c (default_preferred_vector_alignment): New function.
6373 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6375 (DR_TARGET_ALIGNMENT): New macro.
6376 (aligned_access_p): Update commentary.
6377 (vect_known_alignment_in_bytes): New function.
6378 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6380 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6381 Calculate the misalignment based on the target alignment rather than
6383 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6384 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6385 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6386 the target alignment, rather than masking the element misalignment
6387 with the number of elements in a vector. Also use the target
6388 alignment when calculating the maximum number of peels.
6389 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6390 instead of TYPE_ALIGN_UNIT.
6391 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6392 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6393 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6394 (vect_create_data_ref_ptr): Likewise.
6395 (vect_setup_realignment): Realign by ANDing with
6396 -DR_TARGET_MISALIGNMENT.
6397 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6398 the number of peels based on DR_TARGET_ALIGNMENT.
6399 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6400 with the guaranteed alignment boundary when deciding whether
6402 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6403 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6404 (ensure_base_align): Remove stmt_info parameter. Get the
6405 target base alignment from DR_TARGET_ALIGNMENT.
6406 (vectorizable_store): Update call accordingly. Interpret
6407 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6409 (vectorizable_load): Likewise.
6411 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6412 Alan Hayward <alan.hayward@arm.com>
6413 David Sherwood <david.sherwood@arm.com>
6415 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6416 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6417 (vect_enhance_data_refs_alignment): Likewise.
6419 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6421 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6422 error. Only quit immediately if parsing is complete.
6423 (BEGIN): Initialize fatal_err and parse_done.
6424 (begin fpu, end fpu): Check number of arguments.
6425 (begin arch, end arch): Likewise.
6426 (begin cpu, end cpu): Likewise.
6427 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6428 (optalias): Likewise.
6430 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6432 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6433 * config/arm/arm-isa.h: Delete. Move definitions to ...
6434 * arm-cpus.in: ... here. Use new feature and fgroup values.
6435 * config/arm/arm.c (arm_option_override): Use lower case for feature
6437 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6438 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6439 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6441 (print_isa_bits_for): New function.
6442 (gen_isa): New function.
6443 (gen_comm_data): Use print_isa_bits_for.
6444 (define feature): New keyword.
6445 (define fgroup): New keyword.
6446 * config/arm/t-arm (TM_H): Remove.
6447 (GTM_H): Add arm-isa.h.
6448 (arm-isa.h): Add rule to generate file.
6449 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6450 case for feature bit names.
6452 2017-09-22 Richard Biener <rguenther@suse.de>
6454 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6456 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6457 print of no dependency loops ...
6458 * graphite.c (graphite_transform_loops): ... here.
6459 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6461 (same_close_phi_node, remove_duplicate_close_phi,
6462 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6463 (canonicalize_loop_closed_ssa): ... here and simplify.
6464 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6465 (optimize_isl): Use dump_printf_loc to tell when we stopped
6466 optimizing because of an ISL timeout.
6468 2017-09-22 Richard Biener <rguenther@suse.de>
6470 PR tree-optimization/82291
6471 * tree-if-conv.c (predicate_mem_writes): Make sure to
6472 remove writes in blocks predicated with false.
6474 2017-09-22 Richard Biener <rguenther@suse.de>
6476 * sese.c: Include cfganal.h.
6477 (if_region_set_false_region): Remove.
6478 (create_if_region_on_edge): Likewise.
6479 (move_sese_in_condition): Re-implement without destroying
6482 2017-09-22 Richard Biener <rguenther@suse.de>
6484 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6485 Verify both BBs contain loop PHI nodes before dispatching to
6487 (graphite_regenerate_ast_isl): Do not recompute dominators,
6488 do not verify three times. Restructure for clarity.
6489 * graphite-scop-detection.c (same_close_phi_node,
6490 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6491 defined_in_loop_p, canonicalize_loop_closed_ssa,
6492 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6493 checking and SSA rewrite, move to ...
6494 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6495 (graphite_initialize): Do not pass in ctx, do not reset the
6496 SCEV cache, compute only dominators.
6497 (graphite_transform_loops): Allocate ISL ctx after
6498 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6499 Maintain post-dominators only around build_scops.
6500 * sese.c (if_region_set_false_region): Make static. Free
6501 and recompute dominators.
6502 (move_sese_in_condition): Assert we don't get called with
6503 post-dominators computed.
6504 * sese.h (if_region_set_false_region): Remove.
6506 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6508 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6509 mode attribute for TARGET_AVX512VL.
6511 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6513 * config/i386/i386.opt (mprefer-avx256): New option.
6514 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6516 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6517 for TARGET_PREFER_AVX256.
6518 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6520 2017-09-21 Jeff Law <law@redhat.com>
6522 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6523 Fix dump output if the only stack space is for pushed registers.
6525 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6527 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6530 2017-09-21 Martin Sebor <msebor@redhat.com>
6533 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6534 code (in C++) or code that triggers warnings.
6536 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6538 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6540 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6542 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6543 * sched-rgn.c: Ditto.
6544 * sel-sched-ir.c: Ditto.
6546 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6548 * toplev.h (set_random_seed): Adjust return type.
6549 * toplev.c (init_local_tick): Move eager initialization of random_seed
6550 to get_random_seed. Adjust comment.
6551 (init_random_seed): Inline to get_random_seed, delete.
6552 (get_random_seed): Initialize random_seed lazily.
6553 (set_random_seed): Do not return previous value.
6554 (print_switch_value): Do not call get_random_seed.
6556 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6558 * cgraph.c (delete_function_version): New, broken out from...
6559 (cgraph_node::delete_function_version): ...here. Rename to
6560 cgraph_node::delete_function_version_by_decl. Update all uses.
6561 (cgraph_node::remove): Call delete_function_version.
6563 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6566 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6567 VAR_DECLs to which addressable non-volatile parameters are mapped
6568 and for id->retvar after the return value assignment. Clear
6569 id->retval and id->retbnd after inlining.
6571 2017-09-21 Richard Biener <rguenther@suse.de>
6573 PR tree-optimization/82276
6574 PR tree-optimization/82244
6575 * tree-vrp.c (build_assert_expr_for): Set
6576 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6578 (remove_range_assertions): Revert earlier change.
6580 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6583 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6585 2017-09-21 Richard Biener <rguenther@suse.de>
6587 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6588 Restore valid IL after code generation errors.
6589 * graphite.c (graphite_transform_loops): Diagnose code
6590 generation issues as MSG_MISSED_OPTIMIZATION and continue
6591 with processing SCOPs.
6593 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6594 Alan Hayward <alan.hayward@arm.com>
6595 David Sherwood <david.sherwood@arm.com>
6597 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6598 rather than choosing between plus_constant and gen_rtx_<CODE>.
6599 * expr.c (emit_push_insn): Likewise.
6600 (expand_expr_real_2): Likewise.
6602 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6603 Alan Hayward <alan.hayward@arm.com>
6604 David Sherwood <david.sherwood@arm.com>
6606 * loop-unroll.c (split_iv): Call copy_rtx on the step.
6608 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6609 Alan Hayward <alan.hayward@arm.com>
6610 David Sherwood <david.sherwood@arm.com>
6612 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6613 calling tree_to_uhwi.
6615 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6616 Alan Hayward <alan.hayward@arm.com>
6617 David Sherwood <david.sherwood@arm.com>
6619 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6620 INTEGER_CST rather than a negative test for ADDR_EXPR.
6622 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6623 Alan Hayward <alan.hayward@arm.com>
6624 David Sherwood <david.sherwood@arm.com>
6626 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6627 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6629 2017-09-21 Richard Biener <rguenther@suse.de>
6631 PR tree-optimization/71351
6632 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6633 graphite_create_new_loop_guard): Remove, fold remaining parts
6635 (translate_isl_ast_node_for): ... here and simplify.
6637 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6640 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6641 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6642 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6643 alternative always use QI mode, for -Os imov (=R,R) alternative
6644 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6647 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6648 Jeff Law <law@redhat.com>
6650 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6651 (allocate_stack_space): New function, partially extracted from
6653 (s390_emit_prologue): Track offset to most recent stack probe.
6654 Code to allocate space moved into allocate_stack_space.
6655 Dump actions when no stack is allocated.
6656 (s390_prologue_plus_offset): New function.
6657 (s390_emit_stack_probe): Likewise.
6659 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
6661 * common.opt (Wa, Wl, Wp, g, gz=): Add
6663 (gno-column-info): Remove.
6664 (gcolumn-info): Drop RejectNegative.
6666 (gno-record-gcc-switches): Remove.
6667 (grecord-gcc-switches): Drop RejectNegative.
6668 (gno-split-dwarf): Remove.
6669 (gsplit-dwarf): Drop RejectNegative.
6670 (gno-strict-dwarf): Remove.
6671 (gstrict-dwarf): Drop RejectNegative.
6672 * config/darwin.opt (gfull, gused): Add RejectNegative.
6673 * dwarf2out.c (gen_producer_string): Drop
6674 gno-record-gcc-switches handler.
6675 * optc-gen.awk: Add g to prefixes with negative forms.
6676 * opts-common.c (remapping_prefix_p): New.
6677 (find_opt): Check it.
6678 (generate_canonical_option): Test g prefix.
6679 (option_map): Add -gno- mapping.
6680 (add_misspelling_candidates): Check remapping_prefix_p.
6682 2017-09-20 Jeff Law <law@redhat.com>
6684 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6685 thinko in stack clash protection support.
6687 * explow.c (compute_stack_clash_protection_loop_data): Use
6688 CONST_INT_P instead of explicit test. Verify object is a
6689 CONST_INT_P before looking at INTVAL.
6690 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6691 instead of explicit test.
6693 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6696 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6697 address instead of to r1 and r11.
6699 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6701 * config.gcc: Support "knm".
6702 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6703 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6705 * config/i386/i386.c (m_KNM): Define.
6706 (processor_target_table): Add "knm".
6708 (ix86_option_override_internal): Add "knm".
6709 (ix86_issue_rate): Add PROCESSOR_KNM.
6710 (ix86_adjust_cost): Ditto.
6711 (ia32_multipass_dfa_lookahead): Ditto.
6712 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6713 (fold_builtin_cpu): Add M_INTEL_KNM.
6714 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6715 (processor_type): Add PROCESSOR_KNM.
6716 * config/i386/x86-tune.def: Add m_KNM.
6717 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6719 2017-09-20 Richard Biener <rguenther@suse.de>
6721 PR tree-optimization/80213
6722 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6723 are allowed in empty BBs as well.
6724 (canonicalize_loop_closed_ssa): Also look for other complex
6726 (scop_detection::get_sese): Include the loop-closed PHI block
6728 (scop_detection::merge_sese): Remove code adding extra blocks.
6729 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6730 (build_scops): Assert the final returned scop is invalid.
6732 2017-09-20 Richard Biener <rguenther@suse.de>
6734 PR tree-optimization/82264
6735 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6737 (vn_phi_lookup): Likewise.
6738 (vn_phi_insert): Likewise.
6740 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6742 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6743 that fits into uhwi or shwi, add DW_AT_const_value regardless
6744 of early_dwarf without going through RTL, using add_AT_unsigned
6747 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6748 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6749 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6750 (DEBUG_MACRO_SECTION): Likewise.
6751 (DEBUG_LTO_MACRO_SECTION): Likewise.
6752 (DEBUG_STR_DWO_SECTION): Likewise.
6753 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6754 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6755 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6756 (DEBUG_LTO_LINE_STR_SECTION): Define.
6757 (init_sections_and_labels): Initialize debug_line_str_section
6758 variable. Initialize debug_loc_section for -gdwarf-5 to
6759 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6761 2017-09-20 Richard Biener <rguenther@suse.de>
6763 * graphite-sese-to-poly.c (extract_affine): Properly handle
6764 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6766 2017-09-20 Richard Biener <rguenther@suse.de>
6768 PR tree-optimization/81373
6769 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6770 Force SESE live-out defs to be handled even if they are
6773 2017-09-19 Jeff Law <law@redhat.com>
6775 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6776 nothing for stack adjustments with REG_STACK_CHECK.
6777 * sched-deps.c (parse_add_or_inc): Reject insns with
6778 REG_STACK_CHECK from dependency breaking.
6779 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6780 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6781 * reg-notes.def (STACK_CHECK): New note.
6783 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6784 (ix86_expand_prologue): Dump stack clash info as needed.
6785 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6787 * function.c (dump_stack_clash_frame_info): New function.
6788 * function.h (dump_stack_clash_frame_info): Prototype.
6789 (enum stack_clash_probes): New enum.
6791 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6792 flag_stack_clash_protection.
6793 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6794 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6795 (arm_frame_pointer_required): Likewise.
6796 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6797 (ia64_expand_prologue): Likewise.
6798 * config/mips/mips.c (mips_expand_prologue): Likewise.
6799 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6800 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6801 (sparc_flat_expand_prologue): Likewise.
6802 * config/spu/spu.c (spu_expand_prologue): Likewise.
6804 * explow.c: Include "params.h".
6805 (anti_adjust_stack_and_probe_stack_clash): New function.
6806 (get_stack_check_protect): Likewise.
6807 (compute_stack_clash_protection_loop_data): Likewise.
6808 (emit_stack_clash_protection_loop_start): Likewise.
6809 (emit_stack_clash_protection_loop_end): Likewise.
6810 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6811 Use anti_adjust_stack_and_probe_stack_clash.
6812 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6813 (emit_stack_clash_protection_loop_start): Likewise.
6814 (emit_stack_clash_protection_loop_end): Likewise.
6815 * rtl.h (get_stack_check_protect): Prototype.
6816 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6817 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6818 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6820 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6822 * doc/tm.texi: Rebuilt.
6823 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6824 get_stack_check_protect.
6825 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6826 * config/arm/arm.c (arm_expand_prologue): Likewise.
6827 (arm_frame_pointer_required): Likewise.
6828 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6829 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6830 * config/mips/mips.c (mips_expand_prologue): Likewise.
6831 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6832 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6833 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6834 (sparc_flat_expand_prologue): Likewise.
6836 * common.opt (-fstack-clash-protection): New option.
6837 * flag-types.h (enum stack_check_type): Note difference between
6838 -fstack-check= and -fstack-clash-protection.
6839 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6840 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6841 * toplev.c (process_options): Issue warnings/errors for cases
6842 not handled with -fstack-clash-protection.
6843 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6844 (-fstack-check): Note additional problem with -fstack-check=generic.
6845 Note that -fstack-check is primarily for Ada and refer users
6846 to -fstack-clash-protection for stack-clash-protection.
6847 Document new params for stack clash protection.
6849 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6851 * config/i386/i386.c (ix86_split_long_move): Do not handle
6852 address used for LEA in a special way.
6854 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6856 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6858 2017-09-19 Martin Sebor <msebor@redhat.com>
6861 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6862 of incompatible types.
6864 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6866 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6867 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6868 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6869 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6871 2017-09-19 Richard Biener <rguenther@suse.de>
6873 PR tree-optimization/82244
6874 * tree-vrp.c (remove_range_assertions): Do not propagate
6875 a constant to abnormals but replace the assert with a copy.
6877 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6879 PR rtl-optimization/57878
6880 PR rtl-optimization/68988
6881 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6882 avoidance test involving non_reload_pseudos. Move frequency test
6883 below the general fragmentation avoidance test.
6885 2017-09-19 Richard Biener <rguenther@suse.de>
6887 PR tree-optimization/69728
6888 * graphite-sese-to-poly.c (schedule_error): New global.
6889 (add_loop_schedule): Handle empty domain by failing the
6891 (build_original_schedule): Handle schedule_error.
6893 2017-09-19 Richard Biener <rguenther@suse.de>
6895 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6896 Do not iterate to sibling loops but only to siblings of inner
6899 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6902 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6903 that the registers are different.
6905 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6907 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6908 to processor_model and "amdfam17h" to arch_names_table.
6909 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6911 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6914 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6916 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6917 Alan Hayward <alan.hayward@arm.com>
6918 David Sherwood <david.sherwood@arm.com>
6920 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6922 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6924 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6925 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6926 vect_schedule_slp_instance.
6927 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6928 Update call to vect_slp_analyze_node_operations. Simplify return
6930 (vect_slp_analyze_bb_1): Update call accordingly.
6931 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6932 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6933 (vect_schedule_slp): Update call accordingly.
6935 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6936 Alan Hayward <alan.hayward@arm.com>
6937 David Sherwood <david.sherwood@arm.com>
6939 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6940 with types that aren't in fact scalar.
6942 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6944 * tree-vect-slp.c (vect_record_max_nunits): New function,
6946 (vect_build_slp_tree_1): ...here.
6947 (vect_build_slp_tree_2): Call it for phis too.
6949 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6951 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6952 to vect_get_vec_def_for_operand when getting the mask operand.
6954 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6955 Alan Hayward <alan.hayward@arm.com>
6956 David Sherwood <david.sherwood@arm.com>
6958 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
6961 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6962 Alan Hayward <alan.hayward@arm.com>
6963 David Sherwood <david.sherwood@arm.com>
6965 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
6966 calculation for vector booleans.
6968 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6969 Alan Hayward <alan.hayward@arm.com>
6970 David Sherwood <david.sherwood@arm.com>
6972 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
6974 (vect_transform_stmt): ...here.
6975 (vect_analyze_stmt): Use it instead of calling
6976 vectorizable_live_operation directly.
6978 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
6980 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
6981 non-SIMT targets in acc vector loops.
6983 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
6985 * configure.ac: Add arc and check if assembler supports gdwarf2.
6986 * configure: Regenerate.
6988 2017-09-18 Richard Biener <rguenther@suse.de>
6990 PR tree-optimization/82220
6991 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
6992 epilogue niters from the min_profitable_iters compute.
6994 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6997 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
6998 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
6999 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7001 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7004 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7005 switching to a new text section.
7007 2017-09-18 Richard Biener <rguenther@suse.de>
7009 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7011 (build_alias_set): Reject aliases with no access function.
7013 2017-09-18 Richard Biener <rguenther@suse.de>
7015 PR tree-optimization/79622
7016 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7018 (build_cross_bb_scalars_use): Likewise.
7020 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7022 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7024 2017-09-18 Alan Modra <amodra@gmail.com>
7027 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7028 stack_pointer_rtx for count 0. Update comments. Break up
7029 large rtl expression.
7031 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7033 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7034 Increase to 20 bytes.
7035 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7036 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7037 or avx version of the stub.
7039 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7042 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7043 compute the minimum stack alignment. Also update preferred stack
7044 boundary for leaf functions.
7046 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7048 PR tree-optimization/82228
7049 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7052 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7054 * common/config/nds32/nds32-common.c
7055 (nds32_option_optimization_table): Refine formatting.
7056 (nds32_option_optimization_table): Use -fsched-pressure and
7057 -fomit-frame-pointer for specific optimization level.
7059 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7061 * config/nds32/nds32.c: Refine formatting and comments.
7062 * config/nds32/nds32.h: Likewise.
7063 * config/nds32/nds32.md: Likewise.
7064 * config/nds32/nds32-cost.c: Likewise.
7065 * config/nds32/nds32-isr.c: Likewise.
7066 * config/nds32/nds32-md-auxiliary.c: Likewise.
7067 * config/nds32/nds32-multiple.md: Likewise.
7068 * config/nds32/nds32-predicates.c: Likewise.
7070 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7071 Jakub Jelinek <jakub@redhat.com>
7073 Add support for -std=c++2a.
7074 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7076 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7078 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7081 * doc/extend.texi (Common Function Attributes): Add
7082 references to ARM, AArch64, and S/390 specific attributes.
7083 (Function Specific Option Pragmas): Add AArch64 and S/390
7084 to list of back ends that support the target pragma.
7086 2017-09-15 Nathan Sidwell <nathan@acm.org>
7088 * doc/standards.texi: Fix C++17 description. Update URLs for
7091 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7093 * common.opt (Wcast-align=strict): New warning option.
7094 * doc/invoke.texi: Document -Wcast-align=strict.
7096 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7098 * cgraph.h (cgraph_thunk_info): Add comments.
7099 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7100 assert for VIRTUAL_* arguments stricter.
7102 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7104 PR tree-optimization/71026
7105 * match.pd: Move RDIV patterns from fold-const.c
7106 * fold-const.c (distribute_real_division): Removed.
7107 (fold_binary_loc): Remove calls to distribute_real_divison.
7109 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7111 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7112 c++1z and gnu++1z as deprecated. Change other references to
7113 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7114 Change -Wc++1z-compat to -Wc++17-compat.
7115 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7116 * dwarf2out.c (highest_c_language): Handle C++17.
7117 (gen_compile_unit_die): Likewise.
7119 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7121 PR rtl-optimization/82192
7122 * combine.c (make_extraction): Don't look through non-paradoxical
7123 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7126 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7127 Alan Hayward <alan.hayward@arm.com>
7128 David Sherwood <david.sherwood@arm.com>
7130 * target.def (function_arg_offset): New hook.
7131 * targhooks.h (default_function_arg_offset): Declare.
7132 * targhooks.c (default_function_arg_offset): New function.
7133 * function.c (locate_and_pad_parm): Use
7134 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7135 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7136 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7137 * doc/tm.texi: Regenerate.
7138 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7139 * config/spu/spu.c (spu_function_arg_offset): New function.
7140 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7141 * system.h (FUNCTION_ARG_OFFSET): Poison.
7143 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7144 Alan Hayard <alan.hayward@arm.com>
7145 David Sherwood <david.sherwood@arm.com>
7147 * target.def (truly_noop_truncation): New hook.
7148 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7149 than TRULY_NOOP_TRUNCATION.
7150 * hooks.h (hook_bool_uint_uint_true): Declare.
7151 * hooks.c (hook_bool_uint_uint_true): New function.
7152 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7153 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7154 * doc/tm.texi: Regenerate.
7155 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
7156 rather than TRULY_NOOP_TRUNCATION in comments.
7157 (simplify_comparison): Likewise.
7158 (record_truncated_value): Likewise.
7159 * expmed.c (extract_bit_field_1): Likewise.
7160 (extract_split_bit_field): Likewise.
7161 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
7162 instead of TRULY_NOOP_TRUNCATION.
7163 * function.c (assign_parm_setup_block): Likewise.
7164 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
7165 * rtlhooks.c: Include target.h.
7166 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
7167 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
7168 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
7169 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
7170 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
7171 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
7172 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
7173 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
7174 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
7175 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
7176 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
7177 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
7178 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
7179 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
7180 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
7181 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
7182 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
7183 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
7184 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
7185 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
7186 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
7187 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
7188 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
7189 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
7190 * config/mips/mips.c (mips_truly_noop_truncation): New function.
7191 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7192 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7193 rather than TRULY_NOOP_TRUNCATION in comments.
7194 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
7195 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
7196 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
7197 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
7198 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
7199 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
7200 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
7201 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
7202 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
7203 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
7204 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
7205 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7206 rather than TRULY_NOOP_TRUNCATION in comments.
7207 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
7208 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
7209 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
7210 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
7211 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
7212 TRULY_NOOP_TRUNCATION condition.
7213 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
7214 (TRULY_NOOP_TRUNCATION): Delete.
7215 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
7216 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
7217 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7218 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7219 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7220 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7221 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7222 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7223 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7224 rather than TRULY_NOOP_TRUNCATION in comments.
7225 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7226 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7227 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7228 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7229 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7230 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7232 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7235 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7236 (*cmp_ior): Likewise.
7237 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7238 (*ior_scc_scc_cmp): Likewise.
7239 (*and_scc_scc): Likewise.
7240 (*and_scc_scc_cmp): Likewise.
7242 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7243 Alan Hayard <alan.hayward@arm.com>
7244 David Sherwood <david.sherwood@arm.com>
7246 * target.def (can_change_mode_class): New hook.
7247 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7248 (hard_regno_nregs): Likewise.
7249 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7250 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7251 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7252 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7253 (LOAD_EXTEND_OP): Update accordingly.
7254 * doc/tm.texi: Regenerate.
7255 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7256 CANNOT_CHANGE_MODE_CLASS.
7257 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7258 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7259 * combine.c (simplify_set): Update accordingly.
7260 * emit-rtl.c (validate_subreg): Likewise.
7261 * recog.c (general_operand): Likewise.
7262 * regcprop.c (mode_change_ok): Likewise.
7263 * reload1.c (choose_reload_regs): Likewise.
7264 (inherit_piecemeal_p): Likewise.
7265 * rtlanal.c (simplify_subreg_regno): Likewise.
7266 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7267 instead of CANNOT_CHANGE_MODE_CLASS.
7268 (reload_cse_simplify_operands): Likewise.
7269 * reload.c (push_reload): Use targetm.can_change_mode_class
7270 instead of CANNOT_CHANGE_MODE_CLASS.
7271 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7272 REG_CANNOT_CHANGE_MODE_P.
7273 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7274 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7275 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7276 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7277 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7278 (arm_can_change_mode_class): New function.
7279 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7280 than CANNOT_CHANGE_MODE_CLASS in comments.
7281 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7282 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7283 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7284 (ix86_can_change_mode_class): ...this new function, inverting the
7285 sense of the return value.
7286 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7287 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7288 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7289 (ia64_can_change_mode_class): New function.
7290 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7291 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7292 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7293 (m32c_can_change_mode_class): ...this new function, inverting the
7294 sense of the return value.
7295 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7296 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7297 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7298 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7299 (mips_can_change_mode_class): ...this new function, inverting the
7300 sense of the return value.
7301 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7302 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7303 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7304 (msp430_can_change_mode_class): New function.
7305 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7306 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7307 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7308 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7309 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7310 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7311 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7312 (pa_cannot_change_mode_class): Replace with...
7313 (pa_can_change_mode_class): ...this new function, inverting the
7314 sense of the return value.
7315 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7316 than CANNOT_CHANGE_MODE_CLASS in comments.
7317 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7318 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7319 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7320 (pdp11_cannot_change_mode_class): Replace with...
7321 (pdp11_can_change_mode_class): ...this new function, inverting the
7322 sense of the return value.
7323 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7324 * config/powerpcspe/powerpcspe-protos.h
7325 (rs6000_cannot_change_mode_class_ptr): Delete.
7326 * config/powerpcspe/powerpcspe.c
7327 (rs6000_cannot_change_mode_class_ptr): Delete.
7328 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7329 (rs6000_option_override_internal): Assign to
7330 targetm.can_change_mode_class instead of
7331 rs6000_cannot_change_mode_class_ptr.
7332 (rs6000_cannot_change_mode_class): Replace with...
7333 (rs6000_can_change_mode_class): ...this new function, inverting the
7334 sense of the return value.
7335 (rs6000_debug_cannot_change_mode_class): Replace with...
7336 (rs6000_debug_can_change_mode_class): ...this new function.
7337 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7338 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7339 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7340 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7341 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7343 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7344 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7345 (rs6000_option_override_internal): Assign to
7346 targetm.can_change_mode_class instead of
7347 rs6000_cannot_change_mode_class_ptr.
7348 (rs6000_cannot_change_mode_class): Replace with...
7349 (rs6000_can_change_mode_class): ...this new function, inverting the
7350 sense of the return value.
7351 (rs6000_debug_cannot_change_mode_class): Replace with...
7352 (rs6000_debug_can_change_mode_class): ...this new function.
7353 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7354 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7355 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7356 (s390_can_change_mode_class): ...this new function, inverting the
7357 sense of the return value.
7358 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7359 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7360 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7361 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7362 (sh_cannot_change_mode_class): Replace with...
7363 (sh_can_change_mode_class): ...this new function, inverting the
7364 sense of the return value.
7365 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7366 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7367 (sparc_can_change_mode_class): New function.
7368 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7369 * config/spu/spu.c (spu_can_change_mode_class): New function.
7370 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7371 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7372 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7373 (visium_can_change_mode_class): New function.
7374 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7376 2017-09-15 Richard Biener <rguenther@suse.de>
7378 PR tree-optimization/82217
7379 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7380 but not undefined case.
7382 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7385 * postreload.c (reload_cse_simplify_operands): Skip
7386 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7388 2017-09-15 Richard Biener <rguenther@suse.de>
7390 PR tree-optimization/68823
7391 * graphite-scop-detection.c (build_alias_set): If we have a
7392 possible dependence check whether we can handle them by just
7393 looking at the DRs DR_ACCESS_FNs.
7394 (build_scops): If build_alias_set fails, fail the SCOP.
7396 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7398 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7399 to support float128 built-in functions that require the ISA 3.0
7401 (BU_FLOAT128_3_HW): Likewise.
7402 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7404 (FMAF128): Likewise.
7406 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7407 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7408 floating point instructions.
7409 (rs6000_invalid_builtin): Likewise.
7410 (rs6000_builtin_mask_names): Likewise.
7411 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7412 (RS6000_BTM_FLOAT128_HW): Likewise.
7413 (RS6000_BTM_COMMON): Likewise.
7414 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7416 * doc/extend.texi (RS/6000 built-in functions): Document the
7417 IEEE 128-bit floating point square root and fused multiply-add
7420 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7422 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7423 reg (r2) isn't in the set of registers defined in the prologue.
7425 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7426 Alan Hayward <alan.hayward@arm.com>
7427 David Sherwood <david.sherwood@arm.com>
7429 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7430 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7431 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7432 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7433 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7435 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7436 max_vectorization_factor.
7437 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7439 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7440 Alan Hayward <alan.hayward@arm.com>
7441 David Sherwood <david.sherwood@arm.com>
7443 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7444 (vect_worthwhile_without_simd_p): Declare.
7445 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7446 (vectorizable_reduction): Use it.
7447 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7448 (vectorizable_operation): Likewise.
7450 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7451 Alan Hayward <alan.hayward@arm.com>
7452 David Sherwood <david.sherwood@arm.com>
7454 * tree-vectorizer.h (vect_get_num_copies): New function.
7455 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7456 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7457 (vectorizable_induction): Likewise.
7458 (vectorizable_live_operation): Likewise.
7459 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7460 (vectorizable_bswap): Likewise.
7461 (vectorizable_call): Likewise.
7462 (vectorizable_conversion): Likewise.
7463 (vectorizable_assignment): Likewise.
7464 (vectorizable_shift): Likewise.
7465 (vectorizable_operation): Likewise.
7466 (vectorizable_store): Likewise.
7467 (vectorizable_load): Likewise.
7468 (vectorizable_condition): Likewise.
7469 (vectorizable_comparison): Likewise.
7470 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7472 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7473 Alan Hayward <alan.hayward@arm.com>
7474 David Sherwood <david.sherwood@arm.com>
7476 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7477 of vect_init_vector.
7479 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7480 Alan Hayward <alan.hayward@arm.com>
7481 David Sherwood <david.sherwood@arm.com>
7483 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7484 an inline wrapper that provides a location.
7485 (gimple_build_vector): Likewise.
7486 * gimple-fold.c (gimple_build_vector_from_val): New function.
7487 (gimple_build_vector): Likewise.
7488 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7489 functions to build the initial value. Always return a gimple value.
7490 (get_initial_defs_for_reduction): Likewise. Only compute
7492 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7493 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7494 (vectorizable_induction): Use gimple_build_vector rather than
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 * target.h (vec_perm_indices): New typedef.
7502 (auto_vec_perm_indices): Likewise.
7503 * optabs-query.h: Include target.h
7504 (can_vec_perm_p): Take a vec_perm_indices *.
7505 * optabs-query.c (can_vec_perm_p): Likewise.
7506 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7507 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7508 * tree-vect-generic.c (lower_vec_perm): Likewise.
7509 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7510 (vect_grouped_load_supported): Likewise.
7511 (vect_shift_permute_load_chain): Likewise.
7512 (vect_permute_store_chain): Use auto_vec_perm_indices.
7513 (vect_permute_load_chain): Likewise.
7514 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7515 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7516 Update uses of can_vec_perm_p.
7517 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7518 mode with a number of elements. Take a vec_perm_indices *.
7519 (vect_create_epilog_for_reduction): Update accordingly.
7520 Use auto_vec_perm_indices.
7521 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7522 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7523 (vect_transform_slp_perm_load): Likewise.
7524 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7525 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7526 (vect_gen_perm_mask_checked): Likewise.
7527 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7528 (vect_gen_perm_mask_checked): Likewise.
7529 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7530 (vectorizable_store): Likewise.
7531 (vectorizable_load): Likewise.
7532 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7533 (vectorizable_bswap): Likewise.
7535 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7536 Alan Hayward <alan.hayward@arm.com>
7537 David Sherwood <david.sherwood@arm.com>
7539 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7540 * tree.c (build_vector): Likewise.
7541 (build_vector_from_ctor): Update accordingly.
7542 (build_vector_from_val): Likewise.
7543 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7544 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7545 * tree-vect-generic.c (add_rshift): Likewise.
7546 (expand_vector_divmod): Likewise.
7547 (optimize_vector_constructor): Likewise.
7548 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7549 (vect_transform_slp_perm_load): Likewise.
7550 (vect_schedule_slp_instance): Likewise.
7551 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7552 (vectorizable_call): Likewise.
7553 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7554 * expmed.c (make_tree): Likewise.
7555 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7556 a vector passed to build_vector.
7557 (fold_convert_const): Likewise.
7558 (exact_inverse): Likewise.
7559 (fold_ternary_loc): Likewise.
7560 (fold_relational_const): Likewise.
7561 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7562 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7563 (const_unop): Likewise. Store the reduction accumulator in a
7564 variable rather than an array.
7565 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7566 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7567 the new vector, rather than constructing it after the input arrays.
7568 (native_interpret_vector): Use auto_vec<tree> when building
7569 a vector passed to build_vector. Add elements in order.
7570 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7571 auto_vec<tree> when building a vector passed to build_vector.
7572 (vect_create_epilog_for_reduction): Likewise.
7573 (vectorizable_induction): Likewise.
7574 (get_initial_def_for_reduction): Likewise. Fix indentation of
7576 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7578 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7579 passed to build_vector.
7581 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7582 Alan Hayward <alan.hayward@arm.com>
7583 David Sherwood <david.sherwood@arm.com>
7585 * tree-core.h (tree_base::u): Add an "nelts" field.
7586 (tree_vector): Use VECTOR_CST_NELTS as the length.
7587 * tree.c (tree_size): Likewise.
7588 (make_vector): Initialize VECTOR_CST_NELTS.
7589 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7590 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7591 TYPE_VECTOR_SUBPARTS.
7592 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7593 as the number of units, setting it from VECTOR_CST_NELTS.
7594 (const_vector_from_tree): Likewise.
7595 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7596 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7597 (fold_negate_expr_1): Likewise.
7598 (fold_convert_const): Likewise.
7599 (const_binop): Likewise. Differentiate the number of output and
7601 (const_unop): Likewise.
7602 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7603 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7604 in cases that did the opposite.
7606 2017-09-14 Richard Biener <rguenther@suse.de>
7608 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7611 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7613 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7615 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7618 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7619 if and where to split a bb, except for splitting before debug insn
7620 sequences followed by non-label real insn. Delete debug insns
7621 in between basic blocks.
7623 * combine.c (make_compound_operation_int): Formatting fixes.
7625 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7626 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7627 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7628 * config/sol2.h (LINK_EH_SPEC): Likewise.
7629 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7630 * config/s390/linux.h (LINK_SPEC): Likewise.
7631 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7632 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7633 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7634 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7635 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7636 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7637 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7638 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7640 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7642 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7644 (ENDFILE_LINUX_SPEC): Likewise.
7645 (LINK_EH_SPEC): Likewise.
7646 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7647 (LINK_OS_LINUX_SPEC32): Likewise.
7648 (LINK_OS_LINUX_SPEC64): Likewise.
7649 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7650 (LINK_OS_LINUX_SPEC): Likewise.
7652 2017-09-13 Martin Liska <mliska@suse.cz>
7655 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
7656 CASE_HIGH is NULL_TREE.
7658 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7659 Alan Hayward <alan.hayward@arm.com>
7660 David Sherwood <david.sherwood@arm.com>
7662 * target.def (secondary_memory_needed): New hook.
7663 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
7664 instead of SECONDARY_MEMORY_NEEDED.
7665 (secondary_memory_needed_mode): Likewise.
7666 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
7667 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
7668 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
7669 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
7670 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7671 * doc/tm.texi: Regenerate.
7672 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7673 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7674 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7675 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7676 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7677 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7678 mode argument first and change the reg_class arguments to reg_class_t.
7679 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7680 Make static. Update the call to inline_secondary_memory_needed.
7681 (ix86_register_move_cost): Update the call to
7682 inline_secondary_memory_needed.
7683 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7684 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7686 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7687 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7689 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7690 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7691 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7692 and match hook interface. Add comment from mips.h.
7693 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7694 * config/mmix/mmix.md (truncdfsf2): Refer to
7695 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7697 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7698 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7699 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7700 (pa_secondary_memory_needed): New function.
7701 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7702 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7703 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7704 (pdp11_secondary_memory_needed): Make static and match hook interface.
7705 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7706 * config/powerpcspe/powerpcspe-protos.h
7707 (rs6000_secondary_memory_needed_ptr): Delete.
7708 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7710 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7711 (rs6000_option_override_internal): Assign to
7712 targetm.secondary_memory_needed rather than
7713 rs6000_secondary_memory_needed_ptr.
7714 (rs6000_secondary_memory_needed): Match hook interface.
7715 (rs6000_debug_secondary_memory_needed): Likewise.
7716 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7717 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7718 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7719 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7720 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7721 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7723 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7724 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7725 (rs6000_option_override_internal): Assign to
7726 targetm.secondary_memory_needed rather than
7727 rs6000_secondary_memory_needed_ptr.
7728 (rs6000_secondary_memory_needed): Match hook interface.
7729 (rs6000_debug_secondary_memory_needed): Likewise.
7730 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7731 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7732 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7733 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7734 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7735 (sparc_secondary_memory_needed): New function.
7736 * lra-constraints.c (check_and_process_move): Refer to
7737 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7739 (curr_insn_transform): Likewise.
7740 (process_alt_operands): Use targetm.secondary_memory_needed
7741 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7742 (check_secondary_memory_needed_p): Likewise.
7743 (choose_split_class): Likewise.
7744 * reload.c: Unconditionally include code that was previously
7745 conditional on SECONDARY_MEMORY_NEEDED.
7746 (push_secondary_reload): Use targetm.secondary_memory_needed
7747 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7748 (push_reload): Likewise.
7749 * reload1.c: Unconditionally include code that was previously
7750 conditional on SECONDARY_MEMORY_NEEDED.
7751 (choose_reload_regs): Use targetm.secondary_memory_needed
7752 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7753 (gen_reload): Likewise.
7754 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7756 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7757 Alan Hayward <alan.hayward@arm.com>
7758 David Sherwood <david.sherwood@arm.com>
7760 * target.def (secondary_memory_needed_mode): New hook:
7761 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7762 * targhooks.h (default_secondary_memory_needed_mode): New function.
7763 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7764 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7765 * doc/tm.texi: Regenerate.
7766 * lra-constraints.c (check_and_process_move): Use
7767 targetm.secondary_memory_needed_mode instead of
7768 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7769 (curr_insn_transform): Likewise.
7770 * reload.c (get_secondary_mem): Likewise.
7771 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7772 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7774 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7775 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7776 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7777 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7778 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7780 * config/powerpcspe/powerpcspe-protos.h
7781 (rs6000_secondary_memory_needed_mode): Delete.
7782 * config/powerpcspe/powerpcspe.c
7783 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7784 (rs6000_secondary_memory_needed_mode): Make static.
7785 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7786 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7788 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7790 (rs6000_secondary_memory_needed_mode): Make static.
7791 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7792 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7793 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7794 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7795 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7797 (sparc_secondary_memory_needed_mode): New function.
7798 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7800 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7802 * config/aarch64/constraints.md (Umq): New constraint.
7803 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7805 (mov<mode>): Update condition.
7807 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7809 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7810 when bitposition is the same.
7812 2017-09-13 Richard Biener <rguenther@suse.de>
7814 * dwarf2out.c (output_die_symbol): Remove.
7815 (output_die): Do not output a DIEs symbol.
7817 2017-09-13 Richard Biener <rguenther@suse.de>
7820 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7821 default-def to avoid breaking iterator update with the weird
7822 interaction with cgraph_update_edges_for_call_stmt_node.
7824 2017-09-13 Richard Biener <rguenther@suse.de>
7826 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7827 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7828 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7829 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7831 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7833 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7834 Disable pc relative literal load irrespective of
7835 TARGET_FIX_ERR_A53_84341 for default.
7837 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7839 * config/sparc/sparc.c (output_return): Output the source location of
7840 the insn in the delay slot, if any.
7841 (output_sibcall): Likewise.
7843 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7846 * common.opt (-static-pie): New alias.
7847 (shared): Negate static-pie.
7848 (-no-pie): Update help text.
7850 (static-pie): New option.
7851 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7852 -static-pie support.
7853 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7854 (LINK_EH_SPEC): Likewise.
7855 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7856 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7857 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7858 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7859 (init_gcc_specs): Likewise.
7860 (init_spec): Likewise.
7861 (display_help): Update help message for -pie.
7862 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7865 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7867 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7868 (movdi_aarch64): Likewise.
7869 (movti_aarch64): Likewise.
7871 2017-09-12 Simon Wright <simon@pushface.org>
7874 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7875 calculation of the minor version, always output as 0.
7877 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7880 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7881 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7882 on it early, rather than manual conversion late. For
7883 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7884 instead of performing manual conversion.
7886 2017-09-12 Carl Love <cel@us.ibm.com>
7888 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7889 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7891 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7892 VMULOSW): Add definitions.
7893 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7894 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7895 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7896 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7897 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7899 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7901 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7903 (movti_aarch64): Likewise.
7904 (movdf_aarch64): Likewise.
7905 (movtf_aarch64): Likewise.
7906 (load_pairdi): Likewise.
7907 (store_pairdi): Likewise.
7908 (load_pairdf): Likewise.
7909 (store_pairdf): Likewise.
7910 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7911 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7912 (ldr_got_small_<mode>): Likewise.
7913 (ldr_got_small_28k_<mode>): Likewise.
7914 (ldr_got_tiny): Likewise.
7915 * config/aarch64/iterators.md (ldst_sz): New.
7916 (ldpstp_sz): Likewise.
7917 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7919 (thunderx_load): Split load_8 to load_16.
7920 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7922 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7923 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7924 (xgene1_store_pair): Split store_8 to store_16.
7925 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7926 (falkor_st_0_st_sd): Split store_8 to store_16.
7928 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7930 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7931 and store1/2/3/4 to store_4/8/12/16.
7932 * config/aarch64/aarch64.md: Update for rename.
7933 * config/arm/arm.md: Likewise.: Likewise.
7934 * config/arm/arm.c: Likewise.
7935 * config/arm/thumb1.md: Likewise.
7936 * config/arm/thumb2.md: Likewise.
7937 * config/arm/vfp.md: Likewise.
7938 * config/arm/arm-generic.md: Likewise.
7939 * config/arm/arm1020e.md: Likewise.
7940 * config/arm/arm1026ejs.md: Likewise.
7941 * config/arm/arm1136jfs.md: Likewise.
7942 * config/arm/arm926ejs.md: Likewise.
7943 * config/arm/cortex-a15.md: Likewise.
7944 * config/arm/cortex-a17.md: Likewise.
7945 * config/arm/cortex-a5.md: Likewise.
7946 * config/arm/cortex-a53.md: Likewise.
7947 * config/arm/cortex-a57.md: Likewise.
7948 * config/arm/cortex-a7.md: Likewise.
7949 * config/arm/cortex-a8.md: Likewise.
7950 * config/arm/cortex-a9.md: Likewise.
7951 * config/arm/cortex-m4.md: Likewise.
7952 * config/arm/cortex-m7.md: Likewise.
7953 * config/arm/cortex-r4.md: Likewise.
7954 * config/arm/exynos-m1.md: Likewise.
7955 * config/arm/fa526.md: Likewise.
7956 * config/arm/fa606te.md: Likewise.
7957 * config/arm/fa626te.md: Likewise.
7958 * config/arm/fa726te.md: Likewise.
7959 * config/arm/fmp626.md: Likewise.
7960 * config/arm/iwmmxt.md: Likewise.
7961 * config/arm/ldmstm.md: Likewise.
7962 * config/arm/marvell-pj4.md: Likewise.
7963 * config/arm/xgene1.md: Likewise.
7964 * config/aarch64/thunderx.md: Likewise.
7965 * config/aarch64/thunderx2t99.md: Likewise.
7966 * config/aarch64/falkor.md: Likewise.
7968 2017-09-12 Martin Liska <mliska@suse.cz>
7970 * attribs.c (private_lookup_attribute): New function.
7971 * attribs.h (private_lookup_attribute): Declared here.
7972 (lookup_attribute): Called from this place.
7974 2017-09-12 Richard Biener <rguenther@suse.de>
7976 PR tree-optimization/82157
7977 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
7978 stmts with side-effects.
7980 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7981 Alan Hayward <alan.hayward@arm.com>
7982 David Sherwood <david.sherwood@arm.com>
7984 * target.def (hard_regno_nregs): New hook.
7985 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
7986 * targhooks.h (default_hard_regno_nregs): Declare.
7987 * targhooks.c (default_hard_regno_nregs): New function.
7988 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
7989 (TARGET_HARD_REGNO_NREGS): ...this hook.
7990 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
7991 (CLASS_MAX_NREGS): Likewise.
7992 * doc/tm.texi: Regenerate.
7993 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
7994 instead of HARD_REGNO_NREGS.
7995 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
7996 HARD_REGNO_NREGS in the comment.
7997 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
7998 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
7999 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8000 Return an unsigned int.
8001 (TARGET_HARD_REGNO_NREGS): Redefine.
8002 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8003 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8004 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8005 (arc_hard_regno_nregs): New function.
8006 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8007 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8008 (arm_hard_regno_nregs): New function.
8009 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8010 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8011 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8012 (TARGET_HARD_REGNO_NREGS): Redefine.
8013 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8014 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8015 (HARD_REGNO_NREGS): Delete.
8016 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8017 (cr16_hard_regno_nregs): New function.
8018 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8019 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8020 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8021 (cris_hard_regno_nregs): New function.
8022 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8023 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8024 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8025 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8026 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8027 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8028 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8029 (frv_hard_regno_nregs): Make static. Take and return an
8031 (frv_class_max_nregs): Remove outdated copy of documentation.
8032 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8033 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8034 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8035 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8036 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8037 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8038 (TARGET_HARD_REGNO_NREGS): Redefine.
8039 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8040 (CLASS_MAX_NREGS): Update comment.
8041 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8042 (ia64_hard_regno_nregs): New function.
8043 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8044 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8045 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8046 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8047 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8049 (m32c_hard_regno_nregs): Likewise. Make static.
8050 (TARGET_HARD_REGNO_NREGS): Redefine.
8051 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8052 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8053 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8054 (m68k_hard_regno_nregs): New function.
8055 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8056 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8057 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8058 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8059 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8060 Take and return an unsigned int.
8061 (TARGET_HARD_REGNO_NREGS): Redefine.
8062 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8063 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8064 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8065 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8066 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8067 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8068 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8069 (msp430_hard_regno_nregs): Make static. Take and return an
8071 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8072 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8073 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8074 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8075 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8076 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8077 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8078 (TARGET_HARD_REGNO_NREGS): Redefine.
8079 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8080 (PA_HARD_REGNO_NREGS): ...this.
8081 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8082 (PA_HARD_REGNO_NREGS): ...this.
8083 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8084 (pa_hard_regno_nregs): New function.
8085 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8086 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8087 (pdp11_hard_regno_nregs): New function.
8088 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8089 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8090 (rs6000_hard_regno_nregs_hook): New function.
8091 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8092 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8093 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8094 Take and return an unsigned int. Move earlier in file.
8095 (TARGET_HARD_REGNO_NREGS): Redefine.
8096 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8097 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8098 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8099 (rl78_hard_regno_nregs): Make static. Take and return an
8101 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8102 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8103 (rs6000_hard_regno_nregs_hook): New function.
8104 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8105 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8106 (TARGET_HARD_REGNO_NREGS): Redefine.
8107 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8108 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8109 instead of HARD_REGNO_NREGS.
8110 (s390_hard_regno_nregs): New function.
8111 (s390_hard_regno_mode_ok): Add comment from s390.h.
8112 (TARGET_HARD_REGNO_NREGS): Redefine.
8113 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8114 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8115 (sh_hard_regno_nregs): New function.
8116 (sh_pass_in_reg_p): Use it.
8117 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8118 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8119 (sparc_hard_regno_nregs): New function.
8120 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8121 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8122 (spu_function_arg_advance): Use it, supplying a valid register number.
8123 (TARGET_HARD_REGNO_NREGS): Redefine.
8124 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8125 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8126 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8127 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8128 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8129 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8130 (CLASS_MAX_NREGS): Remove copy of old documentation.
8131 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8132 (visium_hard_regno_nregs): New function.
8133 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8134 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8135 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8136 xtensa_hard_regno_nregs): New function.
8137 * system.h (HARD_REGNO_NREGS): Poison.
8139 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8141 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8142 hard_regno_nregs instead of HARD_REGNO_NREGS.
8143 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8144 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8145 (c6x_expand_epilogue): Likewise.
8146 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8147 (frv_read_iacc_argument): Likewise.
8148 * config/sh/sh.c: Include regs.h.
8149 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8150 (regs_used): Likewise.
8151 (output_stack_adjust): Likewise.
8152 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8153 * expmed.c: Include regs.h.
8154 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8155 * ree.c: Include regs.h.
8156 (combine_reaching_defs): Use hard_regno_nregs instead of
8158 (add_removable_extension): Likewise.
8160 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8162 * regs.h (hard_regno_nregs): Turn into a function.
8163 (end_hard_regno): Update accordingly.
8164 * caller-save.c (setup_save_areas): Likewise.
8165 (save_call_clobbered_regs): Likewise.
8166 (replace_reg_with_saved_mem): Likewise.
8167 (insert_restore): Likewise.
8168 (insert_save): Likewise.
8169 * combine.c (can_change_dest_mode): Likewise.
8170 (move_deaths): Likewise.
8171 (distribute_notes): Likewise.
8172 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
8173 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
8174 (rs6000_split_multireg_move): Likewise.
8175 (rs6000_register_move_cost): Likewise.
8176 (rs6000_memory_move_cost): Likewise.
8177 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
8178 (rs6000_split_multireg_move): Likewise.
8179 (rs6000_register_move_cost): Likewise.
8180 (rs6000_memory_move_cost): Likewise.
8181 * cselib.c (cselib_reset_table): Likewise.
8182 (cselib_lookup_1): Likewise.
8183 * emit-rtl.c (set_mode_and_regno): Likewise.
8184 * function.c (aggregate_value_p): Likewise.
8185 * ira-color.c (setup_profitable_hard_regs): Likewise.
8186 (check_hard_reg_p): Likewise.
8187 (calculate_saved_nregs): Likewise.
8188 (assign_hard_reg): Likewise.
8189 (improve_allocation): Likewise.
8190 (calculate_spill_cost): Likewise.
8191 * ira-emit.c (modify_move_list): Likewise.
8192 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
8193 (ira_hard_reg_in_set_p): Likewise.
8194 * ira.c (setup_reg_mode_hard_regset): Likewise.
8195 (clarify_prohibited_class_mode_regs): Likewise.
8196 (check_allocation): Likewise.
8197 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8198 (lra_setup_reg_renumber): Likewise.
8199 (setup_try_hard_regno_pseudos): Likewise.
8200 (spill_for): Likewise.
8201 (assign_hard_regno): Likewise.
8202 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
8203 * lra-constraints.c (in_class_p): Likewise.
8204 (lra_constraint_offset): Likewise.
8205 (simplify_operand_subreg): Likewise.
8206 (lra_constraints): Likewise.
8207 (split_reg): Likewise.
8208 (split_if_necessary): Likewise.
8209 (invariant_p): Likewise.
8210 (inherit_in_ebb): Likewise.
8211 * lra-lives.c (process_bb_lives): Likewise.
8212 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
8213 (get_hard_regs): Likewise.
8214 (do_remat): Likewise.
8215 * lra-spills.c (assign_spill_hard_regs): Likewise.
8216 * mode-switching.c (create_pre_exit): Likewise.
8217 * postreload.c (reload_combine_recognize_pattern): Likewise.
8218 * recog.c (peep2_find_free_register): Likewise.
8219 * regcprop.c (kill_value_regno): Likewise.
8220 (set_value_regno): Likewise.
8221 (copy_value): Likewise.
8222 (maybe_mode_change): Likewise.
8223 (find_oldest_value_reg): Likewise.
8224 (copyprop_hardreg_forward_1): Likewise.
8225 * regrename.c (check_new_reg_p): Likewise.
8226 (regrename_do_replace): Likewise.
8227 * reload.c (push_reload): Likewise.
8228 (combine_reloads): Likewise.
8229 (find_dummy_reload): Likewise.
8230 (operands_match_p): Likewise.
8231 (find_reloads): Likewise.
8232 (find_equiv_reg): Likewise.
8233 (reload_adjust_reg_for_mode): Likewise.
8234 * reload1.c (count_pseudo): Likewise.
8235 (count_spilled_pseudo): Likewise.
8236 (find_reg): Likewise.
8237 (clear_reload_reg_in_use): Likewise.
8238 (free_for_value_p): Likewise.
8239 (allocate_reload_reg): Likewise.
8240 (choose_reload_regs): Likewise.
8241 (reload_adjust_reg_for_temp): Likewise.
8242 (emit_reload_insns): Likewise.
8243 (delete_output_reload): Likewise.
8244 * rtlanal.c (subreg_get_info): Likewise.
8245 * sched-deps.c (sched_analyze_reg): Likewise.
8246 * sel-sched.c (init_regs_for_mode): Likewise.
8247 (mark_unavailable_hard_regs): Likewise.
8248 (choose_best_reg_1): Likewise.
8249 (verify_target_availability): Likewise.
8250 * valtrack.c (dead_debug_insert_temp): Likewise.
8251 * var-tracking.c (track_loc_p): Likewise.
8252 (emit_note_insn_var_location): Likewise.
8253 * varasm.c (make_decl_rtl): Likewise.
8254 * reginfo.c (choose_hard_reg_mode): Likewise.
8255 (init_reg_modes_target): Refer directly to
8256 this_target_regs->x_hard_regno_nregs.
8258 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8260 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8261 instead of hard_regno_nregs.
8263 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8265 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8266 end_hard_regno instead of hard_regno_nregs.
8267 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8268 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8269 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8270 * ira-color.c (improve_allocation): Likewise.
8271 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8272 * lra-lives.c (mark_regno_live): Likewise.
8273 (mark_regno_dead): Likewise.
8274 * lra-remat.c (operand_to_remat): Likewise.
8275 * lra.c (collect_non_operand_hard_regs): Likewise.
8276 * postreload.c (reload_combine_note_store): Likewise.
8277 (move2add_valid_value_p): Likewise.
8278 * reload.c (regno_clobbered_p): Likewise.
8280 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8282 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8284 * config/v850/v850.c (v850_reorg): Likewise.
8285 * reload.c (refers_to_regno_for_reload_p): Likewise.
8286 (find_equiv_reg): Likewise.
8287 * reload1.c (reload_reg_reaches_end_p): Likewise.
8289 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8291 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8293 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8294 * config/arm/arm.c (output_move_neon): Likewise.
8295 (arm_attr_length_move_neon): Likewise.
8296 (neon_split_vcombine): Likewise.
8297 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8298 (c6x_mark_reg_written): Likewise.
8299 (c6x_dwarf_register_span): Likewise.
8300 * config/i386/i386.c (ix86_save_reg): Likewise.
8301 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8302 (rws_access_reg): Likewise.
8303 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8304 * mode-switching.c (create_pre_exit): Likewise.
8305 * ree.c (combine_reaching_defs): Likewise.
8306 (add_removable_extension): Likewise.
8307 * regcprop.c (find_oldest_value_reg): Likewise.
8308 (copyprop_hardreg_forward_1): Likewise.
8309 * reload.c (reload_inner_reg_of_subreg): Likewise.
8310 (push_reload): Likewise.
8311 (combine_reloads): Likewise.
8312 (find_dummy_reload): Likewise.
8313 (reload_adjust_reg_for_mode): Likewise.
8314 * reload1.c (find_reload_regs): Likewise.
8315 (forget_old_reloads_1): Likewise.
8316 (reload_reg_free_for_value_p): Likewise.
8317 (reload_adjust_reg_for_temp): Likewise.
8318 (emit_reload_insns): Likewise.
8319 (delete_output_reload): Likewise.
8320 * sel-sched.c (choose_best_reg_1): Likewise.
8321 (choose_best_pseudo_reg): Likewise.
8323 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8324 Alan Hayward <alan.hayward@arm.com>
8325 David Sherwood <david.sherwood@arm.com>
8327 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8328 * target.def (slow_unaligned_access): New hook.
8329 * targhooks.h (default_slow_unaligned_access): Declare.
8330 * targhooks.c (default_slow_unaligned_access): New function.
8331 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8332 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8333 * doc/tm.texi: Regenerate.
8334 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8335 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8336 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8338 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8339 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8341 (rs6000_slow_unaligned_access): New function.
8342 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8343 (expand_block_compare): Likewise.
8344 (expand_strn_compare): Likewise.
8345 (rs6000_rtx_costs): Likewise.
8346 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8347 (riscv_slow_unaligned_access): Likewise.
8348 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8349 (riscv_slow_unaligned_access_p): ...this and make static.
8350 (riscv_option_override): Update accordingly.
8351 (riscv_slow_unaligned_access): New function.
8352 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8353 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8354 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8355 (rs6000_slow_unaligned_access): New function.
8356 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8357 (rs6000_rtx_costs): Likewise.
8358 * config/rs6000/rs6000-string.c (expand_block_compare)
8359 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8360 of SLOW_UNALIGNED_ACCESS.
8361 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8362 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8363 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8364 of SLOW_UNALIGNED_ACCESS.
8365 * expmed.c (simple_mem_bitfield_p): Likewise.
8366 * expr.c (alignment_for_piecewise_move): Likewise.
8367 (emit_group_load_1): Likewise.
8368 (emit_group_store): Likewise.
8369 (copy_blkmode_from_reg): Likewise.
8370 (emit_push_insn): Likewise.
8371 (expand_assignment): Likewise.
8372 (store_field): Likewise.
8373 (expand_expr_real_1): Likewise.
8374 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8375 * lra-constraints.c (simplify_operand_subreg): Likewise.
8376 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8377 * gimple-ssa-store-merging.c: Likewise in block comment at start
8379 * tree-ssa-strlen.c: Include target.h.
8380 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8381 of SLOW_UNALIGNED_ACCESS.
8382 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8384 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8386 PR rtl-optimization/82185
8387 * expmed.c (emit_store_flag_int): Only test tem if it has been
8390 2017-09-12 Richard Biener <rguenther@suse.de>
8393 * match.pd ((FTYPE) N CMP CST): Fix typo.
8395 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8397 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8399 (mips_near_type_p): Add 'short_call' attribute as a synonym
8401 * doc/extend.texi (short_call): Document new function attribute.
8403 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8406 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8407 assertion check that in the condition.
8408 (get_atomic_generic_size): Likewise. Before testing if parameter
8409 has pointer type, if it has array type, call for C++
8410 default_conversion to perform array-to-pointer conversion.
8412 2017-09-12 Richard Biener <rguenther@suse.de>
8414 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8415 for operations we cannot scalarize.
8417 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8419 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8420 vectors heap vectors. Clean up comments.
8421 Make visited_bbs a reference.
8422 (profitable_jump_thread_path): Make GC
8423 vectors heap vectors. Clean up comments.
8425 (convert_and_register_jump_thread_path): Make GC vectors heap
8427 (check_subpath_and_update_thread_path): Same. Clean up comments.
8428 Make visited_bbs a reference.
8429 (handle_phi): Abstract common code to to
8430 register_jump_thread_path_if_profitable.
8431 Rename VAR_BB to DEF_BB.
8433 Make GC vectors heap vectors.
8434 Make visited_bbs a reference.
8435 (handle_assignment): Same.
8436 (register_jump_thread_path_if_profitable): New.
8437 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8439 Make GC vectors heap vectors. Clean up comments.
8440 Make visited_bbs a reference.
8441 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8442 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8445 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8448 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8449 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8451 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8453 Revert r251800 and r251799.
8455 2017-09-11 Martin Jambor <mjambor@suse.cz>
8458 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8459 arguments in advance.
8460 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8461 use it to find predecessor edges.
8462 (naive_outof_ssa): Collect vector of predecessors.
8464 2017-09-08 Jason Merrill <jason@redhat.com>
8466 PR c++/70029 - ICE with ref-qualifier and -flto
8467 * langhooks.h (struct lang_hooks_for_types): Add
8468 copy_lang_qualifiers.
8469 * attribs.c (build_type_attribute_qual_variant): Use it.
8470 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8472 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8473 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8475 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8478 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8479 (*mulsi3_sp64): New instruction.
8480 (mulsi3): New expander.
8482 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8484 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8486 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8488 * sancov.c: Include memmodel.h.
8490 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8493 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8496 2017-09-07 Carl Love <cel@us.ibm.com>
8498 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8499 the sldi instruction.
8501 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8503 * sancov.c: Include tm_p.h.
8505 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8508 * output.h (switch_to_other_text_partition): New declaration.
8509 * varasm.c (switch_to_other_text_partition): New function.
8510 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8511 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8512 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8513 to the other text partition before emitting LCL label and switch back
8514 after emitting the word after it.
8516 2017-09-07 Richard Biener <rguenther@suse.de>
8518 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8519 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8520 critical edges here, after loop init.
8521 (pass_data_pre): Remove PROP_no_crit_edges flags.
8522 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8523 for valueization of call args to avoid leaking VN_TOP.
8524 (visit_use): Assert we do not visit default defs.
8525 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8526 Use error_mark_node to more easily detect leaking VN_TOP.
8527 All default-defs are varying, not VN_TOP. Mark them visited.
8528 (run_scc_vn): Make code match comment.
8530 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8532 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8533 OPTION_MASK_FLOAT128_KEYWORD.
8534 (POWERPC_MASKS): Likewise.
8535 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8536 support for the -mfloat128-type option, and make -mfloat128
8537 default on PowerPC Linux systems. Define or undefine
8538 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8539 Define __float128 to be __ieee128 if IEEE 128-bit support is
8540 enabled, or undefine it.
8541 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8542 Delete defining __FLOAT128_TYPE__.
8543 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8544 -mfloat128-type option and make -mfloat128 default on PowerPC
8546 (TARGET_FLOAT128_TYPE): Likewise.
8547 (-mfloat128-type): Likewise.
8548 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8549 Delete the -mfloat128-type option and make -mfloat128 default on
8550 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8551 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8552 128-bit floating point is enabled. Change tests from using
8553 -mfloat128-type to -mfloat128.
8554 (rs6000_mangle_type): Use the correct mangling for the __float128
8555 type even if normal long double is restricted to 64-bits.
8556 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8558 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8559 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8560 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8562 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8563 documentation for -mfloat128.
8565 2017-09-06 Olivier Hainque <hainque@adacore.com>
8567 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8569 2017-09-06 Wish Wu <wishwu007@gmail.com>
8570 Jakub Jelinek <jakub@redhat.com>
8572 * asan.c (initialize_sanitizer_builtins): Add
8573 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8574 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8575 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8576 BT_FN_VOID_UINT64_PTR variables.
8577 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8578 (BT_FN_VOID_UINT16_UINT16): Likewise.
8579 (BT_FN_VOID_UINT32_UINT32): Likewise.
8580 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8581 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8582 (BT_FN_VOID_UINT64_PTR): Likewise.
8583 * common.opt (flag_sanitize_coverage): New variable.
8584 (fsanitize-coverage=trace-pc): Remove.
8585 (fsanitize-coverage=): Add.
8586 * flag-types.h (enum sanitize_coverage_code): New enum.
8587 * fold-const.c (fold_range_test): Disable non-short-circuit
8588 optimization if flag_sanitize_coverage.
8589 (fold_truth_andor): Likewise.
8590 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8591 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8592 (coverage_sanitizer_opts): New array.
8593 (get_closest_sanitizer_option): Add OPTS argument, handle also
8594 OPT_fsanitize_coverage_.
8595 (parse_sanitizer_options): Adjusted to also handle
8596 OPT_fsanitize_coverage_.
8597 (common_handle_option): Add OPT_fsanitize_coverage_.
8598 * sancov.c (instrument_comparison, instrument_switch): New function.
8599 (sancov_pass): Add trace-cmp support.
8600 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8601 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8602 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8603 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8604 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8605 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8606 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8607 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8608 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8609 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8611 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8613 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8614 error. Only quit immediately if parsing is complete.
8615 (BEGIN): Initialize fatal_err and parse_done.
8616 (begin fpu, end fpu): Check number of arguments.
8617 (begin arch, end arch): Likewise.
8618 (begin cpu, end cpu): Likewise.
8619 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8620 (optalias): Likewise.
8622 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8624 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8625 * config/arm/arm-isa.h: Delete. Move definitions to ...
8626 * arm-cpus.in: ... here. Use new feature and fgroup values.
8627 * config/arm/arm.c (arm_option_override): Use lower case for feature
8629 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8630 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8631 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8633 (print_isa_bits_for): New function.
8634 (gen_isa): New function.
8635 (gen_comm_data): Use print_isa_bits_for.
8636 (define feature): New keyword.
8637 (define fgroup): New keyword.
8638 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8639 (arm-isa.h): Add rule to generate file.
8640 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8641 case for feature bit names.
8643 2017-09-06 Richard Biener <rguenther@suse.de>
8645 * tree-ssa-pre.c (NECESSARY): Remove.
8646 (create_expression_by_pieces): Do not touch pass-local flags.
8647 (insert_into_preds_of_block): Likewise.
8648 (do_pre_regular_insertion): Likewise.
8649 (eliminate_insert): Likewise.
8650 (eliminate_dom_walker::before_dom_children): Likewise.
8651 (fini_eliminate): Do not look at inserted_exprs.
8652 (mark_operand_necessary): Remove.
8653 (remove_dead_inserted_code): Replace with simple work-list
8654 algorithm based on inserted_exprs and SSA uses.
8655 (pass_pre::execute): Re-order fini_eliminate and
8656 remove_dead_inserted_code.
8658 2017-09-06 Olivier Hainque <hainque@adacore.com>
8660 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
8661 for VxWorks 7. Adjust surrounding comments.
8663 2017-09-06 Richard Biener <rguenther@suse.de>
8665 * gimple-ssa-strength-reduction.c
8666 (find_candidates_dom_walker::before_dom_children): Also allow
8669 2017-09-06 Richard Biener <rguenther@suse.de>
8671 PR tree-optimization/82108
8672 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8673 for gap in the non-permutation SLP case.
8675 2017-09-06 Martin Jambor <mjambor@suse.cz>
8677 PR tree-optimization/82078
8678 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8679 add_access_to_work_queue...
8680 (build_accesses_from_assign): ...here.
8681 (propagate_all_subaccesses): Make sure racc is the group
8682 representative, if there is one.
8684 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8687 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8690 2017-09-06 Richard Biener <rguenther@suse.de>
8692 * gimple-ssa-strength-reduction.c
8693 (find_candidates_dom_walker::before_doom_children): Use a
8694 type and not a mode check.
8696 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8699 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8701 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8703 2017-09-05 Jeff Law <law@redhat.com>
8705 PR tree-optimization/64910
8706 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8707 cases where we have 3 or more operands.
8709 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8712 * omp-low.c (lower_omp_for): Recompute tree invariant if
8713 gimple_omp_for_initial/final is ADDR_EXPR.
8716 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8717 into gimple val before gimplification fo the COND_EXPR.
8719 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8721 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8722 REGION_COPY argument.
8723 (thread_through_all_blocks): Remove unused argument to
8724 duplicate_thread_path.
8726 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8727 Alan Hayward <alan.hayward@arm.com>
8728 David Sherwood <david.sherwood@arm.com>
8730 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8731 Take a scalar_mode rather than a machine_mode.
8732 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8733 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8734 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8735 (aarch64_gen_adjusted_ldpstp): Likewise.
8736 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8738 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8739 Alan Hayward <alan.hayward@arm.com>
8740 David Sherwood <david.sherwood@arm.com>
8742 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8743 Take a scalar_int_mode instead of a machine_mode.
8744 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8745 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8746 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8747 (aarch64_simd_attr_length_rglist): Delete.
8748 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8749 a scalar_int_mode instead of a machine_mode.
8750 (aarch64_add_offset): Likewise.
8751 (aarch64_internal_mov_immediate): Likewise
8752 (aarch64_add_constant_internal): Likewise.
8753 (aarch64_add_constant): Likewise.
8754 (aarch64_movw_imm): Likewise.
8755 (aarch64_rtx_arith_op_extract_p): Likewise.
8756 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8757 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8758 Remove assert that the mode isn't a vector.
8759 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8760 (aarch64_expand_mov_immediate): Update calls after above changes.
8761 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8762 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8763 (aarch64_move_imm): Likewise.
8764 (aarch64_can_const_movi_rtx_p): Likewise.
8765 (aarch64_strip_extend): Likewise.
8766 (aarch64_extr_rtx_p): Likewise.
8767 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8768 a CONST_INT when the mode parameter is VOIDmode.
8769 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8771 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8773 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8774 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8775 (bitwise_type_for_mode): Update accordingly.
8777 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8779 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8780 * stor-layout.c (mode_for_size_tree): Likewise.
8781 (mode_for_array): Update accordingly.
8782 (layout_decl): Likewise.
8783 (compute_record_mode): Likewise. Only set the mode once.
8785 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8787 * target.def (get_mask_mode): Change return type to opt_mode.
8789 * doc/tm.texi: Regenerate.
8790 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8791 * targhooks.c (default_get_mask_mode): Likewise.
8792 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8793 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8794 targetm.get_mask_mode.
8795 * tree.c (build_truth_vector_type): Likewise.
8797 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8799 * machmode.h (mode_for_vector): Return an opt_mode.
8800 * stor-layout.c (mode_for_vector): Likewise.
8801 (mode_for_int_vector): Update accordingly.
8802 (layout_type): Likewise.
8803 * config/i386/i386.c (emit_memmov): Likewise.
8804 (ix86_expand_set_or_movmem): Likewise.
8805 (ix86_expand_vector_init): Likewise.
8806 (ix86_get_mask_mode): Likewise.
8807 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8809 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8810 * expmed.c (extract_bit_field_1): Likewise.
8811 * expr.c (expand_expr_real_2): Likewise.
8812 * optabs-query.c (can_vec_perm_p): Likewise.
8813 (can_vec_mask_load_store_p): Likewise.
8814 * optabs.c (expand_vec_perm): Likewise.
8815 * targhooks.c (default_get_mask_mode): Likewise.
8816 * tree-vect-stmts.c (vectorizable_store): Likewise.
8817 (vectorizable_load): Likewise.
8818 (get_vectype_for_scalar_type_and_size): Likewise.
8820 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8822 * machmode.h (mode_for_int_vector): New function.
8823 * stor-layout.c (mode_for_int_vector): Likewise.
8824 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8825 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8826 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8827 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8828 (s390_expand_vcond): Likewise.
8830 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8832 * machmode.h (opt_machine_mode): New type.
8833 (opt_mode<T>): Allow construction from anything that can be
8835 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8836 (mode_for_size): Return an opt_machine_mode.
8837 * stor-layout.c (mode_for_size): Likewise.
8838 (mode_for_size_tree): Update call accordingly.
8839 (bitwise_mode_for_mode): Likewise.
8840 (make_fract_type): Likewise.
8841 (make_accum_type): Likewise.
8842 * caller-save.c (replace_reg_with_saved_mem): Update call
8844 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8845 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8846 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8847 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8848 * expmed.c (extract_bit_field_1): Likewise.
8849 * reload.c (get_secondary_mem): Likewise.
8850 * varasm.c (assemble_integer): Likewise.
8851 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8854 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8856 * machmode.h (decimal_float_mode_for_size): New function.
8857 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8858 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8859 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8860 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8861 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8862 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8864 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8866 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8867 (get_builtin_sync_mode): Likewise.
8868 (expand_ifn_atomic_compare_exchange): Likewise.
8869 (expand_builtin_atomic_clear): Likewise.
8870 (expand_builtin_atomic_test_and_set): Likewise.
8871 (fold_builtin_atomic_always_lock_free): Likewise.
8872 * calls.c (compute_argument_addresses): Likewise.
8873 (emit_library_call_value_1): Likewise.
8874 (store_one_arg): Likewise.
8875 * combine.c (combine_instructions): Likewise.
8876 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8877 * config/arm/arm.c (arm_function_value): Likewise.
8878 (aapcs_allocate_return_reg): Likewise.
8879 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8880 * config/i386/i386.c (construct_container): Likewise.
8881 (ix86_gimplify_va_arg): Likewise.
8882 (ix86_expand_sse_cmp): Likewise.
8883 (emit_memmov): Likewise.
8884 (emit_memset): Likewise.
8885 (expand_small_movmem_or_setmem): Likewise.
8886 (ix86_expand_pextr): Likewise.
8887 (ix86_expand_pinsr): Likewise.
8888 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8889 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8891 * config/mips/mips.c (mips_function_value_1) Likewise.
8892 (mips_block_move_straight): Likewise.
8893 (mips_expand_ins_as_unaligned_store): Likewise.
8894 * config/powerpcspe/powerpcspe.c
8895 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8896 (rs6000_darwin64_record_arg_flush): Likewise.
8897 * config/rs6000/rs6000.c
8898 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8899 (rs6000_darwin64_record_arg_flush): Likewise.
8900 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8901 (sparc_function_value_1): Likewise.
8902 * config/spu/spu.c (adjust_operand): Likewise.
8903 (spu_emit_branch_or_set): Likewise.
8904 (arith_immediate_p): Likewise.
8905 * emit-rtl.c (gen_lowpart_common): Likewise.
8906 * expr.c (expand_expr_real_1): Likewise.
8907 * function.c (assign_parm_setup_block): Likewise.
8908 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8909 * reload1.c (alter_reg): Likewise.
8910 * stor-layout.c (mode_for_vector): Likewise.
8911 (layout_type): Likewise.
8913 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8915 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8916 (spu_convert_move): Likewise.
8917 * lower-subreg.c (resolve_simple_move): Likewise.
8919 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8922 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8923 define_insn to a define_expand.
8924 (altivec_vsum2sws_direct): New define_insn.
8925 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8927 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8929 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8932 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8937 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8938 realign stack if stack alignment needed is less than incoming
8941 2017-09-05 Marek Polacek <polacek@redhat.com>
8944 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8947 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8949 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8951 2017-09-05 Richard Biener <rguenther@suse.de>
8953 PR tree-optimization/82084
8954 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8956 2017-09-05 Richard Biener <rguenther@suse.de>
8958 PR tree-optimization/82102
8959 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8961 2017-09-05 Martin Liska <mliska@suse.cz>
8963 PR tree-optimization/82032
8964 * tree-cfg.c (generate_range_test): New function.
8965 * tree-cfg.h (generate_range_test): Declared here.
8966 * tree-cfgcleanup.c (convert_single_case_switch): New function.
8967 (cleanup_control_expr_graph): Use it.
8968 * tree-switch-conversion.c (try_switch_expansion): Remove
8970 (emit_case_nodes): Use generate_range_test.
8972 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8975 * config/i386/i386.md (*<btsc><mode>_mask): Add
8976 TARGET_USE_BT to insn constraint.
8977 (*btr<mode>_mask): Ditto.
8979 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
8981 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
8982 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
8984 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
8987 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
8988 TARGET_NEON and TARGET_IWMMXT.
8989 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
8990 TARGET_NEON and TARGET_IWMMXT.
8991 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
8993 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8995 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
8996 (ix86_rewrite_tls_address): Ditto.
8997 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
8998 (ix86_rewrite_tls_address_1): Ditto.
8999 (ix86_rewrite_tls_address): Ditto.
9000 * config/i386/predicates.md (tls_address_pattern): New predicate.
9001 * config/i386/i386.md (TLS address splitter): New splitter.
9003 2017-09-04 Richard Biener <rguenther@suse.de>
9005 PR tree-optimization/82084
9006 * fold-const.h (can_native_encode_string_p): Declare.
9007 * fold-const.c (can_native_encode_string_p): Factor out from ...
9008 (native_encode_string): ... here.
9009 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9010 vectorizing stores from constants we later cannot handle.
9012 2017-09-04 Marek Polacek <polacek@redhat.com>
9015 * doc/invoke.texi: Update -Wtautological-compare documentation.
9017 2017-09-04 Jeff Law <law@redhat.com>
9019 PR tree-optimization/64910
9020 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9021 swap the first and last operand if the last is a constant.
9023 2017-09-04 Marek Polacek <polacek@redhat.com>
9026 * convert.c (do_narrow): When sanitizing signed integer overflows,
9027 bail out for signed types.
9028 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9030 2017-09-04 Richard Biener <rguenther@suse.de>
9032 PR tree-optimization/82060
9033 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9034 Move devirtualization after stmt folding and before EH/AB/noreturn
9035 cleanup to get the stmt refs canonicalized. Use a bool instead
9036 of gimple_modified_p since that doesn't work for NOPs. Schedule
9037 NOPs generated by folding for removal.
9039 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9040 Alan Hayward <alan.hayward@arm.com>
9041 David Sherwood <david.sherwood@arm.com>
9043 * coretypes.h (pad_direction): New enum.
9044 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9045 (FUNCTION_ARG_PADDING): Likewise.
9046 * target.def (function_arg_padding): New hook.
9047 * targhooks.h (default_function_arg_padding): Declare.
9048 * targhooks.c (default_function_arg_padding): New function.
9049 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9050 (TARGET_FUNCTION_ARG_PADDING): ...this.
9051 * doc/tm.texi: Regenerate.
9052 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9053 instead of direction.
9054 (compute_argument_addresses): Likewise.
9055 (load_register_parameters): Likewise.
9056 (emit_library_call_value_1): Likewise.
9057 (store_one_arg): Use targetm.calls.function_arg_padding instead
9058 of FUNCTION_ARG_PADDING.
9059 (must_pass_in_stack_var_size_or_pad): Likewise.
9060 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9061 (emit_group_store): Likewise.
9062 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9063 instead of FUNCTION_ARG_PADDING.
9064 (emit_push_insn): Likewise, and propagate enum change throughout
9066 * function.h (direction): Delete.
9067 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9069 * function.c (assign_parm_find_stack_rtl): Likewise.
9070 (assign_parm_setup_block_p): Likewise.
9071 (assign_parm_setup_block): Likewise.
9072 (gimplify_parameters): Likewise.
9073 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9074 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9076 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9077 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9078 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9079 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9080 (aarch64_function_arg_padding): ...this new function.
9081 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9082 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9083 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9084 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9085 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9086 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9087 (arm_pad_arg_upward): Replace with...
9088 (arm_function_arg_padding): ...this new function.
9089 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9091 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9092 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9093 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9094 (ia64_hpux_function_arg_padding): Replace with...
9095 (ia64_function_arg_padding): ...this new function. Use pad_direction
9096 instead of direction. Check for TARGET_HPUX.
9097 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9098 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9099 (iq2000_function_arg_padding): New function.
9100 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9101 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9102 (mips_function_arg_padding): ...this new function.
9103 (mips_pad_reg_upward): Update accordingly.
9104 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9105 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9106 targetm.calls.function_arg_padding.
9107 (FUNCTION_ARG_PADDING): Delete.
9108 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9109 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9110 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9111 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9112 (nios2_block_reg_padding): Return pad_direction instead of direction.
9113 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9114 instead of direction.
9115 (nios2_function_arg_padding): Likewise. Make static.
9116 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9117 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9118 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9119 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9120 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9121 (pa_function_arg_padding): Make static. Return pad_direction instead
9123 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9124 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9125 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9126 instead of direction. Use targetm.calls.function_arg_padding.
9127 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9128 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9129 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9130 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9131 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9133 (function_arg_padding): Rename to...
9134 (rs6000_function_arg_padding): ...this. Make static. Return
9135 pad_direction instead of direction.
9136 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9137 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9138 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9139 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9140 instead of direction. Use targetm.calls.function_arg_padding.
9141 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9142 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9143 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9144 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9145 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9146 (function_arg_padding): Rename to...
9147 (rs6000_function_arg_padding): ...this. Make static. Return
9148 pad_direction instead of direction.
9149 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9150 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9151 * config/s390/s390.c (s390_function_arg_padding): New function.
9152 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9153 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9154 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
9155 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9156 (function_arg_padding): Rename to...
9157 (sparc_function_arg_padding): ...this. Make static. Return
9158 pad_direction instead of direction.
9159 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
9160 * config/spu/spu.c (spu_function_arg_padding): New function.
9161 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9162 * system.h (FUNCTION_ARG_PADDING): Poison.
9164 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9165 Alan Hayward <alan.hayward@arm.com>
9166 David Sherwood <david.sherwood@arm.com>
9168 * target.def (modes_tieable_p): New hook.
9169 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
9170 (TARGET_MODES_TIEABLE_P): ...this.
9171 * doc/tm.texi.in: Regenerate.
9172 * hooks.h (hook_bool_mode_mode_true): Declare.
9173 * hooks.c (hook_bool_mode_mode_true): New function.
9174 * combine.c (subst): Use targetm.modes_tieable_p instead of
9176 * dse.c (find_shift_sequence): Likewise.
9177 * expmed.c (extract_low_bits): Likewise.
9178 * lower-subreg.c: Include target.h.
9179 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
9181 * rtlanal.c (rtx_cost): Likewise.
9182 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
9183 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
9184 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
9185 (TARGET_MODES_TIEABLE_P): Redefine.
9186 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
9187 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
9188 (TARGET_MODES_TIEABLE_P): Redefine.
9189 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
9190 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
9191 (arc_modes_tieable_p): New function.
9192 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
9193 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
9194 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
9195 (arm_modes_tieable_p): Make static.
9196 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
9197 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
9198 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
9199 (TARGET_MODES_TIEABLE_P): Redefine.
9200 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
9201 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
9202 (TARGET_MODES_TIEABLE_P): Redefine.
9203 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
9204 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
9205 (cr16_modes_tieable_p): New function.
9206 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
9207 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
9208 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
9209 (TRULY_NOOP_TRUNCATION): Update comment.
9210 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
9211 (TRULY_NOOP_TRUNCATION): Update comment.
9212 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
9213 (frv_modes_tieable_p): New function.
9214 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
9215 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
9216 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
9217 (TARGET_MODES_TIEABLE_P): Redefine.
9218 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9219 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9220 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9221 (TARGET_MODES_TIEABLE_P): Redefine.
9222 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9223 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9224 (ia64_modes_tieable_p): New function.
9225 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9226 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9227 (iq2000_modes_tieable_p): New function.
9228 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9229 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9230 (lm32_modes_tieable_p): New function.
9231 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9232 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9233 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9234 (TARGET_MODES_TIEABLE_P): Redefine.
9235 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9236 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9237 (m32r_modes_tieable_p): New function.
9238 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9239 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9240 (m68k_modes_tieable_p): New function.
9241 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9242 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9243 (mcore_modes_tieable_p): New function.
9244 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9245 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9247 (TARGET_MODES_TIEABLE_P): Redefine.
9248 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9249 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9250 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9251 (TARGET_MODES_TIEABLE_P): Redefine.
9252 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9253 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9254 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9255 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9256 (mn10300_modes_tieable_p): ...this and make static.
9257 (TARGET_MODES_TIEABLE_P): Redefine.
9258 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9259 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9260 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9261 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9262 (msp430_modes_tieable_p): Make static.
9263 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9264 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9265 (TARGET_MODES_TIEABLE_P): Redefine.
9266 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9267 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9268 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9269 (TARGET_MODES_TIEABLE_P): Redefine.
9270 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9271 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9272 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9273 (TARGET_MODES_TIEABLE_P): Redefine.
9274 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9275 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9276 (pdp11_modes_tieable_p): New function.
9277 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9278 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9279 (rs6000_modes_tieable_p): New function.
9280 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9281 * config/powerpcspe/powerpcspe.md: Update comment.
9282 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9283 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9284 (TARGET_MODES_TIEABLE_P): Redefine.
9285 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9286 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9287 (rl78_modes_tieable_p): New function.
9288 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9289 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9290 (rs6000_modes_tieable_p): New function.
9291 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9292 * config/rs6000/rs6000.md: Update comment.
9293 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9294 * config/rx/rx.c (rx_modes_tieable_p): New function.
9295 (TARGET_MODES_TIEABLE_P): Redefine.
9296 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9297 * config/s390/s390.c (s390_modes_tieable_p): New function.
9298 (TARGET_MODES_TIEABLE_P): Redefine.
9299 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9300 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9301 (sh_modes_tieable_p): New function.
9302 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9303 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9304 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9305 (sparc_modes_tieable_p): Make static.
9306 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9307 * config/spu/spu.c (spu_modes_tieable_p): New function.
9308 (TARGET_MODES_TIEABLE_P): Redefine.
9309 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9310 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9311 (TARGET_MODES_TIEABLE_P): Redefine.
9312 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9313 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9314 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9315 * config/v850/v850.c (v850_modes_tieable_p): New function.
9316 (TARGET_MODES_TIEABLE_P): Redefine.
9317 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9318 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9319 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9320 (visium_modes_tieable_p): New function.
9321 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9322 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9323 (xtensa_modes_tieable_p): New function.
9324 * system.h (MODES_TIEABLE_P): Poison.
9326 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9327 Alan Hayward <alan.hayward@arm.com>
9328 David Sherwood <david.sherwood@arm.com>
9330 * target.def (hard_regno_mode_ok): New hook.
9331 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9332 (TARGET_HARD_REGNO_MODE_OK): ...this.
9333 * doc/tm.texi.in: Regenerate.
9334 * hooks.h (hook_bool_uint_mode_true): Declare.
9335 * hooks.c (hook_bool_uint_mode_true): New function.
9336 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9338 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9339 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9340 instead of HARD_REGNO_MODE_OK.
9341 * caller-save.c: Include target.h.
9342 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9344 * combine.c (can_combine_p): Likewise.
9345 (combinable_i3pat): Likewise.
9346 (can_change_dest_mode): Likewise.
9347 * expr.c (init_expr_target): Likewise.
9348 (convert_move): Likewise.
9349 (convert_modes): Likewise.
9350 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9351 (setup_prohibited_mode_move_regs): Likewise.
9352 * ira.h (target_ira): Likewise.
9353 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9354 * lra-constraints.c (process_alt_operands): Likewise.
9355 (split_reg): Likewise.
9356 * recog.c (peep2_find_free_register): Likewise.
9357 * ree.c (combine_reaching_defs): Likewise.
9358 * regcprop.c (maybe_mode_change): Likewise.
9359 * reginfo.c (init_reg_sets_1): Likewise.
9360 (choose_hard_reg_mode): Likewise.
9361 (simplifiable_subregs): Likewise.
9362 * regrename.c (check_new_reg_p): Likewise.
9363 * reload.c (find_valid_class): Likewise.
9364 (find_valid_class_1): Likewise.
9365 (reload_inner_reg_of_subreg): Likewise.
9366 (push_reload): Likewise.
9367 (combine_reloads): Likewise.
9368 (find_dummy_reload): Likewise.
9369 (find_reloads): Likewise.
9370 * reload1.c (find_reg): Likewise.
9371 (set_reload_reg): Likewise.
9372 (allocate_reload_reg): Likewise.
9373 (choose_reload_regs): Likewise.
9374 (reload_adjust_reg_for_temp): Likewise.
9375 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9376 (simplify_subreg_regno): Likewise.
9377 * sel-sched.c (init_regs_for_mode): Likewise.
9378 * varasm.c (make_decl_rtl): Likewise.
9379 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9380 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9382 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9383 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9384 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9385 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9386 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9387 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9388 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9389 (arc_mode_class): Delete.
9390 (HARD_REGNO_MODE_OK): Delete.
9391 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9392 (arc_hard_regno_mode_ok): Rename old array to...
9393 (arc_hard_regno_mode_ok_modes): ...this.
9394 (arc_conditional_register_usage): Update accordingly.
9395 (arc_mode_class): Make static.
9396 (arc_hard_regno_mode_ok): New function.
9397 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9398 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9399 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9400 (arm_hard_regno_mode_ok): Make static.
9401 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9403 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9404 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9405 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9407 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9408 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9409 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9410 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9411 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9412 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9413 * config/bfin/predicates.md (valid_reg_operand): Use
9414 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9415 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9416 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9417 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9418 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9419 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9420 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9421 (cr16_hard_regno_mode_ok): Make static and return a bool.
9422 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9423 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9424 (cris_hard_regno_mode_ok): New function.
9425 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9426 (epiphany_mode_class): Delete.
9427 (HARD_REGNO_MODE_OK): Delete.
9428 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9429 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9430 (hard_regno_mode_ok): Rename to...
9431 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9432 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9433 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9435 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9436 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9437 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9438 (frv_hard_regno_mode_ok): Make static and return a bool.
9439 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9441 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9442 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9443 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9444 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9446 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9447 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9448 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9449 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9451 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9452 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9453 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9454 (ia64_hard_regno_mode_ok): New function.
9455 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9456 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9457 (iq2000_hard_regno_mode_ok): New function.
9458 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9459 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9460 (lm32_hard_regno_mode_ok): New function.
9461 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9462 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9463 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9464 instead of HARD_REGNO_MODE_OK.
9465 (m32c_hard_regno_ok): Rename to...
9466 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9467 (m32c_cannot_change_mode_class): Update accordingly.
9468 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9469 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9470 (m32r_mode_class): Delete.
9471 (HARD_REGNO_MODE_OK): Delete.
9472 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9473 (m32r_hard_regno_mode_ok): Rename to...
9474 (m32r_hard_regno_modes): ...this.
9475 (m32r_mode_class): Make static.
9476 (m32r_hard_regno_mode_ok): New function.
9477 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9478 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9479 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9480 (m68k_hard_regno_mode_ok): Make static.
9481 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9482 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9483 (mcore_hard_regno_mode_ok): New function.
9484 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9485 (HARD_REGNO_MODE_OK): Delete.
9486 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9488 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9489 (microblaze_hard_regno_mode_ok): New function.
9490 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9491 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9492 (mips_hard_regno_mode_ok): Delete.
9493 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9494 (mips_hard_regno_mode_ok_p): ...this and make static.
9495 (mips_hard_regno_mode_ok_p): Rename to...
9496 (mips_hard_regno_mode_ok_uncached): ...this.
9497 (mips_hard_regno_mode_ok): New function.
9498 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9499 of HARD_REGNO_MODE_OK.
9500 (mips_option_override): Update after above name changes.
9501 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9502 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9503 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9504 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9505 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9506 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9507 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9508 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9509 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9510 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9511 (msp430_hard_regno_mode_ok): Make static and return a bool.
9512 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9513 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9514 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9516 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9517 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9518 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9519 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9520 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9521 (PA_HARD_REGNO_MODE_OK): ...this
9522 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9523 (PA_HARD_REGNO_MODE_OK): ...this.
9524 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9525 (pa_hard_regno_mode_ok): New function.
9526 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9527 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9528 (pdp11_hard_regno_mode_ok): New function.
9529 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9530 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9532 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9534 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9535 (rs6000_hard_regno_mode_ok): Rename to...
9536 (rs6000_hard_regno_mode_ok_uncached): ...this.
9537 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9538 (rs6000_hard_regno_mode_ok): New function.
9539 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9540 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9541 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9542 (riscv_hard_regno_mode_ok): ...this and make static.
9543 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9544 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9545 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9546 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9547 (rl78_hard_regno_mode_ok): Make static and return bool.
9548 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9549 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9551 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
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/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9558 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9559 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9560 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9561 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9562 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9563 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9564 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9565 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9566 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9567 (sh_hard_regno_mode_ok): Make static.
9568 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9569 instead of HARD_REGNO_MODE_OK.
9570 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9571 (sparc_mode_class): Delete.
9572 (HARD_REGNO_MODE_OK): Delete.
9573 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9574 (hard_regno_mode_classes): Make static.
9575 (sparc_mode_class): Likewise.
9576 (sparc_hard_regno_mode_ok): New function.
9577 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9578 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9579 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9581 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9582 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9583 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9584 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9585 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9586 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9587 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9588 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9589 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9590 (visium_hard_regno_mode_ok): New function.
9591 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9592 instead of HARD_REGNO_MODE_OK.
9593 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9594 (HARD_REGNO_MODE_OK): Delete.
9595 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9596 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9597 (xtensa_option_override): Update accordingly.
9598 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9599 (xtensa_hard_regno_mode_ok): New function.
9600 * system.h (HARD_REGNO_MODE_OK): Poison.
9602 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9603 Alan Hayward <alan.hayward@arm.com>
9604 David Sherwood <david.sherwood@arm.com>
9606 * target.def (hard_regno_call_part_clobbered): New hook.
9607 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9608 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9609 * doc/tm.texi: Regenerate.
9610 * hooks.h (hook_bool_uint_mode_false): Declare.
9611 * hooks.c (hook_bool_uint_mode_false): New function.
9612 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9613 * cselib.c (cselib_process_insn): Use
9614 targetm.hard_regno_call_part_clobbered instead of
9615 HARD_REGNO_CALL_PART_CLOBBERED.
9616 * ira-conflicts.c (ira_build_conflicts): Likewise.
9617 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9618 * lra-constraints.c (need_for_call_save_p): Likewise.
9619 * lra-lives.c: Include target.h.
9620 (check_pseudos_live_through_calls): Use
9621 targetm.hard_regno_call_part_clobbered instead of
9622 HARD_REGNO_CALL_PART_CLOBBERED.
9623 * regcprop.c: Include target.h.
9624 (copyprop_hardreg_forward_1): Use
9625 targetm.hard_regno_call_part_clobbered instead of
9626 HARD_REGNO_CALL_PART_CLOBBERED.
9627 * reginfo.c (choose_hard_reg_mode): Likewise.
9628 * regrename.c (check_new_reg_p): Likewise.
9629 * reload.c (find_equiv_reg): Likewise.
9630 * reload1.c (emit_reload_insns): Likewise.
9631 * sched-deps.c (deps_analyze_insn): Likewise.
9632 * sel-sched.c (init_regs_for_mode): Likewise.
9633 (mark_unavailable_hard_regs): Likewise.
9634 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9635 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9636 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9638 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9639 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9640 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9642 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9644 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9645 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9646 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9648 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9649 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9650 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9652 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9653 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
9655 * config/powerpcspe/powerpcspe.c
9656 (rs6000_hard_regno_call_part_clobbered): New function.
9657 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9658 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9659 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
9661 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9662 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9663 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
9665 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9666 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9667 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
9669 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9670 Alan Hayward <alan.hayward@arm.com>
9671 David Sherwood <david.sherwood@arm.com>
9673 * rtl.h (subreg_memory_offset): Declare.
9674 * emit-rtl.c (subreg_memory_offset): New function.
9675 * expmed.c (store_bit_field_1): Use it.
9676 * expr.c (undefined_operand_subword_p): Likewise.
9677 * simplify-rtx.c (simplify_subreg): Likewise.
9679 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9681 PR rtl-optimization/57448
9684 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9685 using atomic_load pattern.
9686 (expand_atomic_store): Likewise.
9688 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9691 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9692 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9693 handling. Use replace_call_with_value with NULL instead of
9694 gsi_replace, unlink_stmt_vdef and release_defs.
9696 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9699 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9701 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9704 * rtl.h (emit_library_call_value_1): Declare.
9705 (emit_library_call): Replace declaration with a series of overloads.
9706 Remove the parameter count argument.
9707 (emit_library_call_value): Likewise.
9708 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9709 with an "rtx_mode_t *".
9710 (emit_library_call_value): Delete.
9711 (emit_library_call): Likewise.
9712 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9713 (asan_emit_allocas_unpoison): Likewise.
9714 * builtins.c (expand_builtin_powi): Likewise.
9715 (expand_asan_emit_allocas_unpoison): Likewise.
9716 * cfgexpand.c (expand_main_function): Likewise.
9717 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9718 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9719 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9720 * config/arm/arm.c (arm_trampoline_init): Likewise.
9721 (arm_call_tls_get_addr): Likewise.
9722 (arm_expand_divmod_libfunc): Likewise.
9723 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9724 (smulsi3_highpart): Likewise.
9725 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9726 (c6x_expand_compare): Likewise.
9727 (c6x_expand_movmem): Likewise.
9728 * config/frv/frv.c (frv_trampoline_init): Likewise.
9729 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9730 (ix86_expand_divmod_libfunc): Likewise.
9731 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9732 (ia64_expand_compare): Likewise.
9733 (ia64_profile_hook): Likewise.
9734 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9735 (nonlocal_goto): Likewise.
9736 (restore_stack_nonlocal): Likewise.
9737 * config/m32r/m32r.c (block_move_call): Likewise.
9738 (m32r_trampoline_init): Likewise.
9739 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9740 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9741 (m68k_call_m68k_read_tp): Likewise.
9742 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9743 (microblaze_expand_divide): Likewise.
9744 * config/mips/mips.h (mips_args): Likewise.
9745 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9746 (MIPS_ICACHE_SYNC): Likewise.
9747 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9748 (nios2_trampoline_init): Likewise.
9749 * config/pa/pa.c (hppa_tls_call): Likewise.
9750 (pa_trampoline_init): Likewise.
9751 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9752 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9753 (expand_strn_compare): Likewise.
9754 (rs6000_generate_compare): Likewise.
9755 (rs6000_expand_float128_convert): Likewise.
9756 (output_profile_hook): Likewise.
9757 (rs6000_trampoline_init): Likewise.
9758 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9759 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9760 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9761 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9762 (rs6000_generate_compare): Likewise.
9763 (rs6000_expand_float128_convert): Likewise.
9764 (output_profile_hook): Likewise.
9765 (rs6000_trampoline_init): Likewise.
9766 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9767 * config/sh/sh.c (sh_trampoline_init): Likewise.
9768 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9769 (sparc_emit_float_lib_cmp): Likewise.
9770 (sparc32_initialize_trampoline): Likewise.
9771 (sparc64_initialize_trampoline): Likewise.
9772 (sparc_profile_hook): Likewise.
9773 * config/spu/spu.c (ea_load_store): Likewise.
9774 * config/spu/spu.md (floatunssidf2): Likewise.
9775 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9776 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9777 * config/visium/visium.c (expand_block_move_4): Likewise.
9778 (expand_block_move_2): Likewise.
9779 (expand_block_move_1): Likewise.
9780 (expand_block_set_4): Likewise.
9781 (expand_block_set_2): Likewise.
9782 (expand_block_set_1): Likewise.
9783 (visium_trampoline_init): Likewise.
9784 (visium_profile_hook): Likewise.
9785 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9786 (xtensa_setup_frame_addresses): Likewise.
9787 (xtensa_trampoline_init): Likewise.
9788 * except.c (sjlj_emit_function_enter): Likewise.
9789 (sjlj_emit_function_exit): Likewise.
9790 * explow.c (allocate_dynamic_stack_space): Likewise.
9791 (probe_stack_range): Likewise.
9792 * expr.c (convert_mode_scalar): Likewise.
9793 * optabs.c (expand_binop): Likewise.
9794 (expand_twoval_binop_libfunc): Likewise.
9795 (expand_unop): Likewise.
9796 (prepare_cmp_insn): Likewise.
9797 (prepare_float_lib_cmp): Likewise.
9798 (expand_float): Likewise.
9799 (expand_fix): Likewise.
9800 (expand_fixed_convert): Likewise.
9801 (maybe_emit_sync_lock_test_and_set): Likewise.
9802 (expand_atomic_compare_and_swap): Likewise.
9803 (expand_mem_thread_fence): Likewise.
9804 (expand_atomic_fetch_op): Likewise.
9806 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9808 * doc/generic.texi (OpenACC): Adjust URL.
9809 * doc/invoke.texi (C Dialect Options): Ditto.
9811 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9813 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9814 predicate for operand 1. Add (m,<S>) constraint.
9815 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9816 Prevent memory operand 1 with register operand 2.
9818 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9820 PR rtl-optimization/82024
9821 * combine.c (try_combine): If the combination result is a PARALLEL,
9822 and we only need to retain the SET in there that would be placed
9823 at I2, check that we can place that at I3 instead, before doing so.
9825 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9828 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9830 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9831 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9834 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9835 Jeff Law <law@redhat.com>
9837 * varasm.c (bss_initializer_p): Do not put constants into .bss
9838 (categorize_decl_for_section): Handle bss_initializer_p returning
9839 false when DECL_INITIAL is NULL.
9841 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9844 * config/s390/s390.c (s390_can_inline_p): New function.
9846 2017-09-01 Jeff Law <law@redhat.com>
9848 PR tree-optimization/82052
9849 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9850 Always initialize the returned slot after a hash table miss
9851 when INSERT is true.
9853 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9855 * config/s390/s390.md (mem_signal_fence): Remove.
9856 * doc/md.texi (mem_signal_fence): Remove.
9857 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9859 * target-insns.def (mem_signal_fence): Remove.
9861 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9864 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9867 * asan.c (create_odr_indicator): Strip name encoding from assembler
9868 name before appending it after __odr_asan_.
9870 2017-09-01 Martin Liska <mliska@suse.cz>
9872 PR tree-optimization/82059
9873 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9874 frequency only when an edge is redirected.
9876 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9878 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9879 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9880 (arc_conditional_register_usage): Remove ARC600 lp_count
9882 (arc_file_start): Emit Tag_ARC_CPU_variation.
9883 (arc_can_use_doloop_p): New conditions to use ZOLs.
9884 (hwloop_fail): New function.
9885 (hwloop_optimize): Likewise.
9886 (hwloop_pattern_reg): Likewise.
9887 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9888 (arc_reorg_loops): New function, calls reorg_loops.
9889 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9890 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9892 (arc_loop_hazard): Remove function, functionality moved into
9894 (arc_hazard): Remove arc_loop_hazard call.
9895 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9896 into hwloop_optimize.
9897 (arc_label_align): Remove ZOL handling.
9898 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9899 * config/arc/arc.md (doloop_begin): Remove pattern.
9900 (doloop_begin_i): Likewise.
9901 (doloop_end_i): Likewise.
9902 (doloop_fallback): Likewise.
9903 (doloop_fallback_m): Likewise.
9904 (doloop_end): Reimplement expand.
9905 (arc_lp): New pattern for LP instruction.
9906 (loop_end): New pattern.
9907 (loop_fail): Likewise.
9908 (decrement_and_branch_until_zero): Likewise.
9909 * config/arc/arc.opt (mlpc-width): New option.
9910 * doc/invoke.texi (mlpc-width): Document option.
9912 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9914 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9915 (arc_ccfsm_advance): Fix checking for delay slots.
9916 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9918 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9920 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9922 (movhi_insn): Update store instruction constraint which are saving
9923 6-bit short immediates.
9924 (movsi_insn): Consider also short scaled load operations.
9925 (zero_extendhisi2_i): Use Usd constraint instead of T.
9926 (extendhisi2_i): Add q constraint.
9927 (arc_clzsi2): Add type and length attributes.
9928 (arc_ctzsi2): Likewise.
9929 * config/arc/constraints.md (Usc): Update constraint, the
9930 assembler can parse two relocations for a single instruction.
9932 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9934 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9935 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9937 2017-08-31 Olivier Hainque <hainque@adacore.com>
9939 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9940 match as powerpc-wrs-vxworks*.
9942 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9944 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9945 register constraint for by-element operand.
9946 (aarch64_mls_elt_merge<mode>): Likewise.
9948 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9950 * config/arc/arc.c (arc_can_follow_jump): Check for short
9953 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9955 * config.gcc: Use g.opt for arc.
9956 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9957 functionality moved to ...
9958 (legitimate_scaled_address_p): New function, ...here.
9959 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9960 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9961 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9963 (arc_override_options): Handle G option.
9964 (arc_output_pic_addr_const): Correct function definition.
9965 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9966 (arc_decl_anon_ns_mem_p): Delete.
9967 (arc_in_small_data_p): Overhaul this function to take into
9968 consideration the value given via G option.
9969 (arc_rewrite_small_data_1): Renamed and corrected old
9970 arc_rewrite_small_data function.
9971 (arc_rewrite_small_data): New function.
9972 (small_data_pattern): Don't use pic_offset_table_rtx.
9973 * config/arc/arc.h (CC1_SPEC): Recognize G option.
9974 * config/arc/simdext.md (movmisalignv2hi): Use
9975 prepare_move_operands function.
9977 (movmisalign*): Likewise.
9978 * doc/invoke.texi (ARC options): Document -G option.
9980 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9982 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
9984 * config/arc/arc.c (arc_print_operand): Output scalled address for
9985 sdata whenever is possible.
9986 (arc_in_small_data_p): Allow sdata for 64bit datum when double
9987 load/stores are available.
9988 (compact_sda_memory_operand): Check for the alignment required by
9989 code density instructions.
9990 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
9992 * config/arc/constraints.md (Usd): Update constraint.
9993 (Us0): New constraint.
9994 (Usc): Update constraint.
9996 2017-08-31 Richard Biener <rguenther@suse.de>
9999 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10000 function only once.
10002 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10004 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10005 Resize type_signature.
10007 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10008 Alan Hayward <alan.hayward@arm.com>
10009 David Sherwood <david.sherwood@arm.com>
10011 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10012 subregs whose inner modes can be stored in GPRs.
10013 (aarch64_classify_index): Likewise.
10015 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10016 Alan Hayward <alan.hayward@arm.com>
10017 David Sherwood <david.sherwood@arm.com>
10019 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10020 (V_INT_EQUIV): ...this.
10021 (v_cmp_result): Rename to...
10022 (v_int_equiv): ...this.
10023 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10024 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10025 (copysign<mode>3): Likewise.
10026 (aarch64_simd_bsl<mode>_internal): Likewise.
10027 (aarch64_simd_bsl<mode>): Likewise.
10028 (vec_cmp<mode><mode>): Likewise.
10029 (vcond<mode><mode>): Likewise.
10030 (vcond<v_cmp_mixed><mode>): Likewise.
10031 (vcondu<mode><v_cmp_mixed>): Likewise.
10032 (aarch64_cm<optab><mode>): Likewise.
10033 (aarch64_cmtst<mode>): Likewise.
10034 (aarch64_fac<optab><mode>): Likewise.
10035 (vec_perm_const<mode>): Likewise.
10036 (vcond_mask_<mode><v_cmp_result>): Rename to...
10037 (vcond_mask_<mode><v_int_equiv>): ...this.
10038 (vec_cmp<mode><v_cmp_result>): Rename to...
10039 (vec_cmp<mode><v_int_equiv>): ...this.
10041 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10042 Alan Hayward <alan.hayward@arm.com>
10043 David Sherwood <david.sherwood@arm.com>
10045 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
10047 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10048 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10049 (UNSPEC_LD4_DREG): New unspecs.
10050 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10051 (aarch64_ld2<mode>_dreg_be): Replace with...
10052 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10054 (aarch64_ld3<mode>_dreg_le)
10055 (aarch64_ld3<mode>_dreg_be): Replace with...
10056 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10058 (aarch64_ld4<mode>_dreg_le)
10059 (aarch64_ld4<mode>_dreg_be): Replace with...
10060 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10063 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10065 PR tree-optimization/81987
10066 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10067 insert an initializer in a location not dominated by the stride
10070 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10072 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10073 on the entire header of the finally block in the fallthru case.
10075 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10077 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10079 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10081 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10082 rs6000_emit_move_from_cr and call renamed function.
10083 (rs6000_emit_prologue): Call renamed functions.
10084 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10085 movesi_from_cr, remove volatile CRs.
10087 2017-08-30 Jon Beniston <jon@beniston.com>
10088 Richard Biener <rguenther@suse.de>
10090 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10091 of VECTOR_MODE_P check.
10092 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10093 element vector types.
10095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10097 * df.h (df_read_modify_subreg_p): Remove in favor of...
10098 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10099 const_rtx instead of an rtx.
10100 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10101 * df-problems.c (df_word_lr_mark_ref): Likewise.
10102 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10103 (mark_pseudo_reg_dead): Likewise.
10104 (mark_ref_dead): Likewise.
10105 * reginfo.c (init_subregs_of_mode): Likewise.
10106 * sched-deps.c (sched_analyze_1): Likewise.
10107 * df-scan.c (df_def_record_1): Likewise.
10108 (df_uses_record): Likewise.
10109 (df_read_modify_subreg_p): Remove in favor of...
10110 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10111 const_rtx instead of an rtx.
10113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10114 Alan Hayward <alan.hayward@arm.com>
10115 David Sherwood <david.sherwood@arm.com>
10117 * rtl.h (partial_subreg_p): New function.
10118 * caller-save.c (save_call_clobbered_regs): Use it.
10119 * calls.c (expand_call): Likewise.
10120 * combine.c (combinable_i3pat): Likewise.
10121 (simplify_set): Likewise.
10122 (make_extraction): Likewise.
10123 (make_compound_operation_int): Likewise.
10124 (gen_lowpart_or_truncate): Likewise.
10125 (force_to_mode): Likewise.
10126 (make_field_assignment): Likewise.
10127 (reg_truncated_to_mode): Likewise.
10128 (record_truncated_value): Likewise.
10129 (move_deaths): Likewise.
10130 * cse.c (record_jump_cond): Likewise.
10131 (cse_insn): Likewise.
10132 * cselib.c (cselib_lookup_1): Likewise.
10133 * expmed.c (extract_bit_field_using_extv): Likewise.
10134 * function.c (assign_parm_setup_reg): Likewise.
10135 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10136 * ira-build.c (create_insn_allocnos): Likewise.
10137 * lra-coalesce.c (merge_pseudos): Likewise.
10138 * lra-constraints.c (match_reload): Likewise.
10139 (simplify_operand_subreg): Likewise.
10140 (curr_insn_transform): Likewise.
10141 * lra-lives.c (process_bb_lives): Likewise.
10142 * lra.c (new_insn_reg): Likewise.
10143 (lra_substitute_pseudo): Likewise.
10144 * regcprop.c (mode_change_ok): Likewise.
10145 (maybe_mode_change): Likewise.
10146 (copyprop_hardreg_forward_1): Likewise.
10147 * reload.c (push_reload): Likewise.
10148 (find_reloads): Likewise.
10149 (find_reloads_subreg_address): Likewise.
10150 * reload1.c (alter_reg): Likewise.
10151 (eliminate_regs_1): Likewise.
10152 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10154 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
10156 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
10157 back to if statements, including unpack.
10159 2017-08-30 Martin Liska <mliska@suse.cz>
10161 PR inline-asm/82001
10162 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
10164 (func_checker::compare_asm_inputs_outputs): ... this function.
10165 (func_checker::compare_gimple_asm): Use the function to compare
10166 also ASM constrains.
10167 * ipa-icf-gimple.h: Rename the function.
10169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10170 Alan Hayward <alan.hayward@arm.com>
10171 David Sherwood <david.sherwood@arm.com>
10173 * coretypes.h (complex_mode): New type.
10174 * gdbhooks.py (build_pretty_printer): Handle it.
10175 * machmode.h (complex_mode): New class.
10176 (complex_mode::includes_p): New function.
10177 (is_complex_int_mode): Likewise.
10178 (is_complex_float_mode): Likewise.
10179 * genmodes.c (get_mode_class): Handle complex mode classes.
10180 * function.c (expand_function_end): Use is_complex_int_mode.
10182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10183 Alan Hayward <alan.hayward@arm.com>
10184 David Sherwood <david.sherwood@arm.com>
10186 * coretypes.h (scalar_mode_pod): New typedef.
10187 * gdbhooks.py (build_pretty_printer): Handle it.
10188 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
10189 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
10190 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
10191 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
10192 as_a <scalar_mode>.
10194 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10195 Alan Hayward <alan.hayward@arm.com>
10196 David Sherwood <david.sherwood@arm.com>
10198 * machmode.h (mode_for_vector): Take a scalar_mode instead
10200 * stor-layout.c (mode_for_vector): Likewise.
10201 * explow.c (promote_mode): Use as_a <scalar_mode>.
10202 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
10204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10205 Alan Hayward <alan.hayward@arm.com>
10206 David Sherwood <david.sherwood@arm.com>
10208 * target.def (preferred_simd_mode): Take a scalar_mode
10209 instead of a machine_mode.
10210 * targhooks.h (default_preferred_simd_mode): Likewise.
10211 * targhooks.c (default_preferred_simd_mode): Likewise.
10212 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
10213 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
10214 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
10215 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
10216 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
10217 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10218 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10219 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10221 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10222 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10223 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10224 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10225 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10226 * doc/tm.texi: Regenerate.
10227 * optabs-query.c (can_vec_mask_load_store_p): Return false for
10230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10231 Alan Hayward <alan.hayward@arm.com>
10232 David Sherwood <david.sherwood@arm.com>
10234 * target.def (scalar_mode_supported_p): Take a scalar_mode
10235 instead of a machine_mode.
10236 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10237 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10238 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10239 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10240 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10241 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10242 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10243 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10244 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10245 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10246 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10247 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10248 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10249 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10251 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10252 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10253 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10254 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10255 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10257 * doc/tm.texi: Regenerate.
10259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10260 Alan Hayward <alan.hayward@arm.com>
10261 David Sherwood <david.sherwood@arm.com>
10263 * coretypes.h (opt_scalar_mode): New typedef.
10264 * gdbhooks.py (build_pretty_printers): Handle it.
10265 * machmode.h (mode_iterator::get_2xwider): Add overload for
10267 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10269 * expr.c (convert_mode_scalar): Likewise.
10270 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10271 * optabs.c (expand_float): Likewise.
10272 (expand_fix): Likewise.
10273 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10276 Alan Hayward <alan.hayward@arm.com>
10277 David Sherwood <david.sherwood@arm.com>
10279 * optabs.c (expand_float): Explicitly check for scalars before
10280 using a branching expansion.
10281 (expand_fix): Likewise.
10283 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10284 Alan Hayward <alan.hayward@arm.com>
10285 David Sherwood <david.sherwood@arm.com>
10287 * expr.c (convert_mode): Split scalar handling out into...
10288 (convert_mode_scalar): ...this new function. Treat the modes
10291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10292 Alan Hayward <alan.hayward@arm.com>
10293 David Sherwood <david.sherwood@arm.com>
10295 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10297 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10299 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10300 Alan Hayward <alan.hayward@arm.com>
10301 David Sherwood <david.sherwood@arm.com>
10303 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10304 rather than a machine_mode.
10305 (fixed_from_string): Likewise.
10306 (fixed_convert): Likewise.
10307 (fixed_convert_from_int): Likewise.
10308 (fixed_convert_from_real): Likewise.
10309 (real_convert_from_fixed): Likewise.
10310 * fixed-value.c (fixed_from_double_int): Likewise.
10311 (fixed_from_string): Likewise.
10312 (fixed_convert): Likewise.
10313 (fixed_convert_from_int): Likewise.
10314 (fixed_convert_from_real): Likewise.
10315 (real_convert_from_fixed): Likewise.
10316 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10318 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10319 Alan Hayward <alan.hayward@arm.com>
10320 David Sherwood <david.sherwood@arm.com>
10322 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10323 of separate mode class checks. Do not allow vector modes here.
10324 (immed_wide_int_const): Use as_a <scalar_mode>.
10325 * explow.c (trunc_int_for_mode): Likewise.
10326 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10327 (wi::shwi): Likewise.
10328 (wi::min_value): Likewise.
10329 (wi::max_value): Likewise.
10330 * dwarf2out.c (loc_descriptor): Likewise.
10331 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10332 for CONST_WIDE_INT.
10334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10335 Alan Hayward <alan.hayward@arm.com>
10336 David Sherwood <david.sherwood@arm.com>
10338 * tree.h (SCALAR_TYPE_MODE): New macro.
10339 * expr.c (expand_expr_addr_expr_1): Use it.
10340 (expand_expr_real_2): Likewise.
10341 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10342 (fold_convert_const_fixed_from_int): Likewise.
10343 (fold_convert_const_fixed_from_real): Likewise.
10344 (native_encode_fixed): Likewise
10345 (native_encode_complex): Likewise
10346 (native_encode_vector): Likewise.
10347 (native_interpret_fixed): Likewise.
10348 (native_interpret_real): Likewise.
10349 (native_interpret_complex): Likewise.
10350 (native_interpret_vector): Likewise.
10351 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10352 (simd_clone_adjust_argument_types): Likewise.
10353 (simd_clone_init_simd_arrays): Likewise.
10354 (simd_clone_adjust): Likewise.
10355 * stor-layout.c (layout_type): Likewise.
10356 * tree.c (build_minus_one_cst): Likewise.
10357 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10358 * tree-inline.c (estimate_move_cost): Likewise.
10359 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10360 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10361 (vectorizable_reduction): Likewise.
10362 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10363 (vect_recog_mixed_size_cond_pattern): Likewise.
10364 (check_bool_pattern): Likewise.
10365 (adjust_bool_pattern): Likewise.
10366 (search_type_for_mask_1): Likewise.
10367 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10368 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10369 (vectorizable_load): Likewise.
10370 (vectorizable_store): Likewise.
10371 * ubsan.c (ubsan_encode_value): Likewise.
10372 * varasm.c (output_constant): Likewise.
10374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10375 Alan Hayward <alan.hayward@arm.com>
10376 David Sherwood <david.sherwood@arm.com>
10378 * coretypes.h (scalar_mode): New class.
10379 * machmode.h (scalar_mode): Likewise.
10380 (scalar_mode::includes_p): New function.
10381 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10382 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10383 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10384 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10385 * expmed.c (store_bit_field_1): Likewise.
10386 (extract_bit_field_1): Likewise.
10387 * expr.c (write_complex_part): Likewise.
10388 (read_complex_part): Likewise.
10389 (emit_move_complex_push): Likewise.
10390 (expand_expr_real_2): Likewise.
10391 * function.c (assign_parm_setup_reg): Likewise.
10392 (assign_parms_unsplit_complex): Likewise.
10393 * optabs.c (expand_binop): Likewise.
10394 * rtlanal.c (subreg_get_info): Likewise.
10395 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10396 * varasm.c (output_constant_pool_2): Likewise.
10398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10399 Alan Hayward <alan.hayward@arm.com>
10400 David Sherwood <david.sherwood@arm.com>
10402 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10404 (expmed_mult_highpart_optab): Likewise.
10405 (expmed_mult_highpart): Likewise.
10407 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10408 Alan Hayward <alan.hayward@arm.com>
10409 David Sherwood <david.sherwood@arm.com>
10411 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10412 instead of a machine_mode.
10413 (builtin_memset_read_str): Likewise.
10414 * builtins.c (c_readstr): Likewise.
10415 (builtin_memcpy_read_str): Likewise.
10416 (builtin_strncpy_read_str): Likewise.
10417 (builtin_memset_read_str): Likewise.
10418 (builtin_memset_gen_str): Likewise.
10419 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10420 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10421 instead of a machine_mode.
10422 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10424 (make_extraction): Likewise.
10425 (try_widen_shift_mode): Take and return scalar_int_modes instead
10427 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10428 a scalar_int_mode instead of a machine_mode.
10429 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10430 (avr_addr_space_pointer_mode): Likewise.
10431 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10432 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10433 (msp430_unwind_word_mode): Likewise.
10434 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10435 (spu_addr_space_pointer_mode): Likewise.
10436 (spu_addr_space_address_mode): Likewise.
10437 (spu_libgcc_cmp_return_mode): Likewise.
10438 (spu_libgcc_shift_count_mode): Likewise.
10439 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10440 (rl78_addr_space_pointer_mode): Likewise.
10441 (fl78_unwind_word_mode): Likewise.
10442 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10444 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10445 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10446 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10447 (mips_valid_pointer_mode): Likewise.
10448 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10449 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10450 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10452 (ft32_addr_space_address_mode): Likewise.
10453 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10454 scalar_int_mode instead of a machine_mode.
10455 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10457 (m32c_addr_space_address_mode): Likewise.
10458 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10459 (rs6000_eh_return_filter_mode): Likewise.
10460 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10461 (rs6000_eh_return_filter_mode): Likewise.
10462 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10463 (s390_libgcc_shift_count_mode): Likewise.
10464 (s390_unwind_word_mode): Likewise.
10465 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10467 * target.def (mode_rep_extended): Likewise.
10468 (valid_pointer_mode): Likewise.
10469 (addr_space.valid_pointer_mode): Likewise.
10470 (eh_return_filter_mode): Return a scalar_int_mode rather than
10472 (libgcc_cmp_return_mode): Likewise.
10473 (libgcc_shift_count_mode): Likewise.
10474 (unwind_word_mode): Likewise.
10475 (addr_space.pointer_mode): Likewise.
10476 (addr_space.address_mode): Likewise.
10477 * doc/tm.texi: Regenerate.
10478 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10480 (do_jump): Use scalar_int_mode for local variables.
10481 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10482 rather than a machine_mode.
10483 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10484 (scompare_loc_descriptor_wide): Likewise.
10485 (scompare_loc_descriptor_narrow): Likewise.
10486 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10488 * except.c (sjlj_emit_dispatch_table): Likewise.
10489 (expand_builtin_eh_copy_values): Likewise.
10490 * explow.c (convert_memory_address_addr_space_1): Likewise.
10491 Take a scalar_int_mode rather than a machine_mode.
10492 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10493 than a machine_mode.
10494 (memory_address_addr_space): Use scalar_int_mode for local variables.
10495 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10496 rather than a machine_mode.
10497 * expmed.c (mask_rtx): Likewise.
10498 (init_expmed_one_conv): Likewise.
10499 (expand_mult_highpart_adjust): Likewise.
10500 (extract_high_half): Likewise.
10501 (expmed_mult_highpart_optab): Likewise.
10502 (expmed_mult_highpart): Likewise.
10503 (expand_smod_pow2): Likewise.
10504 (expand_sdiv_pow2): Likewise.
10505 (emit_store_flag_int): Likewise.
10506 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10508 (extract_low_bits): Likewise.
10509 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10511 * expr.c (pieces_addr::adjust): Likewise.
10512 (can_store_by_pieces): Likewise.
10513 (store_by_pieces): Likewise.
10514 (clear_by_pieces_1): Likewise.
10515 (expand_expr_addr_expr_1): Likewise.
10516 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10517 (expand_expr_real_1): Likewise.
10518 (try_casesi): Likewise.
10519 * final.c (shorten_branches): Likewise.
10520 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10521 type of "mode" to machine_mode.
10522 * internal-fn.c (expand_arith_overflow_result_store): Take a
10523 scalar_int_mode rather than a machine_mode.
10524 (expand_mul_overflow): Use scalar_int_mode for local variables.
10525 * loop-doloop.c (doloop_modify): Likewise.
10526 (doloop_optimize): Likewise.
10527 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10528 than a machine_mode.
10529 (expand_doubleword_shift_condmove): Likewise.
10530 (expand_doubleword_shift): Likewise.
10531 (expand_doubleword_clz): Likewise.
10532 (expand_doubleword_popcount): Likewise.
10533 (expand_doubleword_parity): Likewise.
10534 (expand_absneg_bit): Use scalar_int_mode for local variables.
10535 (prepare_float_lib_cmp): Likewise.
10536 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10537 rather than a machine_mode.
10538 (convert_memory_address_addr_space): Likewise.
10539 (get_mode_bounds): Likewise.
10540 (get_address_mode): Return a scalar_int_mode rather than a
10542 * rtlanal.c (get_address_mode): Likewise.
10543 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10544 than a machine_mode.
10545 * targhooks.c (default_mode_rep_extended): Likewise.
10546 (default_valid_pointer_mode): Likewise.
10547 (default_addr_space_valid_pointer_mode): Likewise.
10548 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10549 than a machine_mode.
10550 (default_libgcc_cmp_return_mode): Likewise.
10551 (default_libgcc_shift_count_mode): Likewise.
10552 (default_unwind_word_mode): Likewise.
10553 (default_addr_space_pointer_mode): Likewise.
10554 (default_addr_space_address_mode): Likewise.
10555 * targhooks.h (default_eh_return_filter_mode): Likewise.
10556 (default_libgcc_cmp_return_mode): Likewise.
10557 (default_libgcc_shift_count_mode): Likewise.
10558 (default_unwind_word_mode): Likewise.
10559 (default_addr_space_pointer_mode): Likewise.
10560 (default_addr_space_address_mode): Likewise.
10561 (default_mode_rep_extended): Take a scalar_int_mode rather than
10563 (default_valid_pointer_mode): Likewise.
10564 (default_addr_space_valid_pointer_mode): Likewise.
10565 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10567 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10568 rather than a machine_mode.
10569 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10570 for local variables.
10571 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10572 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10573 than a machine_mode.
10575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10576 Alan Hayward <alan.hayward@arm.com>
10577 David Sherwood <david.sherwood@arm.com>
10579 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10580 the mode argument to scalar_int_mode.
10581 (do_jump_by_parts_zero_rtx): Likewise.
10582 (do_jump_by_parts_equality_rtx): Likewise.
10583 (do_jump_by_parts_greater): Take a mode argument.
10584 (do_jump_by_parts_equality): Likewise.
10585 (do_jump_1): Update calls accordingly.
10587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10588 Alan Hayward <alan.hayward@arm.com>
10589 David Sherwood <david.sherwood@arm.com>
10591 * is-a.h (safe_dyn_cast): New function.
10592 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10593 (jump_table_for_label): Likewise.
10594 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10595 instead of an rtx_insn *.
10596 (shorten_branches): Use dyn_cast instead of LABEL_P and
10597 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10598 rtx_jump_table_data::get_data_mode.
10599 (final_scan_insn): Likewise.
10601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10602 Alan Hayward <alan.hayward@arm.com>
10603 David Sherwood <david.sherwood@arm.com>
10605 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10606 trying to combine a full-register integer set with a subreg
10609 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10610 Alan Hayward <alan.hayward@arm.com>
10611 David Sherwood <david.sherwood@arm.com>
10613 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10614 that is always either address_mode or pointer_mode.
10616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10617 Alan Hayward <alan.hayward@arm.com>
10618 David Sherwood <david.sherwood@arm.com>
10620 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10621 when the two are known to be equal.
10623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10624 Alan Hayward <alan.hayward@arm.com>
10625 David Sherwood <david.sherwood@arm.com>
10627 * simplify-rtx.c (simplify_const_unary_operation): Use
10628 is_a <scalar_int_mode> instead of checking for a nonzero
10629 precision. Forcibly convert op_mode to a scalar_int_mode
10630 in that case. More clearly differentiate the operand and
10631 result modes and use the former when deciding what the value
10632 of a count-bits operation should be. Use is_int_mode instead
10633 of checking for a MODE_INT. Remove redundant check for whether
10634 this mode has a zero precision.
10636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10637 Alan Hayward <alan.hayward@arm.com>
10638 David Sherwood <david.sherwood@arm.com>
10640 * optabs.c (widen_leading): Change the type of the mode argument
10641 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10642 (widen_bswap): Likewise.
10643 (expand_parity): Likewise.
10644 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10645 (expand_ffs): Likewise.
10646 (epand_unop): Check for scalar integer modes before calling the
10649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10650 Alan Hayward <alan.hayward@arm.com>
10651 David Sherwood <david.sherwood@arm.com>
10653 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
10655 (expand_expr_real_1): Update call accordingly.
10657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10658 Alan Hayward <alan.hayward@arm.com>
10659 David Sherwood <david.sherwood@arm.com>
10661 * expmed.c (store_bit_field_using_insv): Add op0_mode and
10662 value_mode arguments. Use scalar_int_mode internally.
10663 (store_bit_field_1): Rename the new integer mode from imode
10664 to op0_mode and use it instead of GET_MODE (op0). Update calls
10665 to store_split_bit_field, store_bit_field_using_insv and
10666 store_fixed_bit_field.
10667 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
10668 Use scalar_int_mode internally. Use a bit count rather than a mode
10669 when calculating the largest bit size for get_best_mode.
10670 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10671 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10672 Remove assertion that OP0 has a scalar integer mode.
10673 (store_split_bit_field): Add op0_mode and value_mode arguments.
10674 Update calls to extract_fixed_bit_field.
10675 (extract_bit_field_using_extv): Add an op0_mode argument.
10676 Use scalar_int_mode internally.
10677 (extract_bit_field_1): Rename the new integer mode from imode to
10678 op0_mode and use it instead of GET_MODE (op0). Update calls to
10679 extract_split_bit_field, extract_bit_field_using_extv and
10680 extract_fixed_bit_field.
10681 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10682 to extract_split_bit_field and extract_fixed_bit_field_1.
10683 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10684 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10685 on the target mode.
10686 (extract_split_bit_field): Add an op0_mode argument. Update call
10687 to extract_fixed_bit_field.
10689 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10690 Alan Hayward <alan.hayward@arm.com>
10691 David Sherwood <david.sherwood@arm.com>
10693 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10694 * explow.c (hard_function_value): Likewise.
10695 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10696 convert_to_mode call outside the loop.
10697 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10698 for the mode iterator. Require the mode specified by max_pieces
10700 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10702 (copy_blkmode_to_reg): Likewise.
10703 (set_storage_via_setmem): Likewise.
10704 * optabs.c (prepare_cmp_insn): Likewise.
10705 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10706 * stor-layout.c (finish_bitfield_representative): Likewise.
10708 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10709 Alan Hayward <alan.hayward@arm.com>
10710 David Sherwood <david.sherwood@arm.com>
10712 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10713 * expr.c (convert_move): Use them.
10714 (convert_modes): Likewise.
10715 (store_expr_with_bounds): Likewise.
10717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10718 Alan Hayward <alan.hayward@arm.com>
10719 David Sherwood <david.sherwood@arm.com>
10721 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10722 parameter for the mode of "x". Remove the "known_x", "known_mode"
10723 and "known_ret" arguments. Change the type of the mode argument
10724 to scalar_int_mode.
10725 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10726 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10727 (reg_num_sign_bit_copies_for_combine): Likewise.
10728 * rtlanal.c (nonzero_bits1): Likewise.
10729 (num_sign_bit_copies1): Likewise.
10730 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10731 (reg_num_sign_bit_copies_general): Likewise.
10732 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10733 (reg_nonzero_bits_general): 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 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10740 than in subroutines. Return 1 for non-integer modes.
10741 (cached_num_sign_bit_copies): Change the type of the mode parameter
10742 to scalar_int_mode.
10743 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10744 classes. Handle CONST_INT_P first and then check whether X also
10745 has a scalar integer mode. Check the same thing for inner registers
10746 of a SUBREG and for values that are being extended or truncated.
10748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10749 Alan Hayward <alan.hayward@arm.com>
10750 David Sherwood <david.sherwood@arm.com>
10752 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10753 in subroutines. Return the mode mask for non-integer modes.
10754 (cached_nonzero_bits): Change the type of the mode parameter
10755 to scalar_int_mode.
10756 (nonzero_bits1): Likewise. Remove early exit for other mode
10757 classes. Handle CONST_INT_P first and then check whether X
10758 also has a scalar integer mode.
10760 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10761 Alan Hayward <alan.hayward@arm.com>
10762 David Sherwood <david.sherwood@arm.com>
10764 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10765 Return a scalar_int_mode and assert that the size is greater than
10767 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10768 (op_by_pieces_d::op_by_pieces_d): Likewise.
10769 (op_by_pieces_d::run): Likewise.
10770 (can_store_by_pieces): Likewise.
10772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10773 Alan Hayward <alan.hayward@arm.com>
10774 David Sherwood <david.sherwood@arm.com>
10776 * combine.c (extract_left_shift): Add a mode argument and update
10778 (make_compound_operation_int): Change the type of the mode parameter
10779 to scalar_int_mode and update the call to extract_left_shift.
10781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10782 Alan Hayward <alan.hayward@arm.com>
10783 David Sherwood <david.sherwood@arm.com>
10785 * combine.c (simplify_and_const_int): Change the type of the mode
10786 parameter to scalar_int_mode.
10787 (simplify_and_const_int_1): Likewise. Update recursive call.
10789 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10790 Alan Hayward <alan.hayward@arm.com>
10791 David Sherwood <david.sherwood@arm.com>
10793 * combine.c (simplify_compare_const): Check that the mode is a
10794 scalar_int_mode (rather than VOIDmode) before testing its
10796 (simplify_comparison): Move COMPARISON_P handling out of the
10797 loop and restrict the latter part of the loop to scalar_int_modes.
10798 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10799 and when considering SUBREG_REGs. Use is_int_mode instead of
10800 checking GET_MODE_CLASS against MODE_INT.
10802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10803 Alan Hayward <alan.hayward@arm.com>
10804 David Sherwood <david.sherwood@arm.com>
10806 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10807 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10808 shift_unit_mode and for modes involved in scalar shifts.
10810 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10811 Alan Hayward <alan.hayward@arm.com>
10812 David Sherwood <david.sherwood@arm.com>
10814 * combine.c (force_int_to_mode): New function, split out from...
10815 (force_to_mode): ...here. Keep xmode up-to-date and use it
10816 instead of GET_MODE (x).
10818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10819 Alan Hayward <alan.hayward@arm.com>
10820 David Sherwood <david.sherwood@arm.com>
10822 * optabs-query.h (extraction_insn::struct_mode): Change type to
10823 opt_scalar_int_mode and update comment.
10824 (extraction_insn::field_mode): Change type to scalar_int_mode.
10825 (extraction_insn::pos_mode): Likewise.
10826 * combine.c (make_extraction): Update accordingly.
10827 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10828 (get_optab_extraction_insn): Likewise.
10829 * recog.c (simplify_while_replacing): Likewise.
10830 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10831 parameter to opt_scalar_int_mode.
10833 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10834 Alan Hayward <alan.hayward@arm.com>
10835 David Sherwood <david.sherwood@arm.com>
10837 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10838 to a scalar_int_mode instead of a machine_mode.
10839 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10840 (get_best_mode): Return a boolean and use a pointer argument to store
10841 the selected mode. Replace the limit mode parameter with a bit limit.
10842 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10843 for the values returned by bit_field_mode_iterator::next_mode.
10844 (store_bit_field): Update call to get_best_mode.
10845 (store_fixed_bit_field): Likewise.
10846 (extract_fixed_bit_field): Likewise.
10847 * expr.c (optimize_bitfield_assignment_op): Likewise.
10848 * fold-const.c (optimize_bit_field_compare): Likewise.
10849 (fold_truth_andor_1): Likewise.
10850 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10851 Update for new type of m_mode.
10852 (get_best_mode): As above.
10854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10855 Alan Hayward <alan.hayward@arm.com>
10856 David Sherwood <david.sherwood@arm.com>
10858 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10859 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10860 (store_bit_field): Check is_a <scalar_int_mode> before calling
10861 strict_volatile_bitfield_p.
10862 (extract_bit_field): Likewise.
10864 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10865 Alan Hayward <alan.hayward@arm.com>
10866 David Sherwood <david.sherwood@arm.com>
10868 * target.def (cstore_mode): Return a scalar_int_mode.
10869 * doc/tm.texi: Regenerate.
10870 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10871 * targhooks.h (default_cstore_mode): Likewise.
10872 * targhooks.c (default_cstore_mode): Likewise, using a forced
10874 * expmed.c (emit_cstore): Expect the target of the cstore to be
10877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10878 Alan Hayward <alan.hayward@arm.com>
10879 David Sherwood <david.sherwood@arm.com>
10881 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10883 (niter_desc): Likewise mode.
10884 (iv_analyze): Add a mode parameter.
10886 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10887 and change its type to scalar_int_mode.
10888 * loop-iv.c: Update commentary at head of file.
10889 (iv_constant): Pass the mode paraeter before the rtx it describes
10890 and change its type to scalar_int_mode. Remove VOIDmode handling.
10891 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10892 (iv_extend): Likewise.
10893 (shorten_into_mode): Likewise.
10894 (iv_add): Use scalar_int_mode.
10895 (iv_mult): Likewise.
10896 (iv_shift): Likewise.
10897 (canonicalize_iv_subregs): Likewise.
10898 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10899 it describes and change its mode to scalar_int_mode. Also change
10900 the type of the returned inner_mode to scalar_int_mode.
10901 (get_biv_step): Likewise, turning outer_mode from a pointer
10902 into a direct parameter. Update call to get_biv_step_1.
10903 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10904 iv_constant and get_biv_step.
10905 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10906 and change its type to scalar_int_mode. Don't initialise iv->mode
10907 to VOIDmode and remove later checks for its still being VOIDmode.
10908 Update calls to iv_analyze_op and iv_analyze_expr. Check
10909 is_a <scalar_int_mode> when changing the mode under consideration.
10910 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10911 Update call to iv_analyze_expr.
10912 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10913 inner register is not also a scalar_int_mode. Update call to
10915 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10916 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10917 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10918 separate mode class checks. Update calls to iv_analyze. Remove
10919 fix-up of VOIDmodes after iv_analyze_biv.
10920 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10921 don't have a scalar_int_mode. Update call to biv_p.
10923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10924 Alan Hayward <alan.hayward@arm.com>
10925 David Sherwood <david.sherwood@arm.com>
10927 * cfgexpand.c (convert_debug_memory_address): Use
10928 as_a <scalar_int_mode>.
10929 * combine.c (expand_compound_operation): Likewise.
10930 (make_extraction): Likewise.
10931 (change_zero_ext): Likewise.
10932 (simplify_comparison): Likewise.
10933 * cse.c (cse_insn): Likewise.
10934 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10935 (mem_loc_descriptor): Likewise.
10936 (loc_descriptor): Likewise.
10937 * expmed.c (init_expmed_one_mode): Likewise.
10938 (synth_mult): Likewise.
10939 (emit_store_flag_1): Likewise.
10940 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10941 of a comparison with size.
10942 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10943 (reduce_to_bit_field_precision): Likewise.
10944 * function.c (expand_function_end): Likewise.
10945 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10946 * loop-doloop.c (doloop_modify): Likewise.
10947 * optabs.c (expand_binop): Likewise.
10948 (expand_unop): Likewise.
10949 (expand_copysign_absneg): Likewise.
10950 (prepare_cmp_insn): Likewise.
10951 (maybe_legitimize_operand): Likewise.
10952 * recog.c (const_scalar_int_operand): Likewise.
10953 * rtlanal.c (get_address_mode): Likewise.
10954 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10955 (simplify_cond_clz_ctz): Likewise.
10956 * tree-nested.c (get_nl_goto_field): Likewise.
10957 * tree.c (build_vector_type_for_mode): Likewise.
10958 * var-tracking.c (use_narrower_mode): Likewise.
10960 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10961 Alan Hayward <alan.hayward@arm.com>
10962 David Sherwood <david.sherwood@arm.com>
10964 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10965 * builtins.c (expand_builtin_signbit): Use it.
10966 * cfgexpand.c (expand_debug_expr): Likewise.
10967 * dojump.c (do_jump): Likewise.
10968 (do_compare_and_jump): Likewise.
10969 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
10970 * expmed.c (make_tree): Likewise.
10971 * expr.c (expand_expr_real_2): Likewise.
10972 (expand_expr_real_1): Likewise.
10973 (try_casesi): Likewise.
10974 * fold-const-call.c (fold_const_call_ss): Likewise.
10975 * fold-const.c (unextend): Likewise.
10976 (extract_muldiv_1): Likewise.
10977 (fold_single_bit_test): Likewise.
10978 (native_encode_int): Likewise.
10979 (native_encode_string): Likewise.
10980 (native_interpret_int): Likewise.
10981 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
10982 * internal-fn.c (expand_addsub_overflow): Likewise.
10983 (expand_neg_overflow): Likewise.
10984 (expand_mul_overflow): Likewise.
10985 (expand_arith_overflow): Likewise.
10986 * match.pd: Likewise.
10987 * stor-layout.c (layout_type): Likewise.
10988 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10989 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
10990 * tree-ssanames.c (get_range_info): Likewise.
10991 * tree-switch-conversion.c (array_value_type) Likewise.
10992 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
10993 (vect_recog_divmod_pattern): Likewise.
10994 (vect_recog_mixed_size_cond_pattern): Likewise.
10995 * tree-vrp.c (extract_range_basic): Likewise.
10996 (simplify_float_conversion_using_ranges): Likewise.
10997 * tree.c (int_fits_type_p): Likewise.
10998 * ubsan.c (instrument_bool_enum_load): Likewise.
10999 * varasm.c (mergeable_string_section): Likewise.
11000 (narrowing_initializer_constant_valid_p): Likewise.
11001 (output_constant): Likewise.
11003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11004 Alan Hayward <alan.hayward@arm.com>
11005 David Sherwood <david.sherwood@arm.com>
11007 * machmode.h (NARROWEST_INT_MODE): New macro.
11008 * expr.c (alignment_for_piecewise_move): Use it instead of
11009 GET_CLASS_NARROWEST_MODE (MODE_INT).
11010 (push_block): Likewise.
11011 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11013 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11015 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11016 Alan Hayward <alan.hayward@arm.com>
11017 David Sherwood <david.sherwood@arm.com>
11019 * postreload.c (move2add_valid_value_p): Change the type of the
11020 mode parameter to scalar_int_mode.
11021 (move2add_use_add2_insn): Add a mode parameter and use it instead
11023 (move2add_use_add3_insn): Likewise.
11024 (reload_cse_move2add): Update accordingly.
11026 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11027 Alan Hayward <alan.hayward@arm.com>
11028 David Sherwood <david.sherwood@arm.com>
11030 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11032 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11033 * optabs.c (expand_unop): Likewise.
11035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11036 Alan Hayward <alan.hayward@arm.com>
11037 David Sherwood <david.sherwood@arm.com>
11039 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11040 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11041 (popcount_loc_descriptor): Likewise.
11042 (bswap_loc_descriptor): Likewise.
11043 (rotate_loc_descriptor): Likewise.
11044 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11045 calling the functions above.
11047 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11048 Alan Hayward <alan.hayward@arm.com>
11049 David Sherwood <david.sherwood@arm.com>
11051 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11053 (try_combine): Likewise.
11054 (simplify_if_then_else): Likewise.
11055 * cse.c (cse_insn): Likewise.
11056 * dwarf2out.c (mem_loc_descriptor): Likewise.
11057 * emit-rtl.c (gen_lowpart_common): Likewise.
11058 * simplify-rtx.c (simplify_truncation): Likewise.
11059 (simplify_binary_operation_1): Likewise.
11060 (simplify_const_relational_operation): Likewise.
11061 (simplify_ternary_operation): Likewise.
11062 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11065 Alan Hayward <alan.hayward@arm.com>
11066 David Sherwood <david.sherwood@arm.com>
11068 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11069 * reload.c (push_reload): Likewise.
11070 (find_reloads): Likewise.
11072 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11073 Alan Hayward <alan.hayward@arm.com>
11074 David Sherwood <david.sherwood@arm.com>
11076 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11077 (make_compound_operation_int): Likewise.
11078 (change_zero_ext): Likewise.
11079 * expr.c (convert_move): Likewise.
11080 (convert_modes): Likewise.
11081 * fwprop.c (forward_propagate_subreg): Likewise.
11082 * loop-iv.c (get_biv_step_1): Likewise.
11083 * optabs.c (widen_operand): Likewise.
11084 * postreload.c (move2add_valid_value_p): Likewise.
11085 * recog.c (simplify_while_replacing): Likewise.
11086 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11087 (simplify_binary_operation_1): Likewise. Remove redundant
11088 mode equality check.
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 (combine_simplify_rtx): Add checks for
11095 is_a <scalar_int_mode>.
11096 (simplify_if_then_else): Likewise.
11097 (make_field_assignment): Likewise.
11098 (simplify_comparison): Likewise.
11099 * ifcvt.c (noce_try_bitop): Likewise.
11100 * loop-invariant.c (canonicalize_address_mult): Likewise.
11101 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11104 Alan Hayward <alan.hayward@arm.com>
11105 David Sherwood <david.sherwood@arm.com>
11107 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11108 is_a <scalar_int_mode> instead of != BLKmode.
11110 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11111 Alan Hayward <alan.hayward@arm.com>
11112 David Sherwood <david.sherwood@arm.com>
11114 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11115 instead of != VOIDmode.
11116 * combine.c (if_then_else_cond): Likewise.
11117 (change_zero_ext): Likewise.
11118 * dwarf2out.c (mem_loc_descriptor): Likewise.
11119 (loc_descriptor): Likewise.
11120 * rtlanal.c (canonicalize_condition): Likewise.
11121 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11124 Alan Hayward <alan.hayward@arm.com>
11125 David Sherwood <david.sherwood@arm.com>
11127 * simplify-rtx.c (simplify_binary_operation_1): Use
11128 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11131 Alan Hayward <alan.hayward@arm.com>
11132 David Sherwood <david.sherwood@arm.com>
11134 * wide-int.h (int_traits<unsigned char>) New class.
11135 (int_traits<unsigned short>) Likewise.
11136 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11137 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11139 * combine.c (set_nonzero_bits_and_sign_copies): Use
11140 is_a <scalar_int_mode>.
11141 (find_split_point): Likewise.
11142 (combine_simplify_rtx): Likewise.
11143 (simplify_logical): Likewise.
11144 (expand_compound_operation): Likewise.
11145 (expand_field_assignment): Likewise.
11146 (make_compound_operation): Likewise.
11147 (extended_count): Likewise.
11148 (change_zero_ext): Likewise.
11149 (simplify_comparison): Likewise.
11150 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11151 (ucompare_loc_descriptor): Likewise.
11152 (minmax_loc_descriptor): Likewise.
11153 (mem_loc_descriptor): Likewise.
11154 (loc_descriptor): Likewise.
11155 * expmed.c (init_expmed_one_mode): Likewise.
11156 * lra-constraints.c (lra_constraint_offset): Likewise.
11157 * optabs.c (prepare_libcall_arg): Likewise.
11158 * postreload.c (move2add_note_store): Likewise.
11159 * reload.c (operands_match_p): Likewise.
11160 * rtl.h (load_extend_op): Likewise.
11161 * rtlhooks.c (gen_lowpart_general): Likewise.
11162 * simplify-rtx.c (simplify_truncation): Likewise.
11163 (simplify_unary_operation_1): Likewise.
11164 (simplify_binary_operation_1): Likewise.
11165 (simplify_const_binary_operation): Likewise.
11166 (simplify_const_relational_operation): Likewise.
11167 (simplify_subreg): Likewise.
11168 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11169 * var-tracking.c (adjust_mems): Likewise.
11170 (prepare_call_arguments): Likewise.
11172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11173 Alan Hayward <alan.hayward@arm.com>
11174 David Sherwood <david.sherwood@arm.com>
11176 * machmode.h (is_int_mode): New fuction.
11177 * combine.c (find_split_point): Use it.
11178 (combine_simplify_rtx): Likewise.
11179 (simplify_if_then_else): Likewise.
11180 (simplify_set): Likewise.
11181 (simplify_shift_const_1): Likewise.
11182 (simplify_comparison): Likewise.
11183 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
11184 * cse.c (notreg_cost): Likewise.
11185 (cse_insn): Likewise.
11186 * cselib.c (cselib_lookup_1): Likewise.
11187 * dojump.c (do_jump_1): Likewise.
11188 (do_compare_rtx_and_jump): Likewise.
11189 * dse.c (get_call_args): Likewise.
11190 * dwarf2out.c (rtl_for_decl_init): Likewise.
11191 (native_encode_initializer): Likewise.
11192 * expmed.c (emit_store_flag_1): Likewise.
11193 (emit_store_flag): Likewise.
11194 * expr.c (convert_modes): Likewise.
11195 (store_field): Likewise.
11196 (expand_expr_real_1): Likewise.
11197 * fold-const.c (fold_read_from_constant_string): Likewise.
11198 * gimple-ssa-sprintf.c (get_format_string): Likewise.
11199 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
11200 * optabs.c (expand_binop): Likewise.
11201 (expand_unop): Likewise.
11202 (expand_abs_nojump): Likewise.
11203 (expand_one_cmpl_abs_nojump): Likewise.
11204 * simplify-rtx.c (mode_signbit_p): Likewise.
11205 (val_signbit_p): Likewise.
11206 (val_signbit_known_set_p): Likewise.
11207 (val_signbit_known_clear_p): Likewise.
11208 (simplify_relational_operation_1): Likewise.
11209 * tree.c (vector_type_mode): Likewise.
11211 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11212 Alan Hayward <alan.hayward@arm.com>
11213 David Sherwood <david.sherwood@arm.com>
11215 * machmode.h (smallest_mode_for_size): Fix formatting.
11216 (smallest_int_mode_for_size): New function.
11217 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11218 instead of smallest_mode_for_size.
11219 * combine.c (make_extraction): Likewise.
11220 * config/arc/arc.c (arc_expand_movmem): Likewise.
11221 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11222 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11223 * config/s390/s390.c (s390_expand_insv): Likewise.
11224 * config/sparc/sparc.c (assign_int_registers): Likewise.
11225 * config/spu/spu.c (spu_function_value): Likewise.
11226 (spu_function_arg): Likewise.
11227 * coverage.c (get_gcov_type): Likewise.
11228 (get_gcov_unsigned_t): Likewise.
11229 * dse.c (find_shift_sequence): Likewise.
11230 * expmed.c (store_bit_field_1): Likewise.
11231 * expr.c (convert_move): Likewise.
11232 (store_field): Likewise.
11233 * internal-fn.c (expand_arith_overflow): Likewise.
11234 * optabs-query.c (get_best_extraction_insn): Likewise.
11235 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11236 * stor-layout.c (layout_type): Likewise.
11237 (initialize_sizetypes): Likewise.
11238 * targhooks.c (default_get_mask_mode): Likewise.
11239 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11242 Alan Hayward <alan.hayward@arm.com>
11243 David Sherwood <david.sherwood@arm.com>
11245 * machmode.h (opt_mode::else_blk): New function.
11246 (int_mode_for_mode): Declare.
11247 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11248 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11250 * cfgexpand.c (expand_debug_expr): Likewise.
11251 * combine.c (gen_lowpart_or_truncate): Likewise.
11252 (gen_lowpart_for_combine): Likewise.
11253 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11254 * config/avr/avr.c (avr_to_int_mode): Likewise.
11255 (avr_out_plus_1): Likewise.
11256 (avr_out_plus): Likewise.
11257 (avr_out_round): Likewise.
11258 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11259 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11260 (s390_expand_vcond): Likewise.
11261 * config/spu/spu.c (spu_split_immediate): Likewise.
11262 (spu_expand_mov): Likewise.
11263 * dse.c (get_stored_val): Likewise.
11264 * expmed.c (store_bit_field_1): Likewise.
11265 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11267 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11268 (extract_low_bits): Likewise.
11269 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11270 handling rather than repeating the check.
11271 (emit_group_store): Likewise.
11272 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11273 * optabs.c (expand_absneg_bit): Likewise.
11274 (expand_copysign_absneg): Likewise.
11275 (expand_copysign_bit): Likewise.
11276 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11277 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11278 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11279 * var-tracking.c (prepare_call_arguments): Likewise.
11280 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11281 int_mode_for_mode instead of mode_for_size.
11282 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11285 Alan Hayward <alan.hayward@arm.com>
11286 David Sherwood <david.sherwood@arm.com>
11288 * machmode.h (int_mode_for_size): New function.
11289 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11290 instead of mode_for_size.
11291 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11293 * combine.c (expand_field_assignment): Use int_mode_for_size
11294 instead of mode_for_size.
11295 (make_extraction): Likewise.
11296 (simplify_shift_const_1): Likewise.
11297 (simplify_comparison): Likewise.
11298 * dojump.c (do_jump): Likewise.
11299 * dwarf2out.c (mem_loc_descriptor): Likewise.
11300 * emit-rtl.c (init_derived_machine_modes): Likewise.
11301 * expmed.c (flip_storage_order): Likewise.
11302 (convert_extracted_bit_field): Likewise.
11303 * expr.c (copy_blkmode_from_reg): Likewise.
11304 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11305 * internal-fn.c (expand_mul_overflow): Likewise.
11306 * lower-subreg.c (simple_move): Likewise.
11307 * optabs-libfuncs.c (init_optabs): Likewise.
11308 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11309 * tree.c (vector_type_mode): Likewise.
11310 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11311 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11312 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11313 * tree-vect-stmts.c (vectorizable_load): Likewise.
11314 (vectorizable_store): Likewise.
11316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11317 Alan Hayward <alan.hayward@arm.com>
11318 David Sherwood <david.sherwood@arm.com>
11320 * coretypes.h (pod_mode): New type.
11321 (scalar_int_mode_pod): New typedef.
11322 * machmode.h (pod_mode): New class.
11323 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11324 * genmodes.c (emit_mode_int_n): Update accordingly.
11325 * lower-subreg.h (target_lower_subreg): Change type to
11326 scalar_int_mode_pod.
11327 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11328 scalar_int_mode_pod.
11330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11331 Alan Hayward <alan.hayward@arm.com>
11332 David Sherwood <david.sherwood@arm.com>
11334 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11335 machine_mode to scalar_int_mode.
11336 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11337 (rs6000_option_override_internal): Remove cast to int.
11338 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11339 machine_mode to scalar_int_mode.
11340 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11341 (rs6000_option_override_internal): Remove cast to int.
11342 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11343 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11345 * config/s390/s390.c (s390_expand_builtin): Likewise.
11346 * coretypes.h (scalar_int_mode): New type.
11347 (opt_scalar_int_mode): New typedef.
11348 * machmode.h (scalar_int_mode): New class.
11349 (scalar_int_mode::includes_p): New function.
11350 (byte_mode): Change type to scalar_int_mode.
11351 (word_mode): Likewise.
11352 (ptr_mode): Likewise.
11353 * emit-rtl.c (byte_mode): Likewise.
11354 (word_mode): Likewise.
11355 (ptr_mode): Likewise.
11356 (init_derived_machine_modes): Update accordingly.
11357 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11358 and MODE_PARTIAL_INT.
11359 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11360 opt_scalar_int_mode.
11362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11363 Alan Hayward <alan.hayward@arm.com>
11364 David Sherwood <david.sherwood@arm.com>
11366 * target.def (libgcc_floating_mode_supported_p): Take a
11368 * doc/tm.texi: Regenerate.
11369 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11371 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11372 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11376 Alan Hayward <alan.hayward@arm.com>
11377 David Sherwood <david.sherwood@arm.com>
11379 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11380 * doc/tm.texi: Regenerate.
11381 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11382 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11383 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11384 * targhooks.h (default_floatn_mode): Likewise.
11385 * targhooks.c (default_floatn_mode): Likewise.
11386 * tree.c (build_common_tree_nodes): Update accordingly.
11388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11389 Alan Hayward <alan.hayward@arm.com>
11390 David Sherwood <david.sherwood@arm.com>
11392 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11393 (mode_iterator::iterate_p): Likewise.
11394 (mode_iterator::get_wider): Likewise.
11395 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11398 Alan Hayward <alan.hayward@arm.com>
11399 David Sherwood <david.sherwood@arm.com>
11401 * coretypes.h (opt_scalar_float_mode): New typedef.
11402 * machmode.h (float_mode_for_size): New function.
11403 * emit-rtl.c (double_mode): Delete.
11404 (init_emit_once): Use float_mode_for_size.
11405 * stor-layout.c (layout_type): Likewise.
11406 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11409 Alan Hayward <alan.hayward@arm.com>
11410 David Sherwood <david.sherwood@arm.com>
11412 * output.h (assemble_real): Take a scalar_float_mode.
11413 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11414 * config/arm/arm.md (consttable_4): Likewise.
11415 (consttable_8): Likewise.
11416 (consttable_16): Likewise.
11417 * config/mips/mips.md (consttable_float): Likewise.
11418 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11419 * varasm.c (assemble_real): Take a scalar_float_mode.
11420 (output_constant_pool_2): Update accordingly.
11421 (output_constant): Likewise.
11423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11424 Alan Hayward <alan.hayward@arm.com>
11425 David Sherwood <david.sherwood@arm.com>
11427 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11428 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11429 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11430 (native_encode_real): Likewise.
11431 (native_interpret_real): Likewise.
11432 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11433 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11436 Alan Hayward <alan.hayward@arm.com>
11437 David Sherwood <david.sherwood@arm.com>
11439 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11440 <scalar_float_mode>. Simplify.
11441 (gen_extend_conv_libfunc): Likewise.
11443 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11444 Alan Hayward <alan.hayward@arm.com>
11445 David Sherwood <david.sherwood@arm.com>
11447 * coretypes.h (scalar_float_mode): New type.
11448 * machmode.h (mode_traits::from_int): Use machine_mode if
11449 USE_ENUM_MODES is defined.
11450 (is_a): New function.
11452 (dyn_cast): Likewise.
11453 (scalar_float_mode): New class.
11454 (scalar_float_mode::includes_p): New function.
11455 (is_float_mode): Likewise.
11456 * gdbhooks.py (MachineModePrinter): New class.
11457 (build_pretty_printer): Use it for scalar_float_mode.
11458 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11459 (format_helper::format_helper): Turn into a template.
11460 * genmodes.c (get_mode_class): New function.
11461 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11462 or machine_mode if none.
11463 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11464 as_a <scalar_float_mode>.
11465 * dwarf2out.c (mem_loc_descriptor): Likewise.
11466 (insert_float): Likewise.
11467 (add_const_value_attribute): Likewise.
11468 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11469 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11470 (expand_unop): Update accordingly.
11471 (expand_abs_nojump): Likewise.
11472 (expand_copysign_absneg): Take a scalar_float_mode.
11473 (expand_copysign_bit): Likewise.
11474 (expand_copysign): Update accordingly.
11476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11477 Alan Hayward <alan.hayward@arm.com>
11478 David Sherwood <david.sherwood@arm.com>
11480 * coretypes.h (opt_mode): New class.
11481 * machmode.h (opt_mode): Likewise.
11482 (opt_mode::else_void): New function.
11483 (opt_mode::require): Likewise.
11484 (opt_mode::exists): Likewise.
11485 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11486 (GET_MODE_2XWIDER_MODE): Likewise.
11487 (mode_iterator::get_wider): Update accordingly.
11488 (mode_iterator::get_2xwider): Likewise.
11489 (mode_iterator::get_known_wider): Likewise, turning into a template.
11490 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11491 forcing a wider mode to exist.
11492 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11493 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11494 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11495 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11496 * lower-subreg.c (init_lower_subreg): Likewise.
11497 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11498 on the final iteration.
11499 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11500 a wider mode exists before asking for a move pattern.
11501 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11502 forcing a wider mode to exist.
11503 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11504 returning false if no such mode exists.
11505 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11506 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11507 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11508 Avoid checking for a MODE_INT if we already know the mode is not a
11510 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11511 forcing a wider mode to exist.
11512 (expmed_mult_highpart_optab): Likewise.
11513 (expmed_mult_highpart): Likewise.
11514 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11516 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11517 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11518 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11519 * internal-fn.c (expand_mul_overflow): Update use of
11520 GET_MODE_2XWIDER_MODE.
11521 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11522 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11523 GET_MODE_WIDER_MODE.
11524 (convert_plusminus_to_widen): Likewise.
11525 * tree-switch-conversion.c (array_value_type): Likewise.
11526 * var-tracking.c (emit_note_insn_var_location): Likewise.
11527 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11528 Return false inside rather than outside the loop if no wider mode
11530 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11531 and GET_MODE_2XWIDER_MODE
11532 (can_compare_p): Use else_void.
11533 * gdbhooks.py (OptMachineModePrinter): New class.
11534 (build_pretty_printer): Use it for opt_mode.
11536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11537 Alan Hayward <alan.hayward@arm.com>
11538 David Sherwood <david.sherwood@arm.com>
11540 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11541 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11544 Alan Hayward <alan.hayward@arm.com>
11545 David Sherwood <david.sherwood@arm.com>
11547 * machmode.h (mode_traits): New structure.
11548 (get_narrowest_mode): New function.
11549 (mode_iterator::start): Likewise.
11550 (mode_iterator::iterate_p): Likewise.
11551 (mode_iterator::get_wider): Likewise.
11552 (mode_iterator::get_known_wider): Likewise.
11553 (mode_iterator::get_2xwider): Likewise.
11554 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11555 (FOR_EACH_MODE): Likewise.
11556 (FOR_EACH_MODE_FROM): Likewise.
11557 (FOR_EACH_MODE_UNTIL): Likewise.
11558 (FOR_EACH_WIDER_MODE): Likewise.
11559 (FOR_EACH_2XWIDER_MODE): Likewise.
11560 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11561 * combine.c (simplify_comparison): Likewise
11562 * config/i386/i386.c (type_natural_mode): Likewise.
11563 * cse.c (cse_insn): Likewise.
11564 * dse.c (find_shift_sequence): Likewise.
11565 * emit-rtl.c (init_derived_machine_modes): Likewise.
11566 (init_emit_once): Likewise.
11567 * explow.c (hard_function_value): Likewise.
11568 * expmed.c (extract_fixed_bit_field_1): Likewise.
11569 (extract_bit_field_1): Likewise.
11570 (expand_divmod): Likewise.
11571 (emit_store_flag_1): Likewise.
11572 * expr.c (init_expr_target): Likewise.
11573 (convert_move): Likewise.
11574 (alignment_for_piecewise_move): Likewise.
11575 (widest_int_mode_for_size): Likewise.
11576 (emit_block_move_via_movmem): Likewise.
11577 (copy_blkmode_to_reg): Likewise.
11578 (set_storage_via_setmem): Likewise.
11579 (compress_float_constant): Likewise.
11580 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11581 * optabs-query.c (get_best_extraction_insn): Likewise.
11582 * optabs.c (expand_binop): Likewise.
11583 (expand_twoval_unop): Likewise.
11584 (expand_twoval_binop): Likewise.
11585 (widen_leading): Likewise.
11586 (widen_bswap): Likewise.
11587 (expand_parity): Likewise.
11588 (expand_unop): Likewise.
11589 (prepare_cmp_insn): Likewise.
11590 (prepare_float_lib_cmp): Likewise.
11591 (expand_float): Likewise.
11592 (expand_fix): Likewise.
11593 (expand_sfix_optab): Likewise.
11594 * postreload.c (move2add_use_add2_insn): Likewise.
11595 * reg-stack.c (reg_to_stack): Likewise.
11596 * reginfo.c (choose_hard_reg_mode): Likewise.
11597 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11598 * stor-layout.c (mode_for_size): Likewise.
11599 (smallest_mode_for_size): Likewise.
11600 (mode_for_vector): Likewise.
11601 (finish_bitfield_representative): Likewise.
11602 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11603 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11604 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11605 * var-tracking.c (prepare_call_arguments): Likewise.
11607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11608 Alan Hayward <alan.hayward@arm.com>
11609 David Sherwood <david.sherwood@arm.com>
11611 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11612 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11613 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11614 * machmode.h (mode_size): Move earlier in file.
11615 (mode_precision): Likewise.
11616 (mode_inner): Likewise.
11617 (mode_nunits): Likewise.
11618 (mode_unit_size): Likewise.
11619 (unit_unit_precision): Likewise.
11620 (mode_wider): Likewise.
11621 (mode_2xwider): Likewise.
11622 (machine_mode): New class.
11623 (mode_to_bytes): New function.
11624 (mode_to_bits): Likewise.
11625 (mode_to_precision): Likewise.
11626 (mode_to_inner): Likewise.
11627 (mode_to_unit_size): Likewise.
11628 (mode_to_unit_precision): Likewise.
11629 (mode_to_nunits): Likewise.
11630 (GET_MODE_SIZE): Use mode_to_bytes.
11631 (GET_MODE_BITSIZE): Use mode_to_bits.
11632 (GET_MODE_PRECISION): Use mode_to_precision.
11633 (GET_MODE_INNER): Use mode_to_inner.
11634 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11635 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11636 (GET_MODE_NUNITS): Use mode_to_nunits.
11637 * system.h (ALWAYS_INLINE): New macro.
11638 * config/powerpcspe/powerpcspe-c.c
11639 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11640 int for arg1_mode and arg2_mode.
11642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11643 Alan Hayward <alan.hayward@arm.com>
11644 David Sherwood <david.sherwood@arm.com>
11646 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11647 Prefix mode names with E_ in case statements.
11648 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11649 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11650 (aarch64_split_simd_move): Likewise.
11651 (aarch64_gen_storewb_pair): Likewise.
11652 (aarch64_gen_loadwb_pair): Likewise.
11653 (aarch64_gen_store_pair): Likewise.
11654 (aarch64_gen_load_pair): Likewise.
11655 (aarch64_get_condition_code_1): Likewise.
11656 (aarch64_constant_pool_reload_icode): Likewise.
11657 (get_rsqrte_type): Likewise.
11658 (get_rsqrts_type): Likewise.
11659 (get_recpe_type): Likewise.
11660 (get_recps_type): Likewise.
11661 (aarch64_gimplify_va_arg_expr): Likewise.
11662 (aarch64_simd_container_mode): Likewise.
11663 (aarch64_emit_load_exclusive): Likewise.
11664 (aarch64_emit_store_exclusive): Likewise.
11665 (aarch64_expand_compare_and_swap): Likewise.
11666 (aarch64_gen_atomic_cas): Likewise.
11667 (aarch64_emit_bic): Likewise.
11668 (aarch64_emit_atomic_swap): Likewise.
11669 (aarch64_emit_atomic_load_op): Likewise.
11670 (aarch64_evpc_trn): Likewise.
11671 (aarch64_evpc_uzp): Likewise.
11672 (aarch64_evpc_zip): Likewise.
11673 (aarch64_evpc_ext): Likewise.
11674 (aarch64_evpc_rev): Likewise.
11675 (aarch64_evpc_dup): Likewise.
11676 (aarch64_gen_ccmp_first): Likewise.
11677 (aarch64_gen_ccmp_next): Likewise.
11678 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11679 (alpha_emit_xfloating_libcall): Likewise.
11680 (emit_insxl): Likewise.
11681 (alpha_arg_type): Likewise.
11682 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11683 (arc_preferred_simd_mode): Likewise.
11684 (arc_secondary_reload): Likewise.
11685 (get_arc_condition_code): Likewise.
11686 (arc_print_operand): Likewise.
11687 (arc_legitimate_constant_p): Likewise.
11688 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11689 * config/arc/arc.md (casesi_load): Likewise.
11690 (casesi_compact_jump): Likewise.
11691 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11692 (cc_use_register): Likewise.
11693 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11694 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11695 (arm_init_iwmmxt_builtins): Likewise.
11696 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11697 (neon_expand_vector_init): Likewise.
11698 (arm_attr_length_move_neon): Likewise.
11699 (maybe_get_arm_condition_code): Likewise.
11700 (arm_emit_vector_const): Likewise.
11701 (arm_preferred_simd_mode): Likewise.
11702 (arm_output_iwmmxt_tinsr): Likewise.
11703 (thumb1_output_casesi): Likewise.
11704 (thumb2_output_casesi): Likewise.
11705 (arm_emit_load_exclusive): Likewise.
11706 (arm_emit_store_exclusive): Likewise.
11707 (arm_expand_compare_and_swap): Likewise.
11708 (arm_evpc_neon_vuzp): Likewise.
11709 (arm_evpc_neon_vzip): Likewise.
11710 (arm_evpc_neon_vrev): Likewise.
11711 (arm_evpc_neon_vtrn): Likewise.
11712 (arm_evpc_neon_vext): Likewise.
11713 (arm_validize_comparison): Likewise.
11714 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11715 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11716 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11717 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11718 (c6x_preferred_simd_mode): Likewise.
11719 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11720 (epiphany_rtx_costs): Likewise.
11721 * config/epiphany/predicates.md (proper_comparison_operator):
11723 * config/frv/frv.c (condexec_memory_operand): Likewise.
11724 (frv_emit_move): Likewise.
11725 (output_move_single): Likewise.
11726 (output_condmove_single): Likewise.
11727 (frv_hard_regno_mode_ok): Likewise.
11728 (frv_matching_accg_mode): Likewise.
11729 * config/h8300/h8300.c (split_adds_subs): Likewise.
11730 (h8300_rtx_costs): Likewise.
11731 (h8300_print_operand): Likewise.
11732 (compute_mov_length): Likewise.
11733 (output_logical_op): Likewise.
11734 (compute_logical_op_length): Likewise.
11735 (compute_logical_op_cc): Likewise.
11736 (h8300_shift_needs_scratch_p): Likewise.
11737 (output_a_shift): Likewise.
11738 (compute_a_shift_length): Likewise.
11739 (compute_a_shift_cc): Likewise.
11740 (expand_a_rotate): Likewise.
11741 (output_a_rotate): Likewise.
11742 * config/i386/i386.c (classify_argument): Likewise.
11743 (function_arg_advance_32): Likewise.
11744 (function_arg_32): Likewise.
11745 (function_arg_64): Likewise.
11746 (function_value_64): Likewise.
11747 (ix86_gimplify_va_arg): Likewise.
11748 (ix86_legitimate_constant_p): Likewise.
11749 (put_condition_code): Likewise.
11750 (split_double_mode): Likewise.
11751 (ix86_avx256_split_vector_move_misalign): Likewise.
11752 (ix86_expand_vector_logical_operator): Likewise.
11753 (ix86_split_idivmod): Likewise.
11754 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11755 (ix86_build_const_vector): Likewise.
11756 (ix86_build_signbit_mask): Likewise.
11757 (ix86_match_ccmode): Likewise.
11758 (ix86_cc_modes_compatible): Likewise.
11759 (ix86_expand_branch): Likewise.
11760 (ix86_expand_sse_cmp): Likewise.
11761 (ix86_expand_sse_movcc): Likewise.
11762 (ix86_expand_int_sse_cmp): Likewise.
11763 (ix86_expand_vec_perm_vpermi2): Likewise.
11764 (ix86_expand_vec_perm): Likewise.
11765 (ix86_expand_sse_unpack): Likewise.
11766 (ix86_expand_int_addcc): Likewise.
11767 (ix86_split_to_parts): Likewise.
11768 (ix86_vectorize_builtin_gather): Likewise.
11769 (ix86_vectorize_builtin_scatter): Likewise.
11770 (avx_vpermilp_parallel): Likewise.
11771 (inline_memory_move_cost): Likewise.
11772 (ix86_tieable_integer_mode_p): Likewise.
11773 (x86_maybe_negate_const_int): Likewise.
11774 (ix86_expand_vector_init_duplicate): Likewise.
11775 (ix86_expand_vector_init_one_nonzero): Likewise.
11776 (ix86_expand_vector_init_one_var): Likewise.
11777 (ix86_expand_vector_init_concat): Likewise.
11778 (ix86_expand_vector_init_interleave): Likewise.
11779 (ix86_expand_vector_init_general): Likewise.
11780 (ix86_expand_vector_set): Likewise.
11781 (ix86_expand_vector_extract): Likewise.
11782 (emit_reduc_half): Likewise.
11783 (ix86_emit_i387_round): Likewise.
11784 (ix86_mangle_type): Likewise.
11785 (ix86_expand_round_sse4): Likewise.
11786 (expand_vec_perm_blend): Likewise.
11787 (canonicalize_vector_int_perm): Likewise.
11788 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11789 (expand_vec_perm_1): Likewise.
11790 (expand_vec_perm_interleave3): Likewise.
11791 (expand_vec_perm_even_odd_pack): Likewise.
11792 (expand_vec_perm_even_odd_1): Likewise.
11793 (expand_vec_perm_broadcast_1): Likewise.
11794 (ix86_vectorize_vec_perm_const_ok): Likewise.
11795 (ix86_expand_vecop_qihi): Likewise.
11796 (ix86_expand_mul_widen_hilo): Likewise.
11797 (ix86_expand_sse2_abs): Likewise.
11798 (ix86_expand_pextr): Likewise.
11799 (ix86_expand_pinsr): Likewise.
11800 (ix86_preferred_simd_mode): Likewise.
11801 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11802 * config/i386/sse.md (*andnot<mode>3): Likewise.
11803 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11804 (*<code><mode>3): Likewise.
11805 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11806 (ia64_expand_atomic_op): Likewise.
11807 (ia64_arg_type): Likewise.
11808 (ia64_mode_to_int): Likewise.
11809 (ia64_scalar_mode_supported_p): Likewise.
11810 (ia64_vector_mode_supported_p): Likewise.
11811 (expand_vec_perm_broadcast): Likewise.
11812 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11813 (iq2000_function_arg_advance): Likewise.
11814 (iq2000_function_arg): Likewise.
11815 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11816 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11817 (m68k_libcall_value): Likewise.
11818 (m68k_function_value): Likewise.
11819 (sched_attr_op_type): Likewise.
11820 * config/mcore/mcore.c (mcore_output_move): Likewise.
11821 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11823 (microblaze_function_arg): Likewise.
11824 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11825 (mips_print_operand): Likewise.
11826 (mips_mode_ok_for_mov_fmt_p): Likewise.
11827 (mips_vector_mode_supported_p): Likewise.
11828 (mips_preferred_simd_mode): Likewise.
11829 (mips_expand_vpc_loongson_even_odd): Likewise.
11830 (mips_expand_vec_unpack): Likewise.
11831 (mips_expand_vi_broadcast): Likewise.
11832 (mips_expand_vector_init): Likewise.
11833 (mips_expand_vec_reduc): Likewise.
11834 (mips_expand_msa_cmp): Likewise.
11835 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11836 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11837 (cc_flags_for_mode): Likewise.
11838 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11839 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11840 (nds32_output_casesi_pc_relative): Likewise.
11841 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11842 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11843 (nvptx_gen_unpack): Likewise.
11844 (nvptx_gen_pack): Likewise.
11845 (nvptx_gen_shuffle): Likewise.
11846 (nvptx_gen_wcast): Likewise.
11847 (nvptx_preferred_simd_mode): Likewise.
11848 * config/pa/pa.c (pa_secondary_reload): Likewise.
11849 * config/pa/predicates.md (base14_operand): Likewise.
11850 * config/powerpcspe/powerpcspe-c.c
11851 (altivec_resolve_overloaded_builtin): Likewise.
11852 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11854 (rs6000_preferred_simd_mode): Likewise.
11855 (output_vec_const_move): Likewise.
11856 (rs6000_expand_vector_extract): Likewise.
11857 (rs6000_split_vec_extract_var): Likewise.
11858 (reg_offset_addressing_ok_p): Likewise.
11859 (rs6000_legitimate_offset_address_p): Likewise.
11860 (rs6000_legitimize_address): Likewise.
11861 (rs6000_emit_set_const): Likewise.
11862 (rs6000_const_vec): Likewise.
11863 (rs6000_emit_move): Likewise.
11864 (spe_build_register_parallel): Likewise.
11865 (rs6000_darwin64_record_arg_recurse): Likewise.
11866 (swap_selector_for_mode): Likewise.
11867 (spe_init_builtins): Likewise.
11868 (paired_init_builtins): Likewise.
11869 (altivec_init_builtins): Likewise.
11870 (do_load_for_compare): Likewise.
11871 (rs6000_generate_compare): Likewise.
11872 (rs6000_expand_float128_convert): Likewise.
11873 (emit_load_locked): Likewise.
11874 (emit_store_conditional): Likewise.
11875 (rs6000_output_function_epilogue): Likewise.
11876 (rs6000_handle_altivec_attribute): Likewise.
11877 (rs6000_function_value): Likewise.
11878 (emit_fusion_gpr_load): Likewise.
11879 (emit_fusion_p9_load): Likewise.
11880 (emit_fusion_p9_store): Likewise.
11881 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11882 (fusion_gpr_mem_load): Likewise.
11883 (fusion_addis_mem_combo_load): Likewise.
11884 (fusion_addis_mem_combo_store): Likewise.
11885 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11886 (fusion_gpr_mem_load): Likewise.
11887 (fusion_addis_mem_combo_load): Likewise.
11888 (fusion_addis_mem_combo_store): Likewise.
11889 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11891 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11892 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11893 (rs6000_preferred_simd_mode): Likewise.
11894 (output_vec_const_move): Likewise.
11895 (rs6000_expand_vector_extract): Likewise.
11896 (rs6000_split_vec_extract_var): Likewise.
11897 (reg_offset_addressing_ok_p): Likewise.
11898 (rs6000_legitimate_offset_address_p): Likewise.
11899 (rs6000_legitimize_address): Likewise.
11900 (rs6000_emit_set_const): Likewise.
11901 (rs6000_const_vec): Likewise.
11902 (rs6000_emit_move): Likewise.
11903 (rs6000_darwin64_record_arg_recurse): Likewise.
11904 (swap_selector_for_mode): Likewise.
11905 (paired_init_builtins): Likewise.
11906 (altivec_init_builtins): Likewise.
11907 (rs6000_expand_float128_convert): Likewise.
11908 (emit_load_locked): Likewise.
11909 (emit_store_conditional): Likewise.
11910 (rs6000_output_function_epilogue): Likewise.
11911 (rs6000_handle_altivec_attribute): Likewise.
11912 (rs6000_function_value): Likewise.
11913 (emit_fusion_gpr_load): Likewise.
11914 (emit_fusion_p9_load): Likewise.
11915 (emit_fusion_p9_store): Likewise.
11916 * config/rx/rx.c (rx_gen_move_template): Likewise.
11917 (flags_from_mode): Likewise.
11918 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11919 (s390_slb_comparison): Likewise.
11920 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11921 (s390_vector_mode_supported_p): Likewise.
11922 (s390_cc_modes_compatible): Likewise.
11923 (s390_match_ccmode_set): Likewise.
11924 (s390_canonicalize_comparison): Likewise.
11925 (s390_emit_compare_and_swap): Likewise.
11926 (s390_branch_condition_mask): Likewise.
11927 (s390_rtx_costs): Likewise.
11928 (s390_secondary_reload): Likewise.
11929 (__SECONDARY_RELOAD_CASE): Likewise.
11930 (s390_expand_cs): Likewise.
11931 (s390_preferred_simd_mode): Likewise.
11932 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11933 * config/sh/sh.c (sh_print_operand): Likewise.
11934 (dump_table): Likewise.
11935 (sh_secondary_reload): Likewise.
11936 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11937 * config/sh/sh.md (casesi_worker_1): Likewise.
11938 (casesi_worker_2): Likewise.
11939 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11940 (fcc_comparison_operator): Likewise.
11941 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11942 (emit_soft_tfmode_cvt): Likewise.
11943 (sparc_preferred_simd_mode): Likewise.
11944 (output_cbranch): Likewise.
11945 (sparc_print_operand): Likewise.
11946 (sparc_expand_vec_perm_bmask): Likewise.
11947 (vector_init_bshuffle): Likewise.
11948 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11949 (spu_vector_mode_supported_p): Likewise.
11950 (spu_expand_insv): Likewise.
11951 (spu_emit_branch_or_set): Likewise.
11952 (spu_handle_vector_attribute): Likewise.
11953 (spu_builtin_splats): Likewise.
11954 (spu_builtin_extract): Likewise.
11955 (spu_builtin_promote): Likewise.
11956 (spu_expand_sign_extend): Likewise.
11957 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11958 (tilegx_simd_int): Likewise.
11959 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11960 (tilepro_simd_int): Likewise.
11961 * config/v850/v850.c (const_double_split): Likewise.
11962 (v850_print_operand): Likewise.
11963 (ep_memory_offset): Likewise.
11964 * config/vax/vax.c (vax_rtx_costs): Likewise.
11965 (vax_output_int_move): Likewise.
11966 (vax_output_int_add): Likewise.
11967 (vax_output_int_subtract): Likewise.
11968 * config/visium/predicates.md (visium_branch_operator): Likewise.
11969 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
11970 (visium_print_operand_address): Likewise.
11971 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11972 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
11973 (xtensa_expand_conditional_branch): Likewise.
11974 (xtensa_copy_incoming_a7): Likewise.
11975 (xtensa_output_literal): Likewise.
11976 * dfp.c (decimal_real_maxval): Likewise.
11977 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11980 Alan Hayward <alan.hayward@arm.com>
11981 David Sherwood <david.sherwood@arm.com>
11983 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
11984 (mode_nunits_inline): Likewise.
11985 (mode_inner_inline): Likewise.
11986 (mode_unit_size_inline): Likewise.
11987 (mode_unit_precision_inline): Likewise.
11988 (emit_insn_modes_h): Likewise. Also emit a #define of the
11990 (emit_mode_wider): Add an E_ prefix to mode names.
11991 (emit_mode_complex): Likewise.
11992 (emit_mode_inner): Likewise.
11993 (emit_mode_adjustments): Likewise.
11994 (emit_mode_int_n): Likewise.
11995 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
11996 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
11997 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
11998 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
11999 (CRC32_BUILTIN, ENTRY): Likewise.
12000 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12001 (aarch64_pop_regs): Likewise.
12002 (aarch64_process_components): Likewise.
12003 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12004 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12005 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12006 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12007 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12008 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12010 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12011 (pcmpistr): Likewise.
12012 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12013 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12014 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12016 * config/rl78/rl78.c (mduc_regs): Likewise.
12017 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12018 (htm_expand_builtin): Likewise.
12019 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12020 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12021 * config/xtensa/xtensa.c (print_operand): Likewise.
12022 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12023 (NUM_MODE_VECTOR_INT): Likewise.
12024 * genoutput.c (null_operand): Likewise.
12025 (output_operand_data): Likewise.
12026 * genrecog.c (print_parameter_value): Likewise.
12027 * lra.c (debug_operand_data): Likewise.
12029 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12031 * dwarf2out.c (scompare_loc_descriptor_wide)
12032 (scompare_loc_descriptor_narrow): New functions, split out from...
12033 (scompare_loc_descriptor): ...here.
12034 * expmed.c (emit_store_flag_int): New function, split out from...
12035 (emit_store_flag): ...here.
12037 2017-08-30 Richard Biener <rguenther@suse.de>
12039 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12040 (dwarf2out_early_finish): Move setting of AT_pubnames from
12041 early debug output to early finish.
12043 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12045 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12046 and -mdata-region to the assembler.
12048 2017-08-30 Richard Biener <rguenther@suse.de>
12050 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12052 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12053 (dwarf2out_early_global_decl): Only generate a DIE for the
12054 abstract origin if it doesn't already exist or is a declaration DIE.
12055 (resolve_addr): Do not add the linkage name twice when
12056 generating a stub DIE for the DW_TAG_GNU_call_site target.
12058 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12060 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12061 Use machine_mode rather than int for arg1_mode.
12063 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12066 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12067 that the second argument of the built-in functions to unpack
12068 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12069 switch statement instead a lot of if statements.
12070 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12071 Allow 64-bit values to be in Altivec registers as well as
12072 traditional floating point registers.
12073 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12075 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12077 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12078 MAX_REGS_PER_ADDRESS == 1.
12080 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12082 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12083 * config/i386/i386.c (ix86_option_override_internal): Simplify
12084 setting of opts->x_flag_entry.
12086 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12087 Jakub Jelinek <jakub@redhat.com>
12088 Richard Biener <rguenther@suse.de>
12090 PR tree-optimization/81503
12091 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12092 folded constant fits in the target type; reorder tests for clarity.
12094 2017-08-29 Martin Liska <mliska@suse.cz>
12096 * passes.def: Include pass_lower_switch.
12097 * stmt.c (dump_case_nodes): Remove and move to
12098 tree-switch-conversion.
12099 (case_values_threshold): Likewise.
12100 (expand_switch_as_decision_tree_p): Likewise.
12101 (emit_case_decision_tree): Likewise.
12102 (expand_case): Likewise.
12103 (balance_case_nodes): Likewise.
12104 (node_has_low_bound): Likewise.
12105 (node_has_high_bound): Likewise.
12106 (node_is_bounded): Likewise.
12107 (emit_case_nodes): Likewise.
12108 (struct simple_case_node): New struct.
12109 (add_case_node): Remove.
12110 (emit_case_dispatch_table): Use vector instead of case_list.
12111 (reset_out_edges_aux): Remove.
12112 (compute_cases_per_edge): Likewise.
12113 (expand_case): Build list of simple_case_node.
12114 (expand_sjlj_dispatch_table): Use it.
12115 * tree-switch-conversion.c (struct case_node): Moved from
12116 stmt.c and adjusted.
12117 (emit_case_nodes): Likewise.
12118 (node_has_low_bound): Likewise.
12119 (node_has_high_bound): Likewise.
12120 (node_is_bounded): Likewise.
12121 (case_values_threshold): Likewise.
12122 (reset_out_edges_aux): Likewise.
12123 (compute_cases_per_edge): Likewise.
12124 (add_case_node): Likewise.
12125 (dump_case_nodes): Likewise.
12126 (balance_case_nodes): Likewise.
12127 (expand_switch_as_decision_tree_p): Likewise.
12128 (emit_jump): Likewise.
12129 (emit_case_decision_tree): Likewise.
12130 (try_switch_expansion): Likewise.
12131 (do_jump_if_equal): Likewise.
12132 (emit_cmp_and_jump_insns): Likewise.
12133 (fix_phi_operands_for_edge): New function.
12134 (record_phi_operand_mapping): Likewise.
12135 (class pass_lower_switch): New pass.
12136 (pass_lower_switch::execute): New function.
12137 (make_pass_lower_switch): Likewise.
12138 (conditional_probability):
12139 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12140 * tree-pass.h: Add make_pass_lower_switch.
12142 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12145 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12148 2017-08-29 Richard Biener <rguenther@suse.de>
12150 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12151 we do not add a DW_AT_inline attribute twice.
12152 (gen_subprogram_die): Remove code setting DW_AT_inline on
12153 DECL_ABSTRACT_P nodes.
12155 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
12157 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
12158 calls to internal functions.
12159 (gimplify_modify_expr): Likewise.
12160 * tree-call-cdce.c (use_internal_fn): Likewise.
12161 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
12162 (convert_to_divmod): Set the nothrow flag.
12163 * tree-if-conv.c (predicate_mem_writes): Likewise.
12164 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12165 (vectorizable_call): Likewise.
12166 (vectorizable_store): Likewise.
12167 (vectorizable_load): Likewise.
12168 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
12169 (vect_recog_mask_conversion_pattern): Likewise.
12171 2017-08-29 Martin Liska <mliska@suse.cz>
12174 * gcc.c (driver_handle_option): Add new argument.
12175 * opts-common.c (handle_option): Pass
12176 target_option_override_hook.
12177 * opts-global.c (lang_handle_option): Add new option.
12178 (set_default_handlers): Add new argument.
12179 (decode_options): Likewise.
12180 * opts.c (target_handle_option): Likewise.
12181 (common_handle_option): Call target_option_override_hook.
12182 * opts.h (struct cl_option_handler_func): Add hook for
12183 target option override.
12184 (struct cl_option_handlers): Likewise.
12185 (set_default_handlers): Add new argument.
12186 (decode_options): Likewise.
12187 (common_handle_option): Likewise.
12188 (target_handle_option): Likewise.
12189 * toplev.c (toplev::main): Pass targetm.target_option.override
12192 2017-08-29 Richard Biener <rguenther@suse.de>
12193 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
12195 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
12196 life to the active subtree.
12198 2017-08-28 Jeff Law <law@redhat.com>
12200 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
12201 derive_equivalences.
12202 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
12203 Code moved into....
12204 (edge_info::derive_equivalences): New private member function
12206 * tree-ssa-dom.c (class edge_info): Changed from a struct
12207 to a class. Add ctor/dtor, methods and data members.
12208 (edge_info::edge_info): Renamed from allocate_edge_info.
12209 Initialize additional members.
12210 (edge_info::~edge_info): New.
12211 (free_dom_edge_info): Delete the edge info.
12212 (record_edge_info): Use new class & associated member functions.
12213 Tighten forms for testing for edge equivalences.
12214 (record_temporary_equivalences): Iterate over the simple
12215 equivalences rather than assuming there's only one per edge.
12216 (cprop_into_successor_phis): Iterate over the simple
12217 equivalences rather than assuming there's only one per edge.
12218 (optimize_stmt): Use operand_equal_p rather than pointer
12219 equality for mini-DSE code.
12221 2017-08-28 Nathan Sidwell <nathan@acm.org>
12223 * gcc.c (execute): Fold SIGPIPE handling into switch
12224 statement. Adjust internal error message.
12226 2017-08-28 Richard Biener <rguenther@suse.de>
12229 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12230 Do nothing for removed DIEs.
12232 2017-08-28 Richard Biener <rguenther@suse.de>
12234 PR tree-optimization/81977
12235 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12238 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12241 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12242 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12243 using targetm.gen_mem_thread_fence.
12245 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12248 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12249 predicate to register_operand. Reorder operands.
12250 (*btr<mode>): Ditto.
12251 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12252 (*btr<mode>_mask): Ditto.
12254 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12256 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12257 * config/rs6000/xmmintrin.h: New file.
12258 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12260 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12263 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12264 parameter and_insn and return it.
12265 (recombine_lvx_pattern): Insert a copy to ensure availability of
12266 the base register of the copied masking operation at the point of
12267 the instruction replacement.
12268 (recombine_stvx_pattern): Likewise.
12270 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12272 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12273 undocumented switches.
12274 (-mpower9-dform-vector): Likewise.
12275 (-mpower9-dform): Likewise.
12276 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12277 comments to delete references to -mpower9-dform* switches.
12278 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12279 Delete reference to -mpower9-dform* switches, test for
12280 -mpower9-vector instead.
12281 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12282 (OTHER_P9_VECTOR_MASKS): Likewise.
12283 (POWERPC_MASKS): Likewise.
12284 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12285 tests against -mpower9-dform* to -mpower9-vector. Delete code
12286 that checked for -mpower9-dform* consistancy with other options.
12287 Add test for -mpower9-misc to enable other power9 switches.
12288 (rs6000_init_hard_regno_mode_ok): Likewise.
12289 (rs6000_option_override_internal): Likewise.
12290 (rs6000_emit_prologue): Likewise.
12291 (rs6000_emit_epilogue): Likewise.
12292 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12293 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12294 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12296 (emit_fusion_p9_store): Likewise.
12297 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12298 resetting these macros if the assembler does not support ISA 3.0
12300 (TARGET_P9_DFORM_VECTOR): Likewise.
12301 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12302 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12304 2017-08-25 Alan Modra <amodra@gmail.com>
12306 PR rtl-optimization/81747
12307 * cse.c (cse_extended_basic_block): Don't attempt to record
12308 equivalences for degenerate conditional branches that branch
12309 to their fall-through.
12311 2017-08-24 Martin Sebor <msebor@redhat.com>
12313 PR middle-end/81908
12314 * gimple-fold.c (size_must_be_zero_p): New function.
12315 (gimple_fold_builtin_memory_op): Call it.
12317 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12319 * config/rs6000/mm_malloc.h: New file.
12321 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12323 PR tree-optimization/81913
12324 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12325 analysis when either IVs in condition can wrap.
12327 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12329 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12330 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12332 2017-08-24 Richard Biener <rguenther@suse.de>
12335 * targhooks.c (default_target_can_inline_p): Properly
12336 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12337 is present and always compare.
12338 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12339 infer -mfpmath=sse from TARGET_SSE_P.
12340 (ix86_can_inline_p): Properly use target_option_default_node when
12341 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12343 2017-08-24 Richard Biener <rguenther@suse.de>
12346 * dwarf2out.c (output_die): Handle flag_generate_offload like
12348 (output_comp_unit): Likewise.
12349 (gen_array_type_die): Likewise.
12350 (dwarf2out_early_finish): Likewise.
12351 (note_variable_value_in_expr): Likewise.
12352 (dwarf2out_finish): Likewise. Adjust assert.
12353 * cgraphunit.c (symbol_table::compile): Move setting of
12354 flag_generate_offload earlier ...
12355 (symbol_table::finalize_compilation_unit): ... here, before
12356 early debug finalization.
12358 2017-08-24 Richard Biener <rguenther@suse.de>
12360 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12361 and ipa-fnsummary.h.
12362 (ix86_can_inline_p): When ix86_fpmath flags do not match
12363 check whether the callee uses FP math at all.
12365 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12367 PR middle-end/81931
12368 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12369 instead of TYPE_PRECISION.
12371 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12372 Alan Hayward <alan.hayward@arm.com>
12373 David Sherwood <david.sherwood@arm.com>
12375 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12377 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12379 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12381 (ix86_frame::stack_realign_allocate): New field.
12382 (struct machine_frame_state): Modify comments.
12383 (machine_frame_state::sp_realigned_fp_end): New field.
12384 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12385 layout calculation.
12386 (sp_valid_at): Add assertion to assure no attempt to access invalid
12387 offset of a realigned stack.
12388 (fp_valid_at): Likewise.
12389 (choose_baseaddr): Modify comments.
12390 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12391 ix86_expand_prologue.
12392 (ix86_expand_prologue): Modify stack realignment and allocation.
12393 (ix86_expand_epilogue): Modify comments.
12394 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12395 avx2_runtime, avx512f, and avx512f_runtime.
12397 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12399 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12400 (mstackrealign): Do not init to -1.
12401 * config/i386/i386.c (ix86_option_override_internal):
12402 Check opts_set, not opts when setting default value of
12403 opts->x_ix86_force_align_arg_pointer.
12405 2017-08-23 Richard Biener <rguenther@suse.de>
12407 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12408 lang_hooks.decl_printable_name.
12409 * print-rtl-function.c (print_rtx_function): Likewise.
12410 * tree-pretty-print.c (dump_function_header): Likewise.
12412 2017-08-23 Richard Biener <rguenther@suse.de>
12415 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12416 -g0 at compile-time.
12418 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12420 PR middle-end/19706
12421 * doc/sourcebuild.texi (Other hardware attributes):
12424 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12426 PR middle-end/19706
12427 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12428 Fix single-use check.
12430 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12432 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12434 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12436 * doc/install.texi: Modify to add more details on running selected
12439 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12441 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12442 is combined with -mabi=ms.
12443 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12446 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12448 PR tree-optimization/81488
12449 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12450 and cached_basis fields.
12451 (MAX_SPREAD): New constant.
12452 (alloc_cand_and_find_basis): Initialize new fields.
12453 (clear_visited): New function.
12454 (create_phi_basis_1): Rename from create_phi_basis, set visited
12455 and cached_basis fields.
12456 (create_phi_basis): New wrapper function.
12457 (phi_add_costs_1): Rename from phi_add_costs, add spread
12458 parameter, set visited field, short-circuit when limits reached.
12459 (phi_add_costs): New wrapper function.
12460 (record_phi_increments_1): Rename from record_phi_increments, set
12462 (record_phi_increments): New wrapper function.
12463 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12464 (phi_incr_cost): New wrapper function.
12465 (all_phi_incrs_profitable_1): Rename from
12466 all_phi_incrs_profitable, set visited field.
12467 (all_phi_incrs_profitable): New wrapper function.
12469 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12470 Alan Hayward <alan.hayward@arm.com>
12471 David Sherwood <david.sherwood@arm.com>
12473 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12474 that takes the outer and inner modes.
12475 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12476 comparison as the canonical test for a paradoxical subreg.
12477 * combine.c (simplify_set): Use paradoxical_subreg_p.
12478 (make_extraction): Likewise.
12479 (force_to_mode): Likewise.
12480 (rtx_equal_for_field_assignment_p): Likewise.
12481 (gen_lowpart_for_combine): Likewise.
12482 (simplify_comparison): Likewise.
12483 * cse.c (equiv_constant): Likewise.
12484 * expmed.c (store_bit_field_1): Likewise.
12485 * final.c (alter_subreg): Likewise.
12486 * fwprop.c (propagate_rtx): Likewise.
12487 (forward_propagate_subreg): Likewise.
12488 * ira-conflicts.c (ira_build_conflicts): Likewise.
12489 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12490 * lra-constraints.c (curr_insn_transform): Likewise.
12491 (split_reg): Likewise.
12492 * lra-eliminations.c (move_plus_up): Likewise.
12493 (lra_eliminate_regs_1): Likewise.
12494 * recog.c (general_operand): Likewise.
12495 * ree.c (combine_reaching_defs): Likewise.
12496 * reload.c (push_reload): Likewise.
12497 (find_reloads): Likewise.
12498 * reload1.c (elimination_effects): Likewise.
12499 (compute_reload_subreg_offset): Likewise.
12500 (choose_reload_regs): Likewise.
12501 * rtlanal.c (subreg_lsb_1): Likewise.
12502 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12503 (simplify_subreg): Likewise.
12504 * var-tracking.c (track_loc_p): Likewise.
12505 * emit-rtl.c (byte_lowpart_offset): Likewise.
12506 (paradoxical_subreg_p): Delete out-of-line definition.
12508 2017-08-22 Jeff Law <law@redhat.com>
12510 PR tree-optimization/81741
12511 PR tree-optimization/71947
12512 * tree-ssa-dom.c: Include tree-inline.h.
12513 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12514 equivalences if one is more expensive to compute than the other.
12515 * tree-ssa-scopedtables.h (class const_or_copies): Make
12516 record_const_or_copy_raw method private.
12517 (class avail_exprs_stack): New method simplify_binary_operation.
12518 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12519 avail_exprs_stack::simplify_binary_operation as needed.
12520 (avail_exprs_stack::simplify_binary_operation): New function.
12522 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12524 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12525 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12526 (DOT_SYMBOLS): Likewise.
12527 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12528 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12529 (RS6000_ABI_NAME): Likewise.
12530 (TARGET_CMODEL): Likewise.
12531 (TOC_SECTION_ASM_OP): Likewise.
12532 (SET_CMODEL): New macro.
12533 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12535 2017-08-22 Richard Biener <rguenther@suse.de>
12537 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12538 to allow for free-lang-data replacements similar to verify_type_variant.
12540 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12541 Alan Hayward <alan.hayward@arm.com>
12542 David Sherwood <david.sherwood@arm.com>
12544 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12545 VOIDmode for the LABEL_REF.
12547 2017-08-22 Richard Biener <rguenther@suse.de>
12549 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12551 2017-08-22 Richard Biener <rguenther@suse.de>
12553 * common.opt (feliminate-dwarf2-dups): Ignore.
12554 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12555 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12556 same_die_p_wrap, compute_section_prefix,
12557 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12558 (comdat_symbol_id, comdat_symbol_number): Likewise.
12559 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12561 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12562 (output_die): Mark unreachable path unreachable.
12563 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12564 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12565 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12566 (dwarf2out_early_finish): Likewise.
12568 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12570 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12572 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12575 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12576 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12577 (avr_attribute_table) <io, io_low, address>: Initialize
12578 .decl_required with true.
12580 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12582 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12583 undocumented debugging options.
12584 (-mvsx-scalar-double): Likewise.
12585 (-mallow-df-permute): Likewise.
12586 (-mvectorize-builtins): Likewise.
12587 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12588 (rs6000_builtin_vectorized_function): Likewise.
12589 (rs6000_builtin_md_vectorized_function): Likewise.
12590 (rs6000_opt_vars): Likewise.
12592 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12595 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12596 (*btrq_imm): Rename from *btrq.
12597 (*btcq_imm): Rename from *btcq.
12598 (btsc): New code attribute.
12599 (*<btsc><mode>): New insn pattern.
12600 (*btr<mode>): Ditto.
12601 (*<btsc><mode>_mask): New insn_and_split pattern.
12602 (*btr<mode>_mask): Ditto.
12604 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12605 Alan Hayward <alan.hayward@arm.com>
12606 David Sherwood <david.sherwood@arm.com>
12608 * function.c (pad_below): Simplify padding calculation.
12610 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12611 Alan Hayward <alan.hayward@arm.com>
12612 David Sherwood <david.sherwood@arm.com>
12614 * target.def (function_prologue): Remove frame size argument.
12615 (function_epilogue): Likewise.
12616 * doc/tm.texi: Regenerate.
12617 * final.c (final_start_function): Update call to function_prologue.
12618 (final_end_function): Update call to function_epilogue.
12619 (default_function_pro_epilogue): Remove frame size argument.
12620 * output.h (default_function_pro_epilogue): Likewise.
12621 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12622 (arm_output_function_prologue): Likewise.
12623 * config/frv/frv.c (frv_function_prologue): Likewise.
12624 (frv_function_epilogue): Likewise.
12625 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12626 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12627 (ia64_output_function_epilogue): Likewise.
12628 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12629 (m32r_output_function_epilogue): Likewise.
12630 * config/microblaze/microblaze.c (microblaze_function_prologue)
12631 (microblaze_function_epilogue): Likewise.
12632 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12633 (mips_output_function_epilogue): Likewise.
12634 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12635 (mmix_target_asm_function_epilogue): Likewise.
12636 * config/msp430/msp430.c (msp430_start_function): Likewise.
12637 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12638 (nds32_asm_function_epilogue): Likewise.
12639 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12640 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12641 (pa_output_function_epilogue): Likewise.
12642 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12643 (rs6000_output_function_epilogue): Likewise.
12644 * config/rl78/rl78.c (rl78_start_function): Likewise.
12645 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12646 (rs6000_output_function_epilogue): Likewise.
12647 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12648 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12649 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12650 (sparc_asm_function_epilogue): Likewise.
12652 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12654 * tree.h (type_has_mode_precision_p): New function.
12655 * convert.c (convert_to_integer_1): Use it.
12656 * expr.c (expand_expr_real_2): Likewise.
12657 (expand_expr_real_1): Likewise.
12658 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
12659 * match.pd: Likewise.
12660 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
12661 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
12662 * tree-tailcall.c (process_assignment): Likewise.
12663 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12664 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
12665 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
12666 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12667 (vectorizable_assignment): Likewise.
12668 (vectorizable_shift): Likewise.
12669 (vectorizable_operation): Likewise.
12670 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12672 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12674 * match.pd: Add pow (C, x) simplification.
12676 2017-08-21 Richard Biener <rguenther@suse.de>
12678 PR tree-optimization/81900
12679 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12680 for blocks with abnormal predecessors.
12681 (compute_antic): Do not set visited flag prematurely.
12683 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12686 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12688 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12690 * stor-layout.h (vector_type_mode): Move to...
12691 * tree.h (vector_type_mode): ...here.
12692 * stor-layout.c (vector_type_mode): Move to...
12693 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12695 2017-08-21 Richard Biener <rguenther@suse.de>
12697 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12698 register_external_die hooks.
12699 (debug_false_tree_charstarstar_uhwistar): Declare.
12700 (debug_nothing_tree_charstar_uhwi): Likewise.
12701 * debug.c (do_nothing_debug_hooks): Adjust.
12702 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12703 (debug_nothing_tree_charstar_uhwi): Likewise.
12704 * dbxout.c (dbx_debug_hooks): Adjust.
12705 (xcoff_debug_hooks): Likewise.
12706 * sdbout.c (sdb_debug_hooks): Likewise.
12707 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12708 * dwarf2out.c (macinfo_label_base): New global.
12709 (dwarf2out_register_external_die): New function for the
12710 register_external_die hook.
12711 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12712 (dwarf2_debug_hooks): Use them.
12713 (dwarf2_lineno_debug_hooks): Adjust.
12714 (struct die_struct): Add with_offset flag.
12715 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12716 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12717 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12718 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12719 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12720 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12721 defining section names for the early LTO debug variants.
12722 (reset_indirect_string): New helper.
12723 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12724 (print_dw_val): Add support for offsetted symbol references.
12725 (get_ultimate_context): Split out from is_cxx.
12726 (is_cxx): Use get_ultimate_context.
12727 (is_fortran): Add decl overload.
12728 (compute_comp_unit_symbol): Split out worker from
12729 compute_section_prefix.
12730 (compute_section_prefix): Call compute_comp_unit_symbol and
12731 set comdat_type_p here.
12732 (output_die): Skip DIE symbol output for the LTO added one.
12733 Handle DIE symbol references with offset.
12734 (output_comp_unit): Guard section name mangling properly.
12735 For LTO debug sections emit a symbol at the section beginning
12736 which we use to refer to its DIEs.
12737 (add_abstract_origin_attribute): For DIEs registered via
12738 dwarf2out_register_external_die directly refer to the early
12739 DIE rather than indirectly through the shadow one we created.
12740 Remove obsolete call to dwarf2out_abstract_function for
12741 non-function/block origins.
12742 (gen_array_type_die): When generating early LTO debug do
12743 not emit DW_AT_string_length.
12744 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12745 late when in LTO. As suggested place a gcc_unreachable for
12746 the DECL_ABSTRACT_P case.
12747 (gen_subprogram_die): Avoid another specification DIE
12748 for early built declarations/definitions for the late LTO case.
12749 (gen_variable_die): Add type references for late duplicated VLA dies
12751 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12752 we have the abstract instance already.
12753 (process_scope_var): Adjust decl DIE contexts in LTO which
12754 first puts them in limbo.
12755 (gen_decl_die): Do not generate type DIEs late apart from
12756 types for VLAs or for decls we do not yet have a DIE. Do not
12757 call dwarf2out_abstract_function late.
12758 (dwarf2out_early_global_decl): Make sure to create DIEs
12759 for abstract instances of a decl first.
12760 (dwarf2out_late_global_decl): Adjust comment.
12761 (output_macinfo_op): With multiple macro sections use
12762 macinfo_label_base to distinguish labels.
12763 (output_macinfo): Likewise. Update macinfo_label_base.
12764 Pass in the line info label.
12765 (note_variable_value_in_expr): When generating LTO resolve
12766 all variable values here by generating DIEs as needed.
12767 (init_sections_and_labels): Add early LTO debug flag parameter
12768 and generate different sections and names if set. Add generation
12769 counter for the labels so we can have multiple of them.
12770 (reset_dies): Helper to allow DIEs to be output multiple times.
12771 (dwarf2out_finish): When outputting DIEs to the fat part of an
12772 LTO object first reset DIEs.
12773 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12774 (modified_type_die): Check for decl_ultimate_origin being self
12776 (gen_type_die_with_usage): Likewise.
12777 (gen_typedef_die): Allow decl_ultimate_origin being self.
12778 (set_decl_abstract_flags): Remove.
12779 (set_block_abstract_flags): Likewise.
12780 (dwarf2out_abstract_function): Treat the early generated DIEs
12781 as the abstract copy and only add DW_AT_inline and
12782 DW_AT_artificial here and call set_decl_origin_self.
12783 If the DIE has an abstract origin don't do anything.
12784 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12785 if we have none yet (Go fails to build one, PR78628).
12786 (variably_modified_type_p): Prevent endless recursion for Ada
12787 cyclic pointer types.
12788 * lto-streamer-in.c: Include debug.h.
12789 (dref_queue): New global.
12790 (lto_read_tree_1): Stream in DIE references.
12791 (lto_input_tree): Register DIE references.
12792 (input_function): Stream DECL_DEBUG_ARGS.
12793 * lto-streamer-out.c: Include debug.h.
12794 (lto_write_tree_1): Output DIE references.
12795 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12796 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12797 (output_function): Stream DECL_DEBUG_ARGS.
12798 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12799 Stream DECL_ABSTRACT_ORIGIN.
12800 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12801 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12802 DECL_CONTEXT for file-scope decls.
12803 * lto-streamer.h (struct dref_entry): Declare.
12804 (dref_queue): Likewise.
12805 * cfgexpand.c (pass_expand::execute): Do not call the
12806 outlining_inline_function hook here.
12807 * lto-wrapper.c (debug_obj): New global.
12808 (tool_cleanup): Unlink it if required.
12809 (debug_objcopy): New function.
12810 (run_gcc): Handle early debug sections in the IL files by
12811 extracting them to separate files, partially linkin them and
12812 feeding the result back as result to the linker.
12813 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12814 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12815 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12816 sections into a separate segment.
12817 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12819 (darwin_asm_dwarf_section): Likewise.
12820 (darwin_asm_output_dwarf_offset): Likewise.
12821 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12823 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12824 Alan Hayward <alan.hayward@arm.com>
12825 David Sherwood <david.sherwood@arm.com>
12827 * read-md.h (md_reader::record_potential_iterator_use): Replace
12828 pointer argument with an rtx and an index.
12829 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12830 (apply_mode_iterator): Likewise.
12831 (apply_code_iterator): Likewise.
12832 (apply_int_iterator): Likewise.
12833 (apply_subst_iterator): Likewise.
12834 (record_iterator_use): Likewise.
12835 (record_attribute_use): Likewise.
12836 (md_reader::record_potential_iterator_use): Likewise. Update calls
12837 to record_iterator_use and apply_iterator.
12838 (iterator_use): Replace ptr with x and index.
12839 (attribute_use): Likewise.
12840 (apply_attribute_uses): Update calls to apply_iterator.
12841 (apply_iterators): Likewise. Update initialization of iterator_use.
12842 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12843 and record_potential_iterator_use.
12844 (rtx_reader::read_rtx_operand): Likewise.
12846 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12847 Alan Hayward <alan.hayward@arm.com>
12848 David Sherwood <david.sherwood@arm.com>
12850 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12853 2017-08-21 Richard Biener <rguenther@suse.de>
12855 PR middle-end/81884
12856 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12857 at struct end conservatively when comparing common bases.
12859 2017-08-21 Richard Biener <rguenther@suse.de>
12861 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12862 (mem_ref_in_stmt): Remove.
12863 (determine_max_movement): Use ref index to get at the reference.
12864 (invariantness_dom_walker::before_dom_children): Deal with
12865 lim data already initialized.
12866 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12868 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12870 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12871 (TARGET_ISA_ROUND): Ditto.
12872 (TARGET_ROUND): Ditto.
12873 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12874 * config/i386/i386.md: Ditto.
12875 * config/i386/sse.md: Ditto.
12876 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12877 with OPTION_MASK_ISA_SSE4_1.
12879 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12882 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12883 __builtin_ia32_lzcnt_u16.
12885 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12888 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12889 (rs6000_set_current_function): Rewrite function to use it.
12891 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12894 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12895 and TYPE_WARN_IF_NOT_ALIGN.
12896 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12897 (handle_warn_if_not_align): New.
12898 (place_union_field): Call handle_warn_if_not_align.
12899 (place_field): Call handle_warn_if_not_align.
12900 Copy TYPE_WARN_IF_NOT_ALIGN.
12901 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12902 (layout_type): Likewise.
12903 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12905 (tree_decl_common): Add warn_if_not_align.
12906 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12907 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12908 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12909 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12910 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12911 * doc/extend.texi: Document warn_if_not_aligned attribute.
12912 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12914 2017-08-17 Martin Liska <mliska@suse.cz>
12917 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12918 (get_data_dependence): Use it as pointer type.
12919 (distribute_loop): Likewise.
12921 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12923 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12924 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12925 (p8_vmrgow_<mode>_direct): New define_insn.
12926 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12927 handle endianness for vmrgew and vmrgow permute patterns.
12929 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12931 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12932 * config/rs6000/rs6000-cpus.def: Remove comment.
12933 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12934 (POWERPC_MASKS): Likewise.
12935 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12936 use of TARGET_VSX_TIMODE.
12937 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12938 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12939 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12940 (rs6000_option_override_internal): Remove dead code.
12941 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12942 (rs6000_legitimize_reload_address): Likewise.
12943 (rs6000_legitimate_address_p): Likewise.
12944 (rs6000_opt_masks): Delete "vsx-timode".
12945 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12946 from function comment.
12947 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12948 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12949 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12951 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12952 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12953 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12954 (VSX_TI): Likewise.
12956 (define_peephole2): Likewise.
12958 2017-08-17 Martin Sebor <msebor@redhat.com>
12961 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12962 past the end of an array.
12963 (test_pp_format): Add test cases.
12965 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
12967 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
12968 missing ECF_NOTHROW flags.
12970 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12973 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
12974 operands residing in integer registers.
12975 (*vsx_le_perm_load_<mode>): Likewise.
12976 (*vsx_le_perm_store_<mode>): Likewise.
12977 (define_peephole2): Add peepholes to optimize the above.
12979 2017-08-17 Marek Polacek <polacek@redhat.com>
12981 PR middle-end/81814
12982 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
12983 to mimic what shorten_compare did. Change the return type to bool.
12984 (fold_cond_expr_with_comparison): Update call to
12985 operand_equal_for_comparison_p.
12986 (fold_ternary_loc): Likewise.
12988 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
12990 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
12992 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
12994 2017-08-17 Richard Biener <rguenther@suse.de>
12996 * tree-ssa-structalias.c (solve_graph): When propagating
12997 to successors update the graphs succ edges and avoid duplicate work.
12999 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13002 * config/i386/i386.c (ix86_option_override_internal): Save target
13003 specific options after ix86_stack_protector_guard_reg was changed.
13005 2017-08-17 Richard Biener <rguenther@suse.de>
13007 PR tree-optimization/81827
13008 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13009 (new_var_info): Initialize it conservatively.
13010 (get_call_vi): Mark register vars.
13011 (new_scalar_tmp_constraint_exp): Likewise.
13012 (handle_rhs_call): Likewise.
13013 (handle_const_call): Likewise.
13014 (create_function_info_for): Likewise.
13015 (solve_constraints): Sort varinfos to separate register from
13016 non-register vars to pack points-to solution bitmaps during
13019 2017-08-17 Marek Polacek <polacek@redhat.com>
13021 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13023 2017-08-17 Richard Biener <rguenther@suse.de>
13025 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13026 to true when overflow is undefined and we saturated the result.
13028 2017-08-17 Alan Modra <amodra@gmail.com>
13031 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13032 Don't use store multiple if only one reg needs saving.
13033 (interesting_frame_related_regno): New function.
13034 (rs6000_frame_related): Don't emit frame info for regs that
13036 (rs6000_emit_epilogue): Likewise.
13038 2017-08-16 Nathan Sidwell <nathan@acm.org>
13040 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13041 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13042 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13043 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13044 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13045 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13046 (verify_type): Adjust for TYPE_BINFO move.
13047 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13048 process TYPE_BINFO directly.
13049 (hash_tree): Likewise.
13050 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13052 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13055 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13057 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13059 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13062 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13063 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13064 Add ix86_binary_operator_ok to insn constraint.
13065 (*iordi_1_bts): Ditto.
13066 (*xordi_1_btc): Ditto.
13067 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13068 Update corresponding peephole2 pattern.
13072 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13074 PR tree-optimization/81832
13075 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13076 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13078 2017-08-16 Marek Polacek <polacek@redhat.com>
13081 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13082 regarding size_zero_node.
13084 2017-08-16 Martin Liska <mliska@suse.cz>
13087 * config.gcc: Respect previously set extra_objs in case
13090 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13092 PR tree-optimization/81835
13093 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13094 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13095 not depend on the phi.
13097 2017-08-16 Alan Modra <amodra@gmail.com>
13099 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13102 2017-08-16 Alan Modra <amodra@gmail.com>
13104 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13105 (save_reg_p): ..into this. Update all callers.
13106 (first_reg_to_save): Simplify.
13108 2017-08-16 Alan Modra <amodra@gmail.com>
13110 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13113 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13117 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13118 constant count if that count is less than 32.
13120 2017-08-15 Nathan Sidwell <nathan@acm.org>
13122 * gcc.c (execute): Emit friendlier message if inferior is killed
13123 by an external cause.
13125 2017-08-15 Richard Biener <rguenther@suse.de>
13127 PR tree-optimization/81790
13128 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13129 CONSTRUCTORs from simplifying and VN.
13131 2017-08-14 Martin Sebor <msebor@redhat.com>
13133 * builtin-attrs.def: Add comments.
13135 2017-08-14 Martin Sebor <msebor@redhat.com>
13138 * doc/extend.texi (attribute nonstring): Document new attribute.
13140 2017-08-14 Martin Sebor <msebor@redhat.com>
13143 * tree-diagnostic.c (default_tree_printer): Handle %G.
13144 * gimple-pretty-print.h (percent_G_format): Declare new function.
13145 * gimple-pretty-print.c (percent_G_format): Define.
13146 * tree-pretty-print.c (percent_K_format): Add argument.
13148 2017-08-14 Martin Sebor <msebor@redhat.com>
13150 PR translation/79998
13151 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13152 Remove a stray space.
13154 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
13157 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
13158 (*iordi_1_bts): Ditto.
13159 (*xordi_1_btc): Ditto.
13161 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13164 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
13165 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13167 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
13168 quoted strings, and make more translator-friendly.
13169 (darwin_rs6000_override_options): Likewise.
13170 (rs6000_option_override_internal): Likewise.
13171 (rs6000_return_in_memory): Fix overlong line.
13172 (init_cmulative_args): Use quoted strings, and make more
13173 translator-friendly.
13174 (rs6000_pass_by_reference): Fix overlong line.
13175 (def_builtin): Use quoted strings.
13176 (altivec_expand_predicate_builtin): Use quoted strings, and make
13177 more translator-friendly.
13178 (htm_expand_builtin): Use quoted strings.
13179 (cpu_expand_builtin): Use quoted strings, and make more
13180 translator-friendly.
13181 (altivec_expand_builtin): Likewise.
13182 (paired_expand_predicate_builtin): Likewise.
13183 (rs6000_invalid_builtin): Likewise.
13184 (builtin_function_type): Use quoted strings.
13185 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
13186 more translator-friendly.
13187 (rs6000_trampoline_init): Likewise.
13188 (rs6000_handle_altivec_attribute): Likewise.
13189 (rs6000_inner_target_options): Use quoted strings.
13190 (rs6000_disable_incompatible_switches): Likewise.
13191 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
13192 strings, and make more translator-friendly.
13193 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
13195 2017-08-14 Bin Cheng <bin.cheng@arm.com>
13197 PR tree-optimization/81799
13198 * tree-loop-distribution.c (version_loop_by_alias_check): Force
13199 cond_expr to simple gimple operand.
13201 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
13203 PR middle-end/46932
13204 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
13206 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
13210 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
13211 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
13212 TARGET_GASISR_PROLOGUES.
13213 * config/avr/avr.c (avr_option_override): Same.
13214 (avr_pass_pre_proep::execute): Same.
13216 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13219 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13220 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13222 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13224 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13225 define_insn_and_split. Split to a memory load from 0 in
13226 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13227 using PTR mode iterator.
13228 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13229 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13230 (*add_tp_<mode>): Redefine as define_insn_and_split.
13231 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13232 address space. Merge with *add_tp_x32 using PTR mode iterator.
13233 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13234 Split to an add with a memory load from 0 in
13235 DEFAULT_TLS_SEG_REG address space.
13237 2017-08-12 Andrew Pinski <apinski@cavium.com>
13239 * config/aarch64/aarch64-option-extensions.def (rdma):
13240 Fix feature string to what Linux prints out in /proc/cpuinfo.
13242 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13245 * dwarf2out.c (modified_type_die): For C typedef types that have
13246 an ultimate origin, process the ultimate origin instead of the
13248 (gen_typedef_die): Assert that input DECLs have no ultimate
13250 (gen_type_die_with_usage): For typedef variants that have an
13251 ultimate origin, just call gen_decl_die on the original DECL.
13252 (process_scope_var): Avoid creating DIEs for local typedefs and
13253 concrete static variables.
13255 2017-08-12 Alan Modra <amodra@gmail.com>
13259 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13260 match gnu-user.h startfile.
13261 (ENDFILE_LINUX_SPEC): Similarly.
13263 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13266 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13268 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13270 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13271 * config/aarch64/aarch64.md (mov<mode>): Change.
13272 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13273 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13274 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13276 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13278 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13279 for storage order barriers.
13281 2017-08-11 Martin Liska <mliska@suse.cz>
13284 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13285 variables of void type.
13287 2017-08-11 Martin Liska <mliska@suse.cz>
13289 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13290 TARGET_SUPPORTS_ALIASES.
13291 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13292 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13293 (optimize_weakref): Likewise.
13294 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13295 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13296 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13298 2017-08-11 Martin Liska <mliska@suse.cz>
13301 * config/i386/i386.c (make_resolver_func): Do complete
13302 refactoring of the function.
13304 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13307 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13308 * config/i386/i386.c (ix86_stack_protect_guard): Use
13309 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13310 * doc/invoke.texi (x86 Options): Document
13311 -mstack-protector-guard-symbol= option.
13313 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13315 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13316 * config/i386/i386.c (ix86_split_stack_guard): New function.
13317 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13318 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13319 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13320 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13321 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13322 (split_stack_space_check): Call ix86_split_stack_guard.
13324 2017-08-10 Martin Sebor <msebor@redhat.com>
13326 * print-tree.c (print_node): Print location using the established
13328 Replace spaces with colons.
13329 (debug_raw, debug): Ditto.
13331 2017-08-10 Martin Sebor <msebor@redhat.com>
13334 * pretty-print.c (pp_format): Correct the handling of %s precision.
13336 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13339 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13341 (ix86_finalize_stack_frame_flags): This. Also clear
13342 frame_pointer_needed if -fno-omit-frame-pointer is used without
13344 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13345 with ix86_finalize_stack_frame_flags.
13346 (ix86_expand_epilogue): Likewise.
13347 (ix86_expand_split_stack_prologue): Likewise.
13348 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13350 2017-08-10 Martin Liska <mliska@suse.cz>
13353 * c-attribs.c (handle_target_attribute):
13354 Report warning for an empty string argument of target attribute.
13356 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13359 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13361 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13362 or DECL_NONLOCAL labels.
13363 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13364 or DECL_NONLOCAL labels here.
13366 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13368 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13369 to indicate when early gimple folding has been disabled.
13370 (rs6000_gimple_fold_builtin): Add debug content.
13371 (rs6000_invalid_builtin): Fix whitespace.
13372 (rs6000_expand_builtin): Fix whitespace.
13373 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13375 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13378 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13379 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13381 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13383 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13384 (qdf24xx): Likewise.
13385 * config/aarch64/aarch64.md: Include falkor.md.
13386 * config/aarch64/falkor.md: New.
13388 2017-08-09 Marek Polacek <polacek@redhat.com>
13391 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13392 * diagnostic.c (emit_diagnostic): Add a comment.
13393 (emit_diagnostic_valist): New function.
13395 2017-08-09 Marek Polacek <polacek@redhat.com>
13398 * input.c (make_location): New overload.
13399 * input.h (make_location): Declare.
13401 2017-08-08 Alan Modra <amodra@gmail.com>
13402 H.J. Lu <hongjiu.lu@intel.com>
13405 * gcc.c (NO_PIE_SPEC): Delete.
13406 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13408 (LINK_PIE_SPEC): ..to here.
13409 (LINK_COMMAND_SPEC): Support -no-pie.
13410 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13411 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13413 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13414 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13415 (ENDFILE_CRTEND_SPEC): Similarly.
13417 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13420 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13421 (mstack-protector-guard-offset=): Ditto.
13422 * config/i386/i386.c (ix86_option_override): Handle
13423 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13425 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13426 ix86_stack_protect_guard_offset variables.
13427 (TARGET_STACK_PROTECT_GUARD): Always define.
13428 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13429 and -mstack-protector-guard-offset= options.
13431 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13433 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13434 boundary case for the last candidate.
13436 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13438 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13439 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13441 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13443 PR middle-end/19706
13444 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13445 * config/aarch64/aarch64-builtins.c
13446 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13447 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13448 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13450 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13451 Andrew Pinski <pinskia@gmail.com>
13453 PR middle-end/19706
13454 * internal-fn.def (XORSIGN): New.
13455 * optabs.def (xorsign_optab): New.
13456 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13457 (convert_expand_mult_copysign): New.
13458 (pass_optimize_widening_mul::execute): Call
13459 convert_expand_mult_copysign.
13461 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13463 PR tree-optimization/81354
13464 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13465 Insert on edges rather than explicitly creating landing pads.
13466 (analyze_candidates_and_replace): Commit edge inserts.
13468 2017-08-08 Richard Biener <rguenther@suse.de>
13470 PR middle-end/81719
13471 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13472 (expand_simple_operations): Also look through ADDR_EXPRs with
13473 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13475 2017-08-08 Richard Biener <rguenther@suse.de>
13477 PR tree-optimization/81723
13478 * tree-vect-slp.c (struct bst_traits): New hash traits.
13479 (bst_fail): New global.
13480 (vect_build_slp_tree_2): New worker, split out from ...
13481 (vect_build_slp_tree): ... this now wrapping it with using
13482 bst_fail set to cache SLP tree build fails. Properly handle
13484 (vect_analyze_slp_instance): Allocate and free bst_fail.
13486 2017-08-08 Martin Liska <mliska@suse.cz>
13489 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13490 LABEL_DECLs that can be from a different function.
13492 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13494 PR tree-optimization/81744
13495 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13496 loop's number of iterations.
13498 2017-08-08 Martin Liska <mliska@suse.cz>
13500 * asan.c: Include header files.
13501 * attribs.c (build_decl_attribute_variant): New function moved
13503 (build_type_attribute_qual_variant): Likewise.
13504 (cmp_attrib_identifiers): Likewise.
13505 (simple_cst_list_equal): Likewise.
13506 (omp_declare_simd_clauses_equal): Likewise.
13507 (attribute_value_equal): Likewise.
13508 (comp_type_attributes): Likewise.
13509 (build_type_attribute_variant): Likewise.
13510 (lookup_ident_attribute): Likewise.
13511 (remove_attribute): Likewise.
13512 (merge_attributes): Likewise.
13513 (merge_type_attributes): Likewise.
13514 (merge_decl_attributes): Likewise.
13515 (merge_dllimport_decl_attributes): Likewise.
13516 (handle_dll_attribute): Likewise.
13517 (attribute_list_equal): Likewise.
13518 (attribute_list_contained): Likewise.
13519 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13520 (lookup_attribute_by_prefix): Likewise.
13521 * bb-reorder.c: Include header files.
13522 * builtins.c: Likewise.
13523 * calls.c: Likewise.
13524 * cfgexpand.c: Likewise.
13525 * cgraph.c: Likewise.
13526 * cgraphunit.c: Likewise.
13527 * convert.c: Likewise.
13528 * dwarf2out.c: Likewise.
13529 * final.c: Likewise.
13530 * fold-const.c: Likewise.
13531 * function.c: Likewise.
13532 * gimple-expr.c: Likewise.
13533 * gimple-fold.c: Likewise.
13534 * gimple-pretty-print.c: Likewise.
13535 * gimple.c: Likewise.
13536 * gimplify.c: Likewise.
13537 * hsa-common.c: Likewise.
13538 * hsa-gen.c: Likewise.
13539 * internal-fn.c: Likewise.
13540 * ipa-chkp.c: Likewise.
13541 * ipa-cp.c: Likewise.
13542 * ipa-devirt.c: Likewise.
13543 * ipa-fnsummary.c: Likewise.
13544 * ipa-inline.c: Likewise.
13545 * ipa-visibility.c: Likewise.
13547 * lto-cgraph.c: Likewise.
13548 * omp-expand.c: Likewise.
13549 * omp-general.c: Likewise.
13550 * omp-low.c: Likewise.
13551 * omp-offload.c: Likewise.
13552 * omp-simd-clone.c: Likewise.
13553 * opts-global.c: Likewise.
13554 * passes.c: Likewise.
13555 * predict.c: Likewise.
13556 * sancov.c: Likewise.
13557 * sanopt.c: Likewise.
13558 * symtab.c: Likewise.
13559 * toplev.c: Likewise.
13560 * trans-mem.c: Likewise.
13561 * tree-chkp.c: Likewise.
13562 * tree-eh.c: Likewise.
13563 * tree-into-ssa.c: Likewise.
13564 * tree-object-size.c: Likewise.
13565 * tree-parloops.c: Likewise.
13566 * tree-profile.c: Likewise.
13567 * tree-ssa-ccp.c: Likewise.
13568 * tree-ssa-live.c: Likewise.
13569 * tree-ssa-loop.c: Likewise.
13570 * tree-ssa-sccvn.c: Likewise.
13571 * tree-ssa-structalias.c: Likewise.
13572 * tree-ssa.c: Likewise.
13573 * tree-streamer-in.c: Likewise.
13574 * tree-vectorizer.c: Likewise.
13575 * tree-vrp.c: Likewise.
13576 * tsan.c: Likewise.
13577 * ubsan.c: Likewise.
13578 * varasm.c: Likewise.
13579 * varpool.c: Likewise.
13580 * tree.c: Remove functions moved to attribs.[ch].
13581 * tree.h: Likewise.
13582 * config/aarch64/aarch64.c: Add attrs.h header file.
13583 * config/alpha/alpha.c: Likewise.
13584 * config/arc/arc.c: Likewise.
13585 * config/arm/arm.c: Likewise.
13586 * config/avr/avr.c: Likewise.
13587 * config/bfin/bfin.c: Likewise.
13588 * config/c6x/c6x.c: Likewise.
13589 * config/cr16/cr16.c: Likewise.
13590 * config/cris/cris.c: Likewise.
13591 * config/darwin.c: Likewise.
13592 * config/epiphany/epiphany.c: Likewise.
13593 * config/fr30/fr30.c: Likewise.
13594 * config/frv/frv.c: Likewise.
13595 * config/ft32/ft32.c: Likewise.
13596 * config/h8300/h8300.c: Likewise.
13597 * config/i386/winnt.c: Likewise.
13598 * config/ia64/ia64.c: Likewise.
13599 * config/iq2000/iq2000.c: Likewise.
13600 * config/lm32/lm32.c: Likewise.
13601 * config/m32c/m32c.c: Likewise.
13602 * config/m32r/m32r.c: Likewise.
13603 * config/m68k/m68k.c: Likewise.
13604 * config/mcore/mcore.c: Likewise.
13605 * config/microblaze/microblaze.c: Likewise.
13606 * config/mips/mips.c: Likewise.
13607 * config/mmix/mmix.c: Likewise.
13608 * config/mn10300/mn10300.c: Likewise.
13609 * config/moxie/moxie.c: Likewise.
13610 * config/msp430/msp430.c: Likewise.
13611 * config/nds32/nds32-isr.c: Likewise.
13612 * config/nds32/nds32.c: Likewise.
13613 * config/nios2/nios2.c: Likewise.
13614 * config/nvptx/nvptx.c: Likewise.
13615 * config/pa/pa.c: Likewise.
13616 * config/pdp11/pdp11.c: Likewise.
13617 * config/powerpcspe/powerpcspe.c: Likewise.
13618 * config/riscv/riscv.c: Likewise.
13619 * config/rl78/rl78.c: Likewise.
13620 * config/rx/rx.c: Likewise.
13621 * config/s390/s390.c: Likewise.
13622 * config/sh/sh.c: Likewise.
13623 * config/sol2.c: Likewise.
13624 * config/sparc/sparc.c: Likewise.
13625 * config/spu/spu.c: Likewise.
13626 * config/stormy16/stormy16.c: Likewise.
13627 * config/tilegx/tilegx.c: Likewise.
13628 * config/tilepro/tilepro.c: Likewise.
13629 * config/v850/v850.c: Likewise.
13630 * config/vax/vax.c: Likewise.
13631 * config/visium/visium.c: Likewise.
13632 * config/xtensa/xtensa.c: Likewise.
13634 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13637 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13638 constraints since the -mupper-regs-* switches have been
13640 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13641 into a vector from a double word element that was extracted from
13642 another vector, and eliminate extra XXPERMDI instructions.
13643 (vsx_concat_<mode>_2): Likewise.
13644 (vsx_concat_<mode>_3): Likewise.
13645 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13646 concat to allow optimizing inserts from previous extracts.
13648 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13650 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13651 memory reference to a SSP offset in TLS address space.
13652 (ix86_print_operand) <case '@'>: Remove.
13653 (ix86_print_operand_punct_valid_p): Remove '@' code.
13654 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
13655 UNSPEC_SP_TLS_TEST.
13656 (stack_tls_protect_set_<mode>): Remove.
13657 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
13658 (stack_tls_protect_test_<mode>): Remove.
13659 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
13661 2017-08-07 Olivier Hainque <hainque@adacore.com>
13664 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
13666 2017-08-07 Douglas Rupp <rupp@adacore.com>
13668 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
13669 variable was referenced as multidir in command.
13671 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13674 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13677 2017-08-07 Martin Liska <mliska@suse.cz>
13679 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13680 * config/rl78/rl78.c: Add include of attribs.h.
13681 * config/sh/sh.c: Likewise.
13682 * config/v850/v850.c: Likewise.
13684 2017-08-07 Tom de Vries <tom@codesourcery.com>
13686 PR middle-end/78266
13687 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13689 2017-08-07 Martin Liska <mliska@suse.cz>
13691 * config/mips/mips.c: Include attribs.h.
13693 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13696 * doc/invoke.texi: Document change in behvaior for -Ofast for
13699 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13701 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13703 (aarch64_pop_regs): Likewise.
13704 (aarch64_gen_load_pair): Likewise.
13705 (aarch64_save_callee_saves): Likewise.
13706 (aarch64_restore_callee_saves): Likewise.
13708 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13710 * config/i386/i386.c: Revert the last change.
13712 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13715 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13717 (ix86_finalize_stack_frame_flags): This. Also clear
13718 frame_pointer_needed if -fno-omit-frame-pointer is used without
13720 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13721 with ix86_finalize_stack_frame_flags.
13722 (ix86_expand_epilogue): Likewise.
13723 (ix86_expand_split_stack_prologue): Likewise.
13725 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13728 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13729 to P_AES for Westmere.
13731 2017-08-07 Jonathan Yong <10walls@gmail.com>
13733 * config/i386/mingw.opt (fset-stack-executable): Removed.
13734 * config/i386/cygming.opt (fset-stack-executable): Moved
13736 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13738 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13740 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13742 2017-08-07 Marek Polacek <polacek@redhat.com>
13744 PR middle-end/81737
13745 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13747 2017-08-07 Martin Liska <mliska@suse.cz>
13749 * attribs.h (canonicalize_attr_name): New function.
13750 (cmp_attribs): Move from c-format.c and adjusted.
13751 (is_attribute_p): Moved from tree.h.
13752 * tree-inline.c: Add new includes.
13753 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13754 (private_is_attribute_p): Remove.
13755 (private_lookup_attribute): Likewise.
13756 (private_lookup_attribute_by_prefix): Simplify.
13757 (remove_attribute): Use is_attribute_p.
13758 * tree.h: Remove removed declarations.
13760 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13762 PR middle-end/81698
13763 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13764 instead of computing it in the function. Formatting fix.
13765 (expand_case): Don't rely on default_edge being the first edge,
13766 clear it if removing it, pass default_edge to
13767 emit_case_dispatch_table.
13768 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13771 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13773 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13774 insn in the function, emit NOP after the insn.
13776 2017-08-06 Tom de Vries <tom@codesourcery.com>
13778 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13781 2017-08-06 Tom de Vries <tom@codesourcery.com>
13783 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13786 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13788 PR tree-optimization/57371
13789 * match.pd: New pattern.
13791 2017-08-04 Marek Polacek <polacek@redhat.com>
13793 PR middle-end/81695
13794 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13795 perform the computation in offset_int.
13797 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13799 PR tree-optimization/81136
13800 * tree-vectorizer.h: Include tree-hash-traits.h.
13801 (vec_base_alignments): New typedef.
13802 (vec_info): Add a base_alignments field.
13803 (vect_record_base_alignments): Declare.
13804 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13806 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13807 (create_data_ref): Add an is_conditional_in_stmt argument.
13808 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13809 the is_conditional_in_stmt field.
13810 (data_ref_loc): Add an is_conditional_in_stmt field.
13811 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13812 (find_data_references_in_stmt): Update call to create_data_ref.
13813 (graphite_find_data_references_in_stmt): Likewise.
13814 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13815 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13816 (vect_record_base_alignment): New function.
13817 (vect_record_base_alignments): Likewise.
13818 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13819 for nested statements even if we fail to compute a misalignment.
13820 Use pooled base alignments for unconditional references.
13821 (vect_find_same_alignment_drs): Compare base addresses instead
13823 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13824 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13826 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13828 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13829 Add an explicit name for the enum. Use auto_vec for slp_instances
13830 and grouped_stores.
13831 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13833 (_bb_vec_info): Add a constructor and destructor.
13834 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13835 (destroy_loop_vec_info): Delete.
13836 (vect_destroy_datarefs): Likewise.
13837 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13838 (vec_info::vec_info): New function.
13839 (vec_info::~vec_info): Likewise.
13840 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13841 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13842 destroy_loop_vec_info.
13843 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13844 (_loop_vec_info::_loop_vec_info): ...this.
13845 (destroy_loop_vec_info): Replace with...
13846 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13847 the stmt_vec_infos. Leave handling of vec_info information to its
13848 destructor. Remove explicit vector releases.
13849 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13850 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13851 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13852 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13853 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13854 (destroy_bb_vec_info): Replace with...
13855 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13856 information to its destructor.
13857 (vect_slp_analyze_bb_1): Use new and delete instead of
13858 new_bb_vec_info and destroy_bb_vec_info.
13859 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13862 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13864 * tree-data-ref.h (subscript): Add access_fn field.
13865 (data_dependence_relation): Add could_be_independent_p.
13866 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13867 (same_access_functions): Move to tree-data-ref.c.
13868 * tree-data-ref.c (ref_contains_union_access_p): New function.
13869 (access_fn_component_p): Likewise.
13870 (access_fn_components_comparable_p): Likewise.
13871 (dr_analyze_indices): Add a reference to access_fn_component_p.
13872 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13874 (constant_access_functions): Likewise.
13875 (add_other_self_distances): Likewise.
13876 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13877 (initialize_data_dependence_relation): Use XCNEW and remove
13878 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13879 of access functions that have the same type. Allow the
13880 subsequence to end with different bases in some circumstances.
13881 Record the chosen access functions in SUB_ACCESS_FN.
13882 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13883 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13884 (subscript_dependence_tester_1): Likewise dra and drb.
13885 (build_classic_dist_vector): Update calls accordingly.
13886 (subscript_dependence_tester): Likewise.
13887 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13888 DDR_COULD_BE_INDEPENDENT_P.
13889 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13890 comp_alias_ddrs instead of may_alias_ddrs.
13891 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13893 (vect_analyze_data_ref_dependence): Use it if
13894 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13895 distance vectors if that fails.
13896 (dependence_distance_ge_vf): New function.
13897 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13898 LOOP_VINFO_MAY_ALIAS_DDRS.
13900 2017-08-04 Richard Biener <rguenther@suse.de>
13902 PR middle-end/81705
13903 * fold-const.c (fold_binary_loc): Properly restrict
13904 minus_var0 && minus_var1 case when associating undefined overflow
13907 2017-08-04 Tom de Vries <tom@codesourcery.com>
13909 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13911 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13913 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13914 Don't start diagnostic messages with a capital letter.
13915 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13917 (rs6000_invalid_builtin): Likewise.
13918 (rs6000_trampoline_init): Likewise.
13920 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13923 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13924 after setting changeable df flags.
13926 2017-08-03 Richard Biener <rguenther@suse.de>
13928 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13929 up if the use is in USE - X.
13931 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13933 * toplev.c (dumpfile.h): New include.
13934 (internal_error_reentered): New static function. Use it...
13935 (internal_error_function): ...here to handle reentered internal_error.
13937 2017-08-03 Richard Biener <rguenther@suse.de>
13939 PR middle-end/81148
13940 * fold-const.c (split_tree): Add minus_var and minus_con
13941 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13942 here but always use minus_*.
13943 (associate_trees): Assert we never associate with MINUS_EXPR
13944 and NULL first operand. Do not recurse for PLUS_EXPR operands
13945 when associating as MINUS_EXPR either.
13946 (fold_binary_loc): Track minus_var and minus_con.
13948 2017-08-03 Tom de Vries <tom@codesourcery.com>
13951 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13952 ACCEL_COMPILER, apply finish_options on
13953 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13955 2017-08-03 Tom de Vries <tom@codesourcery.com>
13958 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13959 function_entry_patch_area_size > 0.
13961 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13964 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13965 instead of 10??LU, perform unit multiplication in wide_int,
13966 don't change alloc_object_size_limit if the limit is larger
13969 PR tree-optimization/81655
13970 PR tree-optimization/81588
13971 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
13972 the case when ranges[i].low and high are 1 for unsigned type with
13975 PR middle-end/81052
13976 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
13977 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
13979 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13981 * tree-vrp.h: Add include guard.
13983 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
13986 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
13987 (ud2): New insn pattern.
13988 * config/i386/i386.c (ix86_expand_epilogue):
13989 For naked functions, generate ud2 instead of trap insn.
13991 2017-08-02 Marek Polacek <polacek@redhat.com>
13994 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
13996 2017-08-02 Tom de Vries <tom@codesourcery.com>
13997 Cesar Philippidis <cesar@codesourcery.com>
13999 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14000 Add missing edge probabilities.
14002 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14004 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14005 Correct endianness.
14007 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14009 PR middle-end/79499
14010 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14011 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14012 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14013 prologue_seq sequences - if any.
14015 2017-08-02 Richard Biener <rguenther@suse.de>
14017 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14018 via vectors if supported, integer extracts via punning if supported
14019 or otherwise vector extracts.
14021 2017-08-02 Richard Biener <rguenther@suse.de>
14023 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14025 (bitmap_insert_into_set): ... this.
14027 2017-08-02 Richard Biener <rguenther@suse.de>
14029 PR tree-optimization/81633
14031 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14033 PR tree-optimization/71752
14034 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14036 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14038 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14039 (machine_function::call_ms2sysv_pad_out): Remove field.
14040 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14041 (ix86_compute_frame_layout): Likewise.
14043 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14046 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14047 attribute with interrupt attribute.
14049 2017-08-01 Andrew Pinski <apinski@cavium.com>
14051 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14052 BIT_INSERT_EXPR's operand 1
14053 to see if the types precision matches.
14055 2017-08-01 Martin Liska <mliska@suse.cz>
14057 PR middle-end/70140
14058 * builtins.c (expand_builtin_memcpy_args): Remove.
14059 (expand_builtin_memcpy): Call newly added function
14060 expand_builtin_memory_copy_args.
14061 (expand_builtin_memcpy_with_bounds): Likewise.
14062 (expand_builtin_mempcpy): Remove last argument.
14063 (expand_builtin_mempcpy_with_bounds): Likewise.
14064 (expand_builtin_memory_copy_args): New function created from
14065 expand_builtin_mempcpy_args with small modifications.
14066 (expand_builtin_mempcpy_args): Remove.
14067 (expand_builtin_stpcpy): Remove unused argument.
14068 (expand_builtin): Likewise.
14069 (expand_builtin_with_bounds): Likewise.
14071 2017-08-01 Martin Liska <mliska@suse.cz>
14074 Make mempcpy more optimal (PR middle-end/70140).
14076 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14079 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14080 __builtin_vec_cmpne verify both arguments are compatible vectors
14081 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14082 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14083 move computation of aligned to after checking the argument types.
14087 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14088 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14091 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14092 Doug Rupp <rupp@adacore.com>
14093 Olivier Hainque <hainque@adacore.com>
14095 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14096 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14098 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14099 for TARGET_OS_CPP_BUILTIN.
14100 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14101 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14103 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14104 passing required abi options to the assembler for EABI configurations.
14105 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14106 of .text.hot and .text.unlikely sections for kernel modules when
14107 using ARM style exceptions.
14108 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14109 options. Add EXTRA_CC1_SPEC.
14110 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14112 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14114 (ARM_TARGET2_DWARF_FORMAT): Define.
14115 * config/arm/t-vxworks: Adjust multilib control to removal of the
14116 Diab command line options.
14118 2017-08-01 Martin Liska <mliska@suse.cz>
14120 PR gcov-profile/81561
14121 * gcov.c (unblock): Make unblocking safe as we need to preserve
14122 index correspondence of blocks and block_lists.
14124 2017-08-01 Richard Biener <rguenther@suse.de>
14126 PR tree-optimization/81181
14127 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14128 (compute_antic): ... end of iteration here.
14130 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14132 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14133 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14134 (ftree-slp-vectorize): Likewise.
14135 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14136 can no longer be set independent of flag_tree_loop_vectorize.
14137 * omp-general.c (emp_max_vf): Likewise.
14138 * opts.c (enable_fdo_optimizations): Remove references to
14139 flag_tree_vectorize, these are now implicit.
14140 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14141 and leave it for the options machinery.
14143 2017-08-01 Martin Liska <mliska@suse.cz>
14145 PR middle-end/70140
14146 * builtins.c (expand_builtin_memcpy_args): Remove.
14147 (expand_builtin_memcpy): Call newly added function
14148 expand_builtin_memory_copy_args.
14149 (expand_builtin_memcpy_with_bounds): Likewise.
14150 (expand_builtin_mempcpy): Remove last argument.
14151 (expand_builtin_mempcpy_with_bounds): Likewise.
14152 (expand_builtin_memory_copy_args): New function created from
14153 expand_builtin_mempcpy_args with small modifications.
14154 (expand_builtin_mempcpy_args): Remove.
14155 (expand_builtin_stpcpy): Remove unused argument.
14156 (expand_builtin): Likewise.
14157 (expand_builtin_with_bounds): Likewise.
14159 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14162 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
14163 print "ds:" only for immediates in generic address space.
14165 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14168 * config/i386/i386.c (ix86_funciton_naked): New prototype.
14169 (ix86_function_ok_for_sibcall): Return false for naked functions.
14171 2017-08-01 Richard Biener <rguenther@suse.de>
14173 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
14174 (compute_antic): Seed worklist with exit block predecessors.
14175 * cfganal.c (dfs_find_deadend): For a cycle return the source
14176 of the edge closing it.
14178 2017-08-01 Tamar Christina <tamar.christina@arm.com>
14180 * config/aarch64/aarch64.c
14181 (aarch64_can_const_movi_rtx_p): Move 0 check.
14183 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14185 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
14186 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
14188 * match.pd: Ditto in address comparison pattern.
14190 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14192 PR tree-optimization/81627
14193 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
14194 closed ssa form for store-store chain.
14196 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14198 PR tree-optimization/81620
14199 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
14200 for store-store chain.
14202 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14204 PR tree-optimization/81588
14205 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
14206 ranges[i].in_p, invert comparison code ccode. For >/>=,
14207 swap rhs1 and rhs2 and comparison code unconditionally,
14208 for </<= don't do that. Don't swap rhs1/rhs2 again if
14209 ranges[i].in_p, instead invert comparison code ccode if
14210 opcode or oe->rank is BIT_IOR_EXPR.
14213 * optabs.def (vec_extract_optab, vec_init_optab): Change from
14214 a direct optab to conversion optab.
14215 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
14216 with GET_MODE_INNER as last argument instead of optab_handler.
14217 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14218 vector extraction if possible and optab is available.
14219 * expr.c (store_constructor): Use convert_optab_handler instead
14220 of optab_handler. Use vector initialization from smaller
14221 vectors if possible and optab is available.
14222 * tree-vect-stmts.c (vectorizable_load): Likewise.
14223 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14224 now have two modes.
14225 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14226 of vec_init from half-sized vectors with the same element mode.
14227 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14228 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14229 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14230 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14231 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14232 after mode in gen_vec_extract* calls.
14233 (vec_extract<mode>): Renamed to ...
14234 (vec_extract<mode><ssescalarmodelower>): ... this.
14235 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14236 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14237 element mode after mode in gen_vec_init* calls.
14238 (VEC_INIT_HALF_MODE): New mode iterator.
14239 (vec_init<mode>): Renamed to ...
14240 (vec_init<mode><ssescalarmodelower>): ... this.
14241 (vec_init<mode><ssehalfvecmodelower>): New expander.
14242 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14243 (vec_extractv2sfsf): ... this.
14244 (vec_initv2sf): Renamed to ...
14245 (vec_initv2sfsf): ... this.
14246 (vec_extractv2si): Renamed to ...
14247 (vec_extractv2sisi): ... this.
14248 (vec_initv2si): Renamed to ...
14249 (vec_initv2sisi): ... this.
14250 (vec_extractv4hi): Renamed to ...
14251 (vec_extractv4hihi): ... this.
14252 (vec_initv4hi): Renamed to ...
14253 (vec_initv4hihi): ... this.
14254 (vec_extractv8qi): Renamed to ...
14255 (vec_extractv8qiqi): ... this.
14256 (vec_initv8qi): Renamed to ...
14257 (vec_initv8qiqi): ... this.
14258 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14259 (vec_init<mode>): Renamed to ...
14260 (vec_init<mode><VEC_base_l>): ... this.
14261 (vec_extract<mode>): Renamed to ...
14262 (vec_extract<mode><VEC_base_l>): ... this.
14263 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14264 (vec_initv2sfsf): ... this.
14265 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14266 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14267 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14268 element mode after mode in gen_vec_init* calls.
14269 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14270 (vec_init<mode><Vel>): ... this.
14271 (vec_extract<mode>): Renamed to ...
14272 (vec_extract<mode><Vel>): ... this.
14273 * config/aarch64/iterators.md (Vel): New mode attribute.
14274 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14275 Add element mode after mode in gen_vec_extract* calls.
14276 * config/s390/vector.md (non_vec_l): New mode attribute.
14277 (vec_extract<mode>): Renamed to ...
14278 (vec_extract<mode><non_vec_l>): ... this.
14279 (vec_init<mode>): Renamed to ...
14280 (vec_init<mode><non_vec_l>): ... this.
14281 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14282 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14284 * config/arm/iterators.md (V_elem_l): New mode attribute.
14285 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14286 (vec_extract<mode><V_elem_l>): ... this.
14287 (vec_extractv2di): Renamed to ...
14288 (vec_extractv2didi): ... this.
14289 (vec_init<mode>): Renamed to ...
14290 (vec_init<mode><V_elem_l>): ... this.
14291 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14292 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14293 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14294 Add element mode after gen_vec_extract* calls.
14295 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14296 (vec_init<mode><unitmode>): ... this.
14297 (vec_extract<mode>): Renamed to ...
14298 (vec_extract<mode><unitmode>): ... this.
14299 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14300 (vec_init<mode><unitmode>): ... this.
14301 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14302 (vec_initv2sfsf): ... this.
14303 (vec_extractv2sf): Renamed to ...
14304 (vec_extractv2sfsf): ... this.
14305 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14306 Add element mode after gen_vec_extract* calls.
14307 * config/mips/mips.md (unitmode): New mode iterator.
14308 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14309 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14310 * config/spu/spu.md (inner_l): New mode attribute.
14311 (vec_init<mode>): Renamed to ...
14312 (vec_init<mode><inner_l>): ... this.
14313 (vec_extract<mode>): Renamed to ...
14314 (vec_extract<mode><inner_l>): ... this.
14315 * config/sparc/sparc.md (veltmode): New mode iterator.
14316 (vec_init<VMALL:mode>): Renamed to ...
14317 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14318 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14319 (vec_initv2sisi): ... this.
14320 (vec_initv2sf): Renamed to ...
14321 (vec_initv2sfsf): ... this.
14322 (vec_extractv2sf): Renamed to ...
14323 (vec_extractv2sfsf): ... this.
14324 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14325 (vec_init<mode>): Renamed to ...
14326 (vec_init<mode><VEC_base_l>): ... this.
14327 (vec_extract<mode>): Renamed to ...
14328 (vec_extract<mode><VEC_base_l>): ... this.
14329 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14330 (vec_initv2sfsf): ... this.
14331 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14332 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14333 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14334 gen_vec_init* calls.
14336 2017-08-01 Richard Biener <rguenther@suse.de>
14338 PR tree-optimization/81297
14339 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14340 TREE_OVERFLOW from INTEGER_CSTs.
14342 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14344 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14346 2017-07-31 Carl Love <cel@us.ibm.com>
14348 * config/rs6000/rs6000-c: Add support for built-in functions
14349 vector signed char vec_xl_be (signed long long, signed char *);
14350 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14351 vector signed int vec_xl_be (signed long long, signed int *);
14352 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14353 vector signed long long vec_xl_be (signed long long, signed long long *);
14354 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14355 vector signed short vec_xl_be (signed long long, signed short *);
14356 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14357 vector double vec_xl_be (signed long long, double *);
14358 vector float vec_xl_be (signed long long, float *);
14359 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14360 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14361 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14363 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14364 (altivec_expand_builtin): Add switch statement to call
14365 altivec_expand_xl_be for each builtin.
14366 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14367 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14368 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14369 __builtin_vsx_le_be_v16qi.
14370 * doc/extend.texi: Update the built-in documentation file for the
14371 new built-in functions.
14373 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14376 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14378 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14380 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14382 * config.gcc: Add z14.
14383 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14384 CPU model numbers for z13s and z14.
14385 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14387 * config/s390/s390-opts.h (enum processor_type): Rename
14388 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14389 * config/s390/s390.c (processor_table): Add field for CPU name to
14390 be passed to Binutils.
14391 (s390_asm_output_machine_for_arch): Use the new field in
14392 processor_table for Binutils.
14393 (s390_expand_builtin): Replace arch12 with z14.
14394 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14395 (s390_get_sched_attrmask): Likewise.
14396 (s390_get_unit_mask): Likewise.
14397 * config/s390/s390.opt: Add z14 to processor_type enum.
14399 2017-07-31 Martin Jambor <mjambor@suse.cz>
14402 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14403 regardless of optimization level.
14405 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14406 Martin Liska <mliska@suse.cz>
14408 * predict.def: Remove old comment and adjust probability.
14409 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14410 PREDICT statements.
14412 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14415 * config/i386/i386.c (ix86_function_naked): New function.
14416 (ix86_can_use_return_insn_p): Return false for naked functions.
14417 (ix86_expand_prologue): Skip prologue for naked functions.
14418 (ix86_expand_epilogue): Skip epilogue for naked functions
14419 and emit trap instruction.
14420 (ix86_warn_func_return): New function.
14421 (ix86_attribute_table): Add "naked" attribute specification.
14422 (TARGET_WARN_FUNC_RETURN): Define.
14423 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14425 2017-07-31 Martin Liska <mliska@suse.cz>
14427 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14428 (dump_gimple_bb_header): Always dump BB info.
14429 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14431 2017-07-31 Martin Liska <mliska@suse.cz>
14434 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14435 also with current_function_decl non-null equality.
14437 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14440 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14441 change type to the element type, instead add eltype variable and
14442 use it where we are interested in the element type.
14444 PR tree-optimization/81603
14445 * ipa-polymorphic-call.c
14446 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14447 offset arithmetic in offset_int, bail out if the resulting bit offset
14448 doesn't fit into shwi.
14450 2017-07-31 Martin Liska <mliska@suse.cz>
14452 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14453 (gimplify_save_expr): Fix comment.
14455 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14458 * config/i386/i386.c (ix86_function_arg): Update arguments for
14460 (ix86_compute_frame_layout): Set the initial stack offset to
14461 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14462 INCOMING_FRAME_SP_OFFSET.
14463 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14464 stack before exception handler returns.
14465 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14466 the 'ERROR_CODE' for exception handler.
14468 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14470 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14471 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14472 (ASM_OUTPUT_REG_POP): Ditto.
14473 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14474 instead of asm_fprintf to output pure string.
14476 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14478 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14479 to imported_module_or_decl hook.
14480 (debug_nothing_tree_tree_tree_bool): Remove.
14481 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14482 * debug.c (do_nothing_debug_hooks): Use
14483 debug_nothing_tree_tree_tree_bool_bool instead of
14484 debug_nothing_tree_tree_tree_bool.
14485 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14486 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14487 * sdbout.c (sdb_debug_hooks): Likewise.
14488 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14489 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14491 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14492 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14493 attribute, don't add anything.
14495 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14497 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14498 (fold_build2_stat_loc): Likewise.
14499 (fold_build3_stat_loc): Likewise.
14500 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14501 (fold_build1_loc): Remove macro.
14502 (fold_build2_loc): Likewise.
14503 (fold_build3_loc): Likewise.
14505 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14507 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14508 (gimple_build_debug_bind_source_stat): Likewise.
14509 * gimple.h (gimple_build_debug_bind): Remove macro.
14510 (gimple_build_debug_bind_source): Likewise.
14512 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14514 * bitmap.c (bitmap_alloc): Adjust.
14515 (bitmap_gc_alloc): Likewise.
14516 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14518 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14520 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14521 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14522 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14523 (bitmap_gc_alloc_stat): Likewise.
14524 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14526 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14528 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14529 * rtl.h (shallow_copy_rtx): Remove macro.
14531 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14533 * emit-rtl.c (gen_raw_REG): Adjust.
14534 * gengenrtl.c (gendef): Likewise.
14535 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14536 * rtl.h (rtx_alloc): Remove macro.
14538 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14540 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14541 (build_tree_list_stat): Likewise.
14542 * tree.h (build_tree_list): Remove macro.
14543 (build_tree_list_vec): Likewise.
14545 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14547 * tree.c (make_vector_stat): Remove _stat from name.
14548 (build_vector_stat): Likewise.
14549 * tree.h (make_vector_stat): Remove macro.
14550 (build_vector_stat): Likewise.
14552 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14554 * tree.h (build_var_debug_value): Remove prototype.
14556 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14558 * tree.c (tree_cons_stat): Remove _stat from name.
14559 * tree.h (tree_cons): Remove macro.
14561 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14563 * tree.c (build_vl_exp_stat): Remove _stat from name.
14564 * tree.h (build_vl_exp): Remove macro.
14566 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14568 * tree.c (build_decl_stat): Remove _stat from name.
14569 * tree.h (build_decl): Remove macro.
14571 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14573 * gimple.c (gimple_build_with_ops_stat): Adjust.
14574 (gimple_alloc_stat): Remove _stat from name.
14575 * gimple.h (gimple_alloc): Remove macro.
14577 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14579 * tree.c (make_tree_vec_stat): Remove _stat from name.
14580 (grow_tree_vec_stat): Likewise.
14581 * tree.h (make_tree_vec_stat): Adjust prototype.
14582 (grow_tree_vec_stat): Likewise.
14583 (make_tree_vec): Remove macro.
14584 (grow_tree_vec): Likewise.
14586 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14588 * fold-const.c (fold_build1_stat_loc): Adjust.
14589 (fold_build2_stat_loc): Likewise.
14590 (fold_build3_stat_loc): Likewise.
14591 * tree.c (build0_stat): Remove _stat from name.
14592 (build1_stat): Likewise.
14593 (build2_stat): Likewise.
14594 (build3_stat): Likewise.
14595 (build4_stat): Likewise.
14596 (build5_stat): Likewise.
14597 * tree.h (build1_loc): Remove macro, and rename _stat function
14599 (build2_loc): Likewise.
14600 (build3_loc): Likewise.
14601 (build4_loc): Likewise.
14602 (build5_loc): Likewise.
14604 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14606 * tree.c (make_int_cst_stat): Remove _stat from name.
14607 * tree.h (make_int_cst_stat): Adjust prototype.
14608 (make_int_cst): Remove macro.
14610 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14612 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14613 * tree.h (make_tree_binfo_stat): Adjust prototype.
14614 (make_tree_binfo): Remove.
14616 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14618 * tree.c (copy_node_stat): Rename to copy_node.
14619 (build_distinct_type_copy): Adjust.
14620 * tree.h (copy_node_stat): Adjust prototype.
14621 (copy_node): Remove macro.
14623 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14625 * tree.c (make_node_stat): rename to make_node.
14626 (build_tree_list_stat): Adjust.
14627 (build0_stat): Likewise.
14628 (build2_stat): Likewise.
14629 (build3_stat): Likewise.
14630 (build4_stat): Likewise.
14631 (build5_stat): Likewise.
14632 (build_decl_stat): Likewise.
14633 * tree.h (make_node_stat): Adjust prototype.
14634 (make_node): remove macro.
14636 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14638 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14639 (PPC_FEATURE2_SCV): Likewise.
14640 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14642 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14644 * config/aarch64/aarch64.c
14645 (aarch64_internal_mov_immediate): Add new special pattern.
14646 * config/aarch64/aarch64.md (*movdi_aarch64):
14647 Add reg/32bit const mov case.
14649 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14650 Richard Sandiford <richard.sandiford@linaro.org>
14652 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14653 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14654 Add integer and movi cases.
14655 (movi-split-hf-df-sf split, fp16): New.
14656 (enabled): Added TARGET_FP_F16INST.
14657 * config/aarch64/iterators.md (GPF_HF): New.
14658 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
14660 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14662 * config/aarch64/aarch64.c
14663 (aarch64_simd_container_mode): Add prototype.
14664 (aarch64_expand_mov_immediate): Add HI support.
14665 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
14666 (aarch64_can_const_movi_rtx_p): New.
14667 (aarch64_preferred_reload_class):
14668 Remove restrictions of using FP registers for certain SIMD operations.
14669 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
14670 (aarch64_valid_floating_const): Add integer move validation.
14671 (aarch64_simd_imm_scalar_p): Remove.
14672 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14673 (aarch64_legitimate_constant_p): Expand list of supported cases.
14674 * config/aarch64/aarch64-protos.h
14675 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14676 (aarch64_reinterpret_float_as_int): New.
14677 (aarch64_simd_imm_scalar_p): Remove.
14678 * config/aarch64/constraints.md (Uvi): New.
14679 (Dd): Split into Ds and new Dd.
14680 * config/aarch64/aarch64.md (*movsi_aarch64):
14682 (*movdi_aarch64): Add SIMD mov case.
14684 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14686 * tree-predcom.c: (struct chain): Handle store-store chain in which
14687 stores for elimination only store loop invariant values.
14688 (execute_pred_commoning_chain): Ditto.
14689 (prepare_initializers_chain_store_elim): Ditto.
14690 (prepare_finalizers): Ditto.
14691 (is_inv_store_elimination_chain): New function.
14692 (initialize_root_vars_store_elim_1): New function.
14694 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14696 * tree-predcom.c: Revise general description of the pass.
14697 (enum chain_type): New enum type for store elimination.
14698 (struct chain): New field supporting store elimination.
14699 (struct component): Ditto.
14700 (dump_chain): Dump store-stores chain.
14701 (release_chain): Release resources.
14702 (split_data_refs_to_components): Compute and create component
14703 contains only stores for elimination.
14704 (get_chain_last_ref_at): New function.
14705 (make_invariant_chain): Initialization.
14706 (make_rooted_chain): Specify chain type in parameter and record it.
14707 (add_looparound_copies): Skip for store-stores chain.
14708 (determine_roots_comp): Compute type of chain and pass it to
14710 (initialize_root_vars_store_elim_2): New function.
14711 (finalize_eliminated_stores): New function.
14712 (remove_stmt): Handle store for elimination.
14713 (execute_pred_commoning_chain): Execute predictive commoning on
14714 store-store chains.
14715 (determine_unroll_factor): Skip unroll for store-stores chain.
14716 (prepare_initializers_chain_store_elim): New function.
14717 (prepare_initializers_chain): Hanlde store-store chain.
14718 (prepare_finalizers_chain, prepare_finalizers): New function.
14719 (tree_predictive_commoning_loop): Return integer value indicating
14720 if loop is unrolled or lcssa form is corrupted.
14721 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14723 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14725 * tree-predcom.c (initialize_root): Delete.
14726 (execute_pred_commoning_chain): Initialize root vars and replace
14727 reference of non-combined chain directly, rather than call above
14730 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14732 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14733 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14735 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14737 * tree-predcom.c (struct chain): New field init_seq.
14738 (release_chain): Release init_seq.
14739 (prepare_initializers_chain): Record intialization stmts in above
14741 (insert_init_seqs): New function.
14742 (tree_predictive_commoning_loop): Call insert_init_seqs.
14744 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14746 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14748 2017-07-28 Richard Biener <rguenther@suse.de>
14750 * match.pd: Remove superfluous :c.
14751 * genmatch.c (simplify::id): Add member.
14752 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14754 (current_id): New global.
14755 (dt_node::parent): Move from ...
14756 (dt_operand::parent): ... here. Add for_id member.
14757 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14758 (decision_tree::find_node): Relax order requirement when
14759 merging DT_TRUE nodes to ones inbetween the current simplify
14760 and the one we try to merge with. Add diagnostic whenever
14761 we need to enforce pattern order by not merging.
14762 (decision_tree::insert): Set current_id.
14763 (decision_tree::print_node): Dump parent node and for_id.
14764 (parser::last_id): Add member.
14765 (parser::push_simplify): Assign unique id.
14766 (parser::parser): Initialize last_id.
14768 2017-07-28 Martin Liska <mliska@suse.cz>
14771 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14772 gimple_build_debug_bind.
14774 2017-07-28 Richard Biener <rguenther@suse.de>
14776 PR tree-optimization/81502
14777 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14779 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14781 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14782 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14783 for BIT_FIELD_REF args.
14784 * fold-const.c (make_bit_field_ref): Likewise.
14785 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14787 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14790 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14791 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14792 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14793 Or it into SANITIZER_UNDEFINED.
14794 * ubsan.c: Include gimple-fold.h and varasm.h.
14795 (ubsan_expand_ptr_ifn): New function.
14796 (instrument_pointer_overflow): New function.
14797 (maybe_instrument_pointer_overflow): New function.
14798 (instrument_object_size): Formatting fix.
14799 (pass_ubsan::execute): Call instrument_pointer_overflow
14800 and maybe_instrument_pointer_overflow.
14801 * internal-fn.c (expand_UBSAN_PTR): New function.
14802 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14803 * sanitizer.def (__ubsan_handle_pointer_overflow,
14804 __ubsan_handle_pointer_overflow_abort): New builtins.
14805 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14806 * internal-fn.def (UBSAN_PTR): New internal function.
14807 * opts.c (sanitizer_opts): Add pointer-overflow.
14808 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14809 * fold-const.c (build_range_check): Compute pointer range check in
14810 integral type if pointer arithmetics would be needed. Formatting
14813 2017-07-28 Martin Liska <mliska@suse.cz>
14816 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14817 parameters that are of a variable-length.
14819 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14821 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14823 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14824 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14825 (CRT_CALL_STATIC_FUNCTION): Likewise.
14826 (ASM_DEFAULT_SPEC): New define.
14827 (ASM_SPEC32): Likewise.
14828 (ASM_SPEC64): Likewise.
14829 (ASM_SPEC_COMMON): Likewise.
14830 (ASM_SPEC): Likewise.
14831 (INVALID_64BIT): Likewise.
14832 (LINK_OS_DEFAULT_SPEC): Likewise.
14833 (LINK_OS_SPEC32): Likewise.
14834 (LINK_OS_SPEC64): Likewise.
14835 (POWERPC_LINUX): Likewise.
14836 (PTRDIFF_TYPE): Likewise.
14837 (RESTORE_FP_PREFIX): Likewise.
14838 (RESTORE_FP_SUFFIX): Likewise.
14839 (SAVE_FP_PREFIX): Likewise.
14840 (SAVE_FP_SUFFIX): Likewise.
14841 (SIZE_TYPE): Likewise.
14842 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14843 (TARGET_64BIT): Likewise.
14844 (TARGET_64BIT): Likewise.
14845 (TARGET_AIX): Likewise.
14846 (WCHAR_TYPE_SIZE): Likewise.
14847 (WCHAR_TYPE): Undefine.
14848 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14849 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14850 (CPP_OS_RTEMS_SPEC): Delete.
14851 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14852 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14854 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14856 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14858 PR tree-optimization/81578
14859 * tree-parloops.c (build_new_reduction): Bail out if
14860 reduction_code isn't one of the standard OpenMP reductions.
14861 Move the details printing after that decision.
14863 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14865 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14866 related to reload_in_progress.
14867 (splat_input_operand): Likewise.
14868 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14870 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14872 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14873 (TARGET_INSTANTIATE_DECLS): Likewise.
14874 (legitimate_indexed_address_p): Delete reload_in_progress code.
14875 (rs6000_debug_legitimate_address_p): Likewise.
14876 (rs6000_eliminate_indexed_memrefs): Likewise.
14877 (rs6000_emit_le_vsx_store): Likewise.
14878 (rs6000_emit_move_si_sf_subreg): Likewise.
14879 (rs6000_emit_move): Likewise.
14880 (register_to_reg_type): Likewise.
14881 (rs6000_pre_atomic_barrier): Likewise.
14882 (rs6000_machopic_legitimize_pic_address): Likewise.
14883 (rs6000_allocate_stack_temp): Likewise.
14884 (rs6000_address_for_fpconvert): Likewise.
14885 (rs6000_address_for_altivec): Likewise.
14886 (rs6000_secondary_memory_needed_rtx): Delete function.
14887 (rs6000_check_sdmode): Likewise.
14888 (rs6000_alloc_sdmode_stack_slot): Likewise.
14889 (rs6000_instantiate_decls): Likewise.
14890 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14891 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14892 Delete reload_in_progress.
14893 (*vec_reload_and_plus_<mptrsize>): Likewise.
14894 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14895 (vsx_div_v2di): Likewise.
14896 (vsx_udiv_v2di): Likewise.
14898 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14900 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14901 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14902 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14903 (rs6000_debug_reg_global): Delete print of LRA status.
14904 (rs6000_option_override_internal): Delete dead LRA related code.
14905 (rs6000_lra_p): Delete function.
14906 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14908 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14910 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14911 * config/riscv/rtems.h: New file.
14913 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14914 Sudakshina Das <sudi.das@arm.com>
14916 * config/aarch64/aarch64.md
14917 (define_split for and<mode>3nr_compare): Move
14918 non aarch64_logical_operand to a register.
14919 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14920 register immediate operand to a register.
14921 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14923 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14925 PR middle-end/81564
14926 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14928 2017-07-27 Richard Biener <rguenther@suse.de>
14930 PR tree-optimization/81573
14931 PR tree-optimization/81494
14932 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14933 multi defuse cycle case.
14935 2017-07-27 Richard Biener <rguenther@suse.de>
14937 PR tree-optimization/81571
14938 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14941 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14943 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14944 earlier and only if MASK_FPU is set. Adjust formatting.
14946 2017-07-27 Martin Liska <mliska@suse.cz>
14948 * opt-functions.awk: Add validation of value of Init.
14949 * optc-gen.awk: Pass new argument.
14951 2017-07-27 Martin Liska <mliska@suse.cz>
14953 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14954 Fix wrong condition.
14956 2017-07-27 Martin Liska <mliska@suse.cz>
14958 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14959 BBs and edges seen by autoFDO.
14961 2017-07-27 Richard Biener <rguenther@suse.de>
14963 PR tree-optimization/81502
14964 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14965 with incompatible but same sized type.
14966 (execute_update_addresses_taken): Likewise.
14968 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
14970 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
14971 flag_tree_loop_vectorize rather than flag_tree_vectorize.
14973 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14976 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
14977 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
14978 Change s_operand to memory_operand.
14980 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
14982 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
14983 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
14984 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
14985 Emit instructions rather than returning an expression. Handle TFmode
14986 and KFmode by casting to TImode.
14987 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
14988 (rs6000_emit_le_vsx_store): Likewise.
14989 * config/rs6000/vsx.md (VSX_TI): New iterator.
14990 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
14991 (*vsx_le_undo_permute_<mode>): Likewise.
14992 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
14993 emit the split sequence.
14994 (*vsx_le_perm_store_<mode>): Likewise.
14996 2017-07-27 Jakub Jelinek <jakub@redhat.com>
14998 PR tree-optimization/81555
14999 PR tree-optimization/81556
15000 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15001 if true, force CHANGED for the recursive invocation.
15002 (reassociate_bb): Remember original length of ops array, pass
15003 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15005 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15006 attributes for noipa attribute. For naked attribute use
15007 lookup_attribute first before lookup_attribute_spec.
15008 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15010 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15011 for functions with noipa attribute.
15012 (cgraph_externally_visible_p): Return true for functions with noipa
15014 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15015 for functions with noipa attribute.
15016 * doc/extend.texi: Document noipa function attribute.
15017 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15018 also for functions with noipa attribute.
15019 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15021 2017-07-26 Andrew Pinski <apinski@cavium.com>
15023 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15024 vec_unalign_load_cost and vec_unalign_store_cost.
15026 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15028 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15029 -mvsx-small-integer option.
15030 (ISA_3_0_MASKS_IEEE): Likewise.
15031 (OTHER_VSX_VECTOR_MASKS): Likewise.
15032 (POWERPC_MASKS): Likewise.
15033 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15034 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15035 code, only testing for DImode being allowed in non-VSX floating
15037 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15038 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15040 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15041 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15042 TARGET_P8_VECTOR test.
15043 (rs6000_secondary_reload_simple_move): Likewise.
15044 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15045 since TARGET_P9_VECTOR was already tested.
15046 (rs6000_opt_masks): Remove -mvsx-small-integer.
15047 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15048 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15050 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15051 test for TARGET_VEXTRACTUB was used, and that uses
15053 (p9 extract splitter): Likewise.
15054 (vsx_extract_<mode>_di_p9): Likewise.
15055 (vsx_extract_<mode>_store_p9): Likewise.
15056 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15057 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15058 the elimination of TARGET_VSX_SMALL_INTEGER.
15059 (vsx_extract_<mode>_p8): Likewise.
15060 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15061 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15062 (vsx_set_<mode>_p9): Likewise.
15063 (vsx_set_v4sf_p9): Likewise.
15064 (vsx_set_v4sf_p9_zero): Likewise.
15065 (vsx_insert_extract_v4sf_p9): Likewise.
15066 (vsx_insert_extract_v4sf_p9_2): Likewise.
15067 * config/rs6000/rs6000.md (sign extend splitter): Change
15068 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15069 (floatsi<mode>2_lfiwax_mem): Likewise.
15070 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15071 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15072 since a test for TARGET_P9_VECTOR was used.
15073 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15074 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15075 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15076 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15077 TARGET_P8_VECTOR test.
15078 (fix_trunc<mode>si2_stfiwx): Likewise.
15079 (fix_trunc<mode>si2_internal): Likewise.
15080 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15081 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15083 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15084 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15085 TARGET_P8_VECTOR test.
15086 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15087 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15088 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15090 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15091 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15092 since a test for TARGET_P9_VECTOR was used.
15093 (splitter for loading small constants): Likewise.
15095 2017-07-26 Andrew Pinski <apinski@cavium.com>
15097 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15100 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15103 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15104 (fp_valid_at): Likewise.
15106 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15108 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15109 (qdf24xx_addrcost_table): Likewise.
15110 (cortexa57_tunings): Update to use generic_branch_cost.
15111 (cortexa72_tunings): Likewise.
15112 (cortexa73_tunings): Likewise.
15113 (qdf24xx_tunings): Likewise.
15115 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15117 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15118 (thunderx2t99_branch_cost): Likewise.
15119 (cortexa35_tunings): Update to use generic_branch_cost.
15120 (cortexa53_tunings): Likewise.
15121 (cortexa57_tunings): Likewise.
15122 (cortexa72_tunings): Likewise.
15123 (cortexa73_tunings): Likewise.
15124 (thunderx2t99_tunings): Likewise.
15126 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15128 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15129 (sparc_option_override): Honour MASK_FSMULD.
15130 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15131 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15132 * config/sparc/sparc.opt (mfsmuld): New option.
15133 * doc/invoke.texi (mfsmuld): Document option.
15135 2017-07-26 Marek Polacek <polacek@redhat.com>
15137 PR middle-end/70992
15138 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15140 2017-07-26 Richard Biener <rguenther@suse.de>
15142 * gimple-match-head.c (do_valueize): Return OP if valueize
15144 (get_def): New helper to get at the def stmt of a SSA name
15145 if valueize allows.
15146 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15147 do_valueize to get at the def stmt.
15148 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15150 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15152 PR middle-end/46932
15153 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
15155 2017-07-26 Martin Liska <mliska@suse.cz>
15158 * function.c (expand_function_start): Make expansion of
15159 nonlocal_goto_save_area after parm_birth_insn.
15161 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15163 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
15164 from all CPU target flags enable members.
15166 2017-07-26 Richard Biener <rguenther@suse.de>
15168 * genmatch.c (dt_simplify::gen): Make iterator vars const.
15169 (decision_tree::gen): Make 'type' const.
15170 (write_predicate): Likewise.
15172 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15174 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15175 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
15176 (rs6000_option_override_internal): Likewise.
15177 (rs6000_expand_vector_set): Likewise.
15178 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
15179 (TARGET_UPPER_REGS_SF): Likewise.
15180 (TARGET_UPPER_REGS_DI): Likewise.
15181 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
15182 (TARGET_DIRECT_MOVE_64BIT): Likewise.
15183 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
15184 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15185 (Splitters for DI constants in Altivec registers): Likewise.
15186 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
15187 (vsx_set_v4sf_p9): Likewise.
15188 (vsx_set_v4sf_p9_zero): Likewise.
15189 (vsx_insert_extract_v4sf_p9): Likewise.
15190 (vsx_insert_extract_v4sf_p9_2): Likewise.
15192 2017-07-25 Carl Love <cel@us.ibm.com>
15194 * doc/extend.texi: Update the built-in documentation file for the
15195 existing built-in functions
15196 vector signed char vec_cnttz (vector signed char);
15197 vector unsigned char vec_cnttz (vector unsigned char);
15198 vector signed short vec_cnttz (vector signed short);
15199 vector unsigned short vec_cnttz (vector unsigned short);
15200 vector signed int vec_cnttz (vector signed int);
15201 vector unsigned int vec_cnttz (vector unsigned int);
15202 vector signed long long vec_cnttz (vector signed long long);
15203 vector unsigned long long vec_cnttz (vector unsigned long long);
15205 2017-07-25 Andrew Pinski <apinski@cavium.com>
15207 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
15208 accesses where the use is for the first operand of a BIT_INSERT.
15210 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15213 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15214 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15216 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15218 * config/i386/gstabs.h: Delete.
15219 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15221 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15223 * config/i386/i386.c (ix86_decompose_address): Do not check for
15224 register RTX when looking at index_reg or base_reg.
15225 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15227 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15229 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15230 to update EH info here.
15232 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15234 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15236 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15238 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15240 2017-07-25 Torsten Duwe <duwe@suse.de>
15242 * common.opt: Introduce -fpatchable-function-entry
15243 command line option, and its variables function_entry_patch_area_size
15244 and function_entry_patch_area_start.
15245 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15246 including a two-value parser.
15247 * target.def (print_patchable_function_entry): New target hook.
15248 * targhooks.h (default_print_patchable_function_entry): New function.
15249 * targhooks.c (default_print_patchable_function_entry): Likewise.
15250 * toplev.c (process_options): Switch off IPA-RA if
15251 patchable function entries are being generated.
15252 * varasm.c (assemble_start_function): Look at the
15253 patchable-function-entry command line switch and current
15254 function attributes and maybe generate NOP instructions by
15255 calling the print_patchable_function_entry hook.
15256 * doc/extend.texi: Document patchable_function_entry attribute.
15257 * doc/invoke.texi: Document -fpatchable_function_entry
15258 command line option.
15259 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15261 * doc/tm.texi: Re-generate.
15263 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15266 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15267 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15269 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15271 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15273 2017-07-25 Richard Biener <rguenther@suse.de>
15275 PR tree-optimization/81455
15276 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15277 not walk in cycles when looking for guards.
15279 2017-07-25 Richard Biener <rguenther@suse.de>
15281 PR tree-optimization/81529
15282 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15283 when optimizing backedge uses.
15285 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15287 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15289 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15290 to dl_section_ref. On AIX, append an expression to subtract
15291 the size of the section length to dl_section_ref.
15293 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15295 * configure.ac: If any of the config.* scripts fail, exit 1.
15296 * configure: Regenerate.
15298 2017-07-25 Richard Biener <rguenther@suse.de>
15300 PR middle-end/81546
15301 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15302 of immediate uses, be more verbose on errors.
15304 2017-07-25 Richard Biener <rguenther@suse.de>
15306 PR tree-optimization/81510
15307 * tree-vect-loop.c (vect_is_simple_reduction): When the
15308 reduction stmt is not inside the loop bail out.
15310 2017-07-25 Richard Biener <rguenther@suse.de>
15312 PR tree-optimization/81303
15313 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15314 profitability check against LOOP_VINFO_NITERSM1.
15316 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15318 * domwalk.c (cmp_bb_postorder): Simplify.
15319 (sort_bbs_postorder): New function. Use it...
15320 (dom_walker::walk): ...here to optimize common cases.
15322 2017-07-25 Martin Liska <mliska@suse.cz>
15325 * ipa-visibility.c (function_and_variable_visibility): Make the
15326 redirection just on target that supports aliasing.
15327 Fix GNU coding style.
15329 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15332 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15334 * config/arm/rtems.h: Likewise.
15335 * config/bfin/rtems.h: Likewise.
15336 * config/i386/rtemself.h: Likewise.
15337 * config/lm32/rtems.h: Likewise.
15338 * config/m32c/rtems.h: Likewise.
15339 * config/m68k/rtemself.h: Likewise.
15340 * config/microblaze/rtems.h: Likewise.
15341 * config/mips/rtems.h: Likewise.
15342 * config/moxie/rtems.h: Likewise.
15343 * config/nios2/rtems.h: Likewise.
15344 * config/powerpcspe/rtems.h: Likewise.
15345 * config/rs6000/rtems.h: Likewise.
15346 * config/rtems.h: Likewise.
15347 * config/sh/rtems.h: Likewise.
15348 * config/sh/rtemself.h: Likewise.
15349 * config/sparc/rtemself.h: Likewise.
15351 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15354 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15355 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15356 * tree-ssa-structalias.c (alias_get_name): Same.
15358 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15361 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15362 instructions if no du chain is found.
15364 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15366 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15368 2017-07-25 Richard Biener <rguenther@suse.de>
15370 PR middle-end/81505
15371 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15374 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15376 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15377 upper-regs options.
15378 (ISA_2_7_MASKS_SERVER): Likewise.
15379 (ISA_3_0_MASKS_IEEE): Likewise.
15380 (OTHER_P8_VECTOR_MASKS): Likewise.
15381 (OTHER_VSX_VECTOR_MASKS): Likewise.
15382 (POWERPC_MASKS): Likewise.
15383 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15384 duplicate list of options.
15385 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15386 explicit -mupper-regs options.
15387 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15388 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15389 alias for -mupper-regs-df.
15390 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15391 (rs6000_init_hard_regno_mode_ok): Likewise.
15392 (rs6000_option_override_internal): Likewise.
15393 (rs6000_opt_masks): Likewise.
15394 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15395 options in terms of whether -mvsx or -mpower8-vector was used.
15396 (TARGET_UPPER_REGS_DI): Likewise.
15397 (TARGET_UPPER_REGS_SF): Likewise.
15398 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15399 -mupper-regs-* options.
15401 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15403 * passes.c (emergency_dump_function): Print some empty lines and a
15404 header before the RTL dump.
15406 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15408 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15410 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15413 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15414 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15416 2017-07-24 Carl Love <cel@us.ibm.com>
15418 * config/rs6000/rs6000-c.c: Add support for built-in functions
15419 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15420 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15421 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15422 vec_extract_fp_from_shortl): Add defines for the two builtins.
15423 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15424 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15426 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15427 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15428 * doc/extend.texi: Update the built-in documentation file for the
15429 new built-in function.
15431 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15434 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15436 * doc/generic.texi: Likewise.
15437 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15438 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15440 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15442 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15443 (aarch64_mls_elt_merge<mode>): Likewise.
15445 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15447 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15448 having __cxa_atexit.
15450 2017-07-23 Michael Collison <michael.collison@arm.com>
15452 * config/arm/arm.c (arm_option_override): Deprecate
15453 use of -mstructure-size-boundary.
15454 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15455 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15457 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15460 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15461 Reduce cost estimate for direct moves.
15463 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15466 * config/i386/i386.c (ix86_option_override_internal): Disable
15467 BMI, BMI2 and TBM instructions for -m16.
15469 2017-07-21 Carl Love <cel@us.ibm.com>
15471 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15472 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15473 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15474 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15475 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15476 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15477 VMULOSW): New enum "unspec" values.
15478 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15479 altivec_vmulosw): New patterns.
15480 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15481 VMULOSW): Add definitions.
15483 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15485 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15486 (qdf24xx): Likewise.
15487 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15488 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15489 (AARCH64_FL_V8_1): Renumber.
15490 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15491 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15492 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15493 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15494 rdma to feature modifiers list.
15496 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15498 PR middle-end/56727
15499 * ipa-visibility (function_and_variable_visibility): Convert
15500 recursive PLT call to direct call if appropriate.
15502 2017-07-21 Andrew Pinski <apinski@cavium.com>
15504 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15505 operand 1 to see if the types precision matches.
15506 * fold-const.c (operand_equal_p): Likewise.
15508 2017-07-21 Richard Biener <rguenther@suse.de>
15510 PR tree-optimization/81303
15511 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15512 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15513 (vect_peeling_hash_get_lowest_cost): Adjust.
15514 (vect_enhance_data_refs_alignment): Likewise. Use
15515 vect_get_peeling_costs_all_drs to compute the penalty for no
15516 peeling to match up costs.
15518 2017-07-21 Richard Biener <rguenther@suse.de>
15520 PR tree-optimization/81500
15521 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15522 we didn't identify a reduction path.
15524 2017-07-21 Tom de Vries <tom@codesourcery.com>
15525 Cesar Philippidis <cesar@codesourcery.com>
15527 PR gcov-profile/81442
15528 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15531 2017-07-21 Tom de Vries <tom@codesourcery.com>
15534 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15536 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15537 nvptx_override_options_after_change.
15539 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15541 * dwarf2out.c (output_file_names): Avoid double testing for
15542 dwarf_version >= 5.
15544 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15546 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15548 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15550 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15552 (try_crossjump_to_edge): Do not punt on partitioned functions.
15554 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15556 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15557 Put all BBs reachable only via paths crossing cold region to cold
15559 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15561 2016-07-21 Richard Biener <rguenther@suse.de>
15563 PR tree-optimization/81303
15564 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15565 into account prologue and epilogue iterations when raising
15566 min_profitable_iters to sth at least covering one vector iteration.
15568 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15570 * config/arm/arm.c (arm_test_cpu_arch_dat):
15573 2017-07-20 Nathan Sidwell <nathan@acm.org>
15575 Remove TYPE_METHODS.
15576 * tree.h (TYPE_METHODS): Delete.
15577 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15578 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15579 (dbxout_type_methods): Scan TYPE_FIELDS.
15580 (dbxout_type): Don't check TYPE_METHODS here.
15581 * function.c (use_register_for_decl): Always ignore register for
15582 class types when not optimizing.
15583 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15584 * tree.c (free_lang_data_in_type): Stitch out member functions and
15585 templates from TYPE_FIELDS.
15586 (build_distinct_type_copy, verify_type_variant,
15587 verify_type): Member fns are on TYPE_FIELDS.
15588 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15589 * tree-pretty-print.c (dump_generic_node): Likewise.
15591 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15594 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15595 V2TImode and V4TImode.
15596 (ix86_expand_vector_extract): Likewise.
15597 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15598 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15599 (ssescalarmode): Handle V4TImode and V2TImode.
15600 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15601 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15602 (VEXTRACTI128_MODE): New mode iterator.
15603 (splitter for *vec_extractv?ti first element): New.
15604 (VEC_INIT_MODE): New mode iterator.
15605 (vec_init<mode>): Consolidate 3 expanders into one using
15606 VEC_INIT_MODE mode iterator.
15608 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15610 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15611 non_spilled_static_chain_regno_p.
15613 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15615 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15617 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15619 * bb-reorder.c (connect_traces): Allow copying of blocks within
15622 2017-07-20 Richard Biener <rguenther@suse.de>
15624 * gimple.h (gimple_phi_result): Add gphi * overload.
15625 (gimple_phi_result_ptr): Likewise.
15626 (gimple_phi_arg): Likewise. Adjust index assert to only
15627 allow actual argument accesses rather than all slots available
15629 (gimple_phi_arg_def): Add gphi * overload.
15630 * tree-phinodes.c (make_phi_node): Initialize only actual
15632 (resize_phi_node): Clear memory not covered by old node,
15633 do not initialize excess argument slots.
15634 (reserve_phi_args_for_new_edge): Initialize new argument slot
15637 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15639 PR tree-optimization/81388
15641 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15643 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15644 by removing computation of may_be_zero.
15646 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15647 Tom de Vries <tom@codesourcery.com>
15649 PR middle-end/81030
15650 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15651 when gimple level profile disagrees with what RTL expander did.
15653 2017-07-20 Richard Biener <rguenther@suse.de>
15655 PR tree-optimization/61171
15656 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
15657 (vect_analyze_stmt): Add slp instance parameter.
15658 (vectorizable_reduction): Likewise.
15659 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
15660 (vect_is_simple_reduction): Deal with chains not detected
15661 as SLP reduction chain, specifically not properly associated
15662 chains containing a mix of plus/minus.
15663 (get_reduction_op): Remove.
15664 (get_initial_defs_for_reduction): Simplify, pass in whether
15665 this is a reduction chain, pass in the SLP node for the PHIs.
15666 (vect_create_epilog_for_reduction): Get the SLP instance as
15668 (vectorizable_reduction): Get the SLP instance as arg.
15669 During analysis remember the SLP node with the PHIs in the
15670 instance. Simplify getting at the vectorized reduction PHIs.
15671 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15672 through SLP instance.
15673 (vect_slp_analyze_operations): Likewise.
15674 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15675 (vect_transform_stmt): Likewise.
15677 2017-07-20 Tom de Vries <tom@codesourcery.com>
15679 PR tree-optimization/81489
15680 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15681 read of phi arg location to before loop that modifies phi.
15683 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15685 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15688 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15690 PR middle-end/81331
15691 * except.c (execute): Fix ordering issue.
15693 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15695 PR rtl-optimization/81423
15696 * combine.c (make_compound_operation_int): Don't try to optimize
15697 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15699 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15701 PR rtl-optimization/81423
15702 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15703 with a constant that is -1 in the truncated to mode.
15705 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15707 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15708 (determine_unlikely_bbs): ... here.
15709 * predict.h (propagate_unlikely_bbs_forward): Declare.
15710 * cfgexpand.c (pass_expand::execute): Use it.
15711 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15713 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15714 propagate_unlikely_bbs_forward.
15716 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15718 PR middle-end/81331
15719 * except.c (maybe_add_nop_after_section_switch): New function.
15722 2017-07-19 Tom de Vries <tom@codesourcery.com>
15724 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15726 2017-07-19 Tom de Vries <tom@codesourcery.com>
15728 * gimple.h (gimple_phi_arg): Make assert more strict.
15730 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15732 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15733 * config/rs6000/mmintrin.h: New file.
15734 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15736 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15738 PR tree-optimization/81346
15739 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15741 2017-07-19 Tom de Vries <tom@codesourcery.com>
15743 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15745 2017-07-19 Tom de Vries <tom@codesourcery.com>
15747 * config/nvptx/nvptx-modes.def: Add V2DImode.
15748 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15749 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15750 (nvptx_output_mov_insn): Handle lack of mov.b128.
15751 (nvptx_print_operand): Handle 'H' and 'L' codes.
15752 (nvptx_vector_mode_supported): Allow V2DImode.
15753 (nvptx_preferred_simd_mode): New function.
15754 (nvptx_data_alignment): New function.
15755 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15756 nvptx_preferred_simd_mode.
15757 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15759 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15761 2017-07-19 Tom de Vries <tom@codesourcery.com>
15763 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15764 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15765 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15766 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15767 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15768 (mov<VECIM>_insn): New define_insn.
15769 (define_expand "mov<VECIM>): New define_expand.
15771 2017-07-19 Tom de Vries <tom@codesourcery.com>
15773 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15775 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15777 PR tree-optimization/81346
15778 * fold-const.h (fold_div_compare, range_check_type): Declare.
15779 * fold-const.c (range_check_type): New function.
15780 (build_range_check): Use range_check_type.
15781 (fold_div_compare): No longer static, rewritten into
15782 a match.pd helper function.
15783 (fold_comparison): Don't call fold_div_compare here.
15784 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15785 as helper function.
15787 2017-07-19 Nathan Sidwell <nathan@acm.org>
15789 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15790 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15791 * tree.c (find_decls_types_r, verify_type): Use
15792 TYPE_{MIN,MAX}_VALUE_RAW.
15793 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15794 (hash_tree): Likewise.
15795 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15797 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15800 2017-07-18 Tom de Vries <tom@codesourcery.com>
15802 PR middle-end/81464
15803 * omp-expand.c (expand_omp_for_static_chunk): Handle
15804 equal-argument loop exit phi.
15806 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15809 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15810 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15811 operand 2 predicate.
15812 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15813 operand 2 predicate.
15814 (ror,rol -> rorx splitters): Use const_int_operand as
15815 operand 2 predicate.
15817 2017-06-18 Richard Biener <rguenther@suse.de>
15819 PR tree-optimization/81410
15820 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15821 the gap in the ! slp_perm SLP case after each group.
15823 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15825 PR middle-end/81463
15826 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15829 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15831 PR middle-end/81462
15832 * predict.c (set_even_probabilities): Cleanup; do not affect
15833 probabilities that are already known.
15834 (combine_predictions_for_bb): Call even when count is set.
15836 2017-07-18 Nathan Sidwell <nathan@acm.org>
15838 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15841 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15844 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15845 optimization for loop niter analysis.
15847 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15850 * config/avr/avr.c (avr_optimize_casesi): Don't use
15851 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15853 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15855 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15856 body_cost_vec from _vect_peel_extended_info.
15857 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15858 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15861 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15863 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15865 2017-07-18 Richard Biener <rguenther@suse.de>
15867 PR tree-optimization/80620
15868 PR tree-optimization/81403
15869 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15870 info when re-using a VN table entry.
15872 2017-07-18 Richard Biener <rguenther@suse.de>
15874 PR tree-optimization/81418
15875 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15876 vectype_in. Verify that with lane-reducing reduction operations
15877 we have a single def-use cycle.
15879 2017-07-17 Carl Love <cel@us.ibm.com>
15881 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15883 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15884 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15885 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15886 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15887 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15888 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15889 VMULOSW): New enum "unspec" values.
15890 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15891 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15892 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15893 altivec_vmulosw): New patterns.
15894 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15895 VMULOSW): Add definitions.
15896 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15898 * config/alpha/alpha.c: Include predict.h.
15900 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15902 * tree-vrp.c (compare_assert_loc): Fix comparison function
15903 to return predictable results.
15905 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15907 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15909 (subdi3): Likewise.
15910 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15911 * doc/invoke.texi (mexpand-adddi): Update text.
15913 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15915 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15916 that also clobbers the CC register. The old expand code is moved
15918 (*arc_clzsi2): ... here.
15919 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15920 the CC register. The old expand code is moved to ...
15921 (arc_ctzsi2): ... here.
15923 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15925 * config/arc/arc.opt (mindexed-loads): Use initial value
15926 TARGET_INDEXED_LOADS_DEFAULT.
15927 (mauto-modify-reg): Use initial value
15928 TARGET_AUTO_MODIFY_REG_DEFAULT.
15929 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15930 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15931 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15932 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15934 2017-07-17 Martin Liska <mliska@suse.cz>
15937 * opts.c (finish_options): Do not allow -fgnu-tm
15938 w/ -fsanitize={kernel-,}address. Say sorry.
15940 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15943 * tree-loop-distribution.c (classify_partition): Only assert on
15944 numer of iterations.
15945 (merge_dep_scc_partitions): Delete prameter. Update function call.
15946 (distribute_loop): Remove code handling loop with unknown niters.
15947 (pass_loop_distribution::execute): Skip loop with unknown niters.
15949 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15952 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15953 function sort_partitions_by_post_order.
15955 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15957 PR tree-optimization/81374
15958 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15959 the max index of basic blocks, rather than number of basic blocks.
15961 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15963 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15965 (arc_legitimate_pic_operand_p): Likewise.
15966 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
15968 (arc_needs_pcl_p): Likewise.
15969 (arc_legitimate_pc_offset_p): Likewise.
15970 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
15971 function is also used in constrains.md.
15972 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
15973 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
15974 PLUS. Only return true/false in known cases, otherwise assert.
15975 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
15976 is already called in arc_legitimate_constant_p.
15977 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
15979 (LEGITIMATE_PIC_OPERAND_P): Use
15980 arc_raw_symbolic_reference_mentioned_p function.
15981 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
15986 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15987 Andrew Burgess <andrew.burgess@embecosm.com>
15989 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
15990 (arc_return_address_register): New function.
15991 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
15992 (arc_handle_fndecl_attribute): Add naked attribute.
15993 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15994 (TARGET_WARN_FUNC_RETURN): Likewise.
15995 (arc_allocate_stack_slots_for_args): New function.
15996 (arc_warn_func_return): Likewise.
15997 (machine_function): Change type fn_type.
15998 (arc_compute_function_type): Consider new naked function type,
15999 change function return type.
16000 (arc_must_save_register): Adapt to handle new
16001 arc_compute_function_type's return type.
16002 (arc_expand_prologue): Likewise.
16003 (arc_expand_epilogue): Likewise.
16004 (arc_return_address_regs): Delete.
16005 (arc_return_address_register): New function.
16006 (arc_epilogue_uses): Use above function.
16007 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16008 (arc_function_type): Change encoding, add naked type.
16009 (ARC_INTERRUPT_P): Change to handle the new encoding.
16010 (ARC_FAST_INTERRUPT_P): Likewise.
16011 (ARC_NORMAL_P): Define.
16012 (ARC_NAKED_P): Likewise.
16013 (arc_compute_function_type): Delete prototype.
16014 * config/arc/arc.md (in_ret_delay_slot): Use
16015 arc_return_address_register function.
16016 (simple_return): Likewise.
16017 (p_return_i): Likewise.
16019 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16021 PR tree-optimization/81428
16022 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16023 can't be built for those types.
16025 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16027 Remove stuff dead since r239246.
16029 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16030 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16031 (avr_inform_devices): Remove dead stuff.
16033 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16035 * config/arm/arm_neon.h: Fix softp typo.
16037 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16039 PR tree-optimization/81365
16040 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16041 aggregate moves onto bb predecessor edges, make sure there are no
16042 loads that could alias the lhs in between the start of bb and the
16045 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16048 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16049 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16050 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16052 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16054 PR tree-optimization/81396
16055 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16056 (init_symbolic_number): Initialize it to 1.
16057 (perform_symbolic_merge): Add n_ops from both operands into the new
16059 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16060 without base_addr as useless if they need more than one operation.
16061 (bswap_replace): Handle !bswap case for NULL base_addr.
16063 2017-07-17 Tom de Vries <tom@codesourcery.com>
16066 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16069 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16071 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16072 conditional builtin define __FIX_LEON3FT_B2BST.
16074 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16076 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16077 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16080 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16082 PR rtl-optimization/81424
16083 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16084 to remove potential trapping from operands if -fnon-call-exceptions.
16086 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16088 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16089 profile_proability for scalling.
16090 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16092 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16094 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16096 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16098 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16099 fixpoint arithmetics.
16101 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16103 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16104 fixpoint arithmetics.
16106 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16108 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16109 fixpoint arithmetics.
16111 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16113 * profile-count.h (profile_probability::from_reg_br_prob_note,
16114 profile_probability::to_reg_br_prob_note): New functions.
16115 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16116 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16117 * predict.c (probability_reliable_p): Update.
16118 (edge_probability_reliable_p): Update.
16119 (br_prob_note_reliable_p): Update.
16120 (invert_br_probabilities): Update.
16121 (add_reg_br_prob_note): New function.
16122 (combine_predictions_for_insn): Update.
16123 * asan.c (asan_clear_shadow): Update.
16124 * cfgbuild.c (compute_outgoing_frequencies): Update.
16125 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16126 (update_br_prob_note): Update.
16127 (rtl_verify_edges): Update.
16128 (purge_dead_edges): Update.
16129 (fixup_reorder_chain): Update.
16130 * emit-rtl.c (try_split): Update.
16131 * ifcvt.c (cond_exec_process_insns): Update.
16132 (cond_exec_process_if_block): Update.
16133 (dead_or_predicable): Update.
16134 * internal-fn.c (expand_addsub_overflow): Update.
16135 (expand_neg_overflow): Update.
16136 (expand_mul_overflow): Update.
16137 * loop-doloop.c (doloop_modify): Update.
16138 * loop-unroll.c (compare_and_jump_seq): Update.
16139 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16140 * predict.h: Update.
16141 * reorg.c (mostly_true_jump): Update.
16143 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16144 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16145 * config/arc/arc.c: (emit_unlikely_jump): Update.
16146 * config/arm/arm.c: (emit_unlikely_jump): Update.
16147 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16148 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16149 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16150 (ix86_print_operand): Update.
16151 (ix86_split_fp_branch): Update.
16152 (predict_jump): Update.
16153 * config/ia64/ia64.c (ia64_print_operand): Update.
16154 * config/mmix/mmix.c (mmix_print_operand): Update.
16155 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
16156 (rs6000_expand_split_stack_prologue): Update.
16157 * config/rs6000/rs6000.c: Update.
16158 * config/s390/s390.c (s390_expand_vec_strlen): Update.
16159 (s390_expand_vec_movstr): Update.
16160 (s390_expand_cs_tdsi): Update.
16161 (s390_expand_split_stack_prologue): Update.
16162 * config/sh/sh.c (sh_print_operand): Update.
16163 (expand_cbranchsi4): Update.
16164 (expand_cbranchdi4): Update.
16165 * config/sparc/sparc.c (output_v9branch): Update.
16166 * config/spu/spu.c (get_branch_target): Update.
16167 (ea_load_store_inline): Update.
16168 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
16169 * config/tilepro/tilepro.c: Update.
16171 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16173 * gimplify.c (mostly_copy_tree_r): Revert latest change.
16174 (gimplify_save_expr): Likewise.
16176 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16178 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
16180 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16182 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
16184 * timevar.def (TV_IPA_FNSUMMARY): Define.
16186 2017-07-16 Daniel Cederman <cederman@gaisler.com>
16188 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
16189 to back store errata sensitive sequence from being generated.
16190 (sqrtdf2_fix): Likewise.
16192 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16194 * tree-ssa-threadupdate.c (compute_path_counts,
16195 update_joiner_offpath_counts): Use profile_probability.
16197 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
16200 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16202 * config/arm/arm-c.c (arm_cpu_builtins): Define
16203 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16205 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16207 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16208 array entries to represent __ieee128 versions of the
16209 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
16210 scalar_extract_sig, and scalar_insert_exp built-in functions.
16211 (altivec_resolve_overloaded_builtin): Add special case handling
16212 for the __builtin_scalar_insert_exp function, as represented by
16213 the P9V_BUILTIN_VEC_VSIEDP constant.
16214 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
16215 exponent support for __ieee128 argument.
16216 (VSESQP): Add scalar extract signature support for __ieee128
16218 (VSTDCNQP): Add scalar test negative support for __ieee128
16220 (VSIEQP): Add scalar insert exponent support for __int128 argument
16221 with __ieee128 result.
16222 (VSIEQPF): Add scalar insert exponent support for __ieee128
16223 argument with __ieee128 result.
16224 (VSTDCQP): Add scalar test data class support for __ieee128
16226 (VSTDCNQP): Add overload support for scalar test negative with
16227 __ieee128 argument.
16228 (VSTDCQP): Add overload support for scalar test data class
16229 __ieee128 argument.
16230 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16231 UNSPEC_VSX_SXSIGDP.
16232 (UNSPEC_VSX_SIEXPQP): New constant.
16233 (xsxexpqp): New insn for VSX scalar extract exponent quad
16235 (xsxsigqp): New insn for VSX scalar extract significand quad
16237 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16238 precision with floating point argument.
16239 (xststdcqp): New expand for VSX scalar test data class quad
16241 (xststdcnegqp): New expand for VSX scalar test negative quad
16243 (xststdcqp): New insn to match expansions for VSX scalar test data
16244 class quad precision and VSX scalar test negative quad precision.
16245 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16246 special case operand checking to enforce that second operand of
16247 VSX scalar test data class with quad precision argument is a 7-bit
16249 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16250 prototypes and descriptions of __ieee128 versions of
16251 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16252 scalar_test_data_class, and scalar_test_neg built-in functions.
16254 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16256 PR tree-optimization/81162
16257 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16258 replace a negate with an add.
16260 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16262 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16264 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16266 * config/arm/arm-c.c (arm_cpu_builtins): Define
16267 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16269 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16271 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16272 (armv8-r): Set ARM Cortex-R52 as default CPU.
16273 * config/arm/arm-tables.opt: Regenerate.
16274 * config/arm/arm-tune.md: Regenerate.
16275 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16277 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16278 extension for -mcpu=cortex-r52.
16280 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16282 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16283 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16284 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16285 (fp-armv8): Define it as FP_ARMv8 only.
16286 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16287 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16289 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16290 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16291 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16292 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16293 than TARGET_FPU_ARMV8.
16294 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16295 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16296 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16298 * config/arm/neon.md (neon_vrint): Likewise.
16299 (neon_vcvt): Likewise.
16300 (neon_<fmaxmin_op><mode>): Likewise.
16301 (<fmaxmin><mode>3): Likewise.
16302 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16303 * config/arm/predicates.md (arm_cond_move_operator): Check against
16304 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16306 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16308 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16309 to top of function.
16311 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16313 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16314 loop in comment with memset.
16316 2017-07-14 Martin Liska <mliska@suse.cz>
16318 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16319 * dwarf2out.c (is_java): Remove the function.
16320 (output_pubname): Remove usage of the function.
16321 (lower_bound_default): Remove usage of DW_LANG_Java.
16322 (gen_compile_unit_die): Likewise.
16323 * gcc.c: Remove compiler defaults for .java and .zip files.
16324 * gimple-expr.c (remove_suffix): Change as there's no longer
16325 extension than 4-letter one.
16326 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16327 (gimplify_save_expr): Likewise.
16328 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16329 as it's possible even for other languages than Java.
16330 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16331 * lto-opts.c (lto_write_options): Remove reference to Java.
16332 * opts.c (strip_off_ending): Update file extension handling.
16333 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16334 * tree-eh.c (lower_resx): Likewise.
16335 * tree.c (free_lang_data_in_type): Remove dead code.
16336 (find_decls_types_r): Likewise.
16337 (build_common_builtin_nodes): Remove Java from a comment.
16338 (verify_type): Remove dead code.
16339 * varasm.c (assemble_external): Remove Java from a comment.
16341 2017-07-14 Martin Liska <mliska@suse.cz>
16343 * opts.c (finish_options): Add quotes.
16344 (common_handle_option): Likewise.
16346 2017-07-14 Martin Liska <mliska@suse.cz>
16348 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16349 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16350 Remove N_SO_PASCAL.
16351 * dwarf2out.c (lower_bound_default): Do not handle
16353 (gen_compile_unit_die): Likewise.
16354 * gcc.c: Remove default extension binding for GNU Pascal.
16355 * stmt.c: Remove Pascal language from a comment.
16356 * xcoffout.c: Likewise.
16358 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16361 * diagnostic-show-locus.c (fixit_cmp): New function.
16362 (layout::layout): Sort m_fixit_hints.
16363 (column_range::column_range): Assert that the values are valid.
16364 (struct char_span): New struct.
16365 (correction::overwrite): New method.
16366 (struct source_line): New struct.
16367 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16368 calls in terms of classes source_line and char_span, and
16369 correction::overwrite.
16370 (selftest::test_overlapped_fixit_printing_2): New function.
16371 (selftest::diagnostic_show_locus_c_tests): Call it.
16373 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16375 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16376 early if there is no lhs.
16378 2017-07-13 Martin Liska <mliska@suse.cz>
16380 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16381 (gen_reference_type_die): Likewise.
16382 * stor-layout.c: Remove Pascal-related comment.
16384 2017-07-13 Martin Liska <mliska@suse.cz>
16386 * opts.c (finish_options): Add quotes to error messages.
16387 (parse_sanitizer_options): Likewise.
16389 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16391 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16393 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16395 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16397 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16399 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16401 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16403 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16406 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16407 provides the hardware capability bits, define the macro
16408 __BUILTIN_CPU_SUPPORTS__.
16409 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16410 if GLIBC does not provide the hardware capability bits. Add a
16411 gcc_unreachable call if the built-in cpu function is neither
16412 __builtin_cpu_is nor __builtin_cpu_supports.
16413 (rs6000_get_function_versions_dispatcher): Change the warning
16414 that an old GLIBC is used which does not export the capability
16415 bits to be an error.
16416 * doc/extend.texi (target_clones attribute): Document the
16417 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16418 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16419 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16420 the macros defined by GCC if the newer GLIBC is available.
16422 2017-07-12 Jeff Law <law@redhat.com>
16424 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16425 remaining includes slightly.
16426 * config/riscv/riscv-builtins.c: Include profile-count.h.
16428 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16431 * config/avr/avr.c (avr_set_current_function): In diagnostic
16432 messages: Quote keywords and (parts of) identifiers.
16433 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16436 2017-07-12 Carl Love <cel@us.ibm.com>
16438 * config/rs6000/rs6000-c.c: Add support for built-in functions
16439 vector bool char vec_revb (vector bool char);
16440 vector bool short vec_revb (vector short char);
16441 vector bool int vec_revb (vector bool int);
16442 vector bool long long vec_revb (vector bool long long);
16443 * doc/extend.texi: Update the built-in documentation file for the
16444 new built-in functions.
16446 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16448 * config/s390/s390.md: Remove movcc splitter.
16450 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16452 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16453 load/store on condition.
16455 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16458 * config/avr/avr.c (avr_encode_section_info)
16459 [progmem && !TREE_READONLY]: Error if progmem object needs
16462 2017-07-11 Michael Collison <michael.collison@arm.com>
16464 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16467 2017-07-11 Carl Love <cel@us.ibm.com>
16469 * config/rs6000/rs6000-c.c: Add support for builtins
16470 vector unsigned int vec_parity_lsbb (vector signed int);
16471 vector unsigned int vec_parity_lsbb (vector unsigned int);
16472 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16473 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16474 vector unsigned long long vec_parity_lsbb (vector signed long long);
16475 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16476 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16477 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16478 * doc/extend.texi: Update the built-in documentation file for the
16479 new built-in functions.
16481 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16483 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16484 (layout::m_primary_loc): New field.
16485 (layout::layout): Initialize new field. Move location filtering
16486 logic from here to...
16487 (layout::maybe_add_location_range): ...this new method. Add
16488 support for filtering to just the lines already specified by other
16490 (layout::will_show_line_p): New method.
16491 (gcc_rich_location::add_location_if_nearby): New method.
16492 (selftest::test_add_location_if_nearby): New test function.
16493 (selftest::diagnostic_show_locus_c_tests): Call it.
16494 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16497 2017-07-11 Tom de Vries <tom@codesourcery.com>
16499 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16500 (bb_first_real_insn): New function.
16501 (nvptx_single): Add extra initialization of broadcasted condition
16504 2017-07-11 Nathan Sidwell <nathan@acm.org>
16506 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16508 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16510 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16511 Binutils doc as TEXI will mess them up.
16512 * doc/invoke.texi (AVR Options): Same here.
16514 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16516 * config/sparc/sparc.opt (mfix-ut700): New option.
16517 (mfix-gr712rc): Likewise.
16518 (sparc_fix_b2bst): New variable.
16519 * doc/invoke.texi (SPARC options): Document them.
16520 (ARM options): Fix warnings.
16521 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16522 instructions to prevent sequences that can trigger the store-store
16523 errata for certain LEON3FT processors.
16524 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16525 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16526 * config/sparc/sparc.md (fix_b2bst): New attribute.
16527 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16529 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16532 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16534 (*rsqrtsf2_sse): Ditto.
16536 (div<mode>3): Macroize insn from divdf3 and divsf3
16537 using MODEF mode iterator.
16539 2017-07-10 Martin Sebor <msebor@redhat.com>
16541 PR tree-optimization/80397
16542 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16543 instead of testing for equality to INTEGER_TYPE.
16545 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16547 * config.gcc: Remove uclibc from arc target spec.
16549 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16551 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16553 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16556 * lto-wrapper.c (remove_option): New function.
16557 (merge_and_complain): Merge PIC/PIE options more realistically.
16559 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16561 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16565 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16566 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16567 * config.in: Regenerate.
16568 * configure: Regenerate.
16569 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16570 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16571 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16572 (TARGET_GASISR_PROLOGUES): ...target mask.
16573 * common/config/avr/avr-common.c
16574 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16575 Set -mgas-isr-prologues.
16576 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16577 INSERT_PASS_BEFORE for it.
16578 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16579 * config/avr/avr.c (avr_option_override)
16580 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16581 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16582 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16583 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16584 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16586 (make_avr_pass_pre_proep): New function.
16587 (emit_push_sfr) <treg>: Add argument to function and use it
16588 instead of TMP_REG.
16589 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16590 and set machine->gasisr.yes.
16591 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16592 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16593 __gcc_isr.n_pushed to .L__stack_usage.
16594 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16595 (avr_asm_final_postscan_insn): ...this new static function.
16596 * config/avr/avr.h (machine_function)
16597 <is_no_gccisr, use_L__stack_usage>: New fields.
16598 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16599 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16600 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16601 (gasisr, *gasisr): New expander and insn.
16602 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16603 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16604 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16606 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16608 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16611 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16613 Move jump-tables out of .text again.
16616 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16617 (ASM_OUTPUT_ADDR_VEC): New function.
16618 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16619 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16620 INSN_ADDRESSes as asm comment.
16621 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16622 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16623 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16624 * config/avr/avr.md (*tablejump): Adjust comment.
16625 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16626 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16628 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16629 (avr_output_addr_vec): New proto.
16630 (avr_log_t) <insn_addresses>: New field.
16632 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16635 * config/i386/i386.c (ix86_function_arg_advance): Set
16636 outgoing_args_on_stack to true if there are outgoing arguments
16638 (ix86_function_arg): Likewise.
16639 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16640 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16641 * config/i386/i386.h (machine_function): Add
16642 outgoing_args_on_stack.
16644 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16646 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16647 supporting pthreds.
16648 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16650 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16652 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16653 (REAL_H): Remove $(MACHMODE_H).
16654 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
16656 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
16657 $(MACHMODE_H) and double-int.h.
16658 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
16660 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
16663 2017-07-07 Andrew Pinski <apinski@cavium.com>
16665 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
16666 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
16668 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16670 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16671 Add warning if GCC was not configured to link against a GLIBC that
16672 exports the hardware capability bits.
16673 (make_resolver_func): Make resolver function private and not a
16674 COMDAT function. Create the name with clone_function_name instead
16675 of make_unique_name.
16678 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16679 correct operand in doing the split.
16681 2017-07-07 Carl Love <cel@us.ibm.com>
16683 * config/rs6000/rs6000-c: Add support for built-in function
16684 vector unsigned short vec_pack_to_short_fp32 (vector float,
16686 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16687 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16688 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16689 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16690 (convert_4f32_8i16): Add define_expand.
16691 * doc/extend.texi: Update the built-in documentation file for the
16692 new built-in function.
16694 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16696 * config/sparc/m8.md: New file.
16697 * config/sparc/sparc.md: Include m8.md.
16699 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16701 * config/sparc/sparc.opt: New option -mvis4b.
16702 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16703 (sparc_option_override): Handle VIS4B.
16704 (enum sparc_builtins): Define
16705 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16706 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16707 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16708 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16709 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16710 (check_constant_argument): New function.
16711 (sparc_vis_init_builtins): Define builtins
16712 __builtin_vis_dictunpack{8,16,32},
16713 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16714 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16715 __builtin_vis_fpcmpde{8,16,32}shl and
16716 __builtin_vis_fpcmpur{8,16,32}shl.
16717 (sparc_expand_builtin): Check that the constant operands to
16718 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16719 constant and in range.
16720 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16722 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16723 (SPARC_IMM5_P): Likewise.
16724 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16725 (enabled): Handle vis4b.
16726 (UNSPEC_DICTUNPACK): New unspec.
16727 (UNSPEC_FPCMPSHL): Likewise.
16728 (UNSPEC_FPUCMPSHL): Likewise.
16729 (UNSPEC_FPCMPDESHL): Likewise.
16730 (UNSPEC_FPCMPURSHL): Likewise.
16731 (cpu_feature): New CPU feature `vis4b'.
16732 (dictunpack{8,16,32}): New insns.
16733 (FPCSMODE): New mode iterator.
16734 (fpcscond): New code iterator.
16735 (fpcsucond): Likewise.
16736 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16737 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16738 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16739 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16740 * config/sparc/constraints.md: Define constraints `q' for unsigned
16741 2-bit integer constants and `t' for unsigned 5-bit integer
16743 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16745 (imm5_operand_dictunpack16): Likewise.
16746 (imm5_operand_dictunpack32): Likewise.
16747 (imm2_operand): Likewise.
16748 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16749 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16750 ditunpack* and fpcmp*shl builtins.
16752 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16754 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16755 * config.in: Add HAVE_AS_SPARC6 define.
16756 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16758 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16760 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16761 (CPP_CPU_SPEC): Handle m8.
16762 (ASM_CPU_SPEC): Likewise.
16763 * config/sparc/sparc-opts.h (enum processor_type): Add
16765 * config/sparc/sparc.c (m8_costs): New struct.
16766 (sparc_option_override): Handle TARGET_CPU_m8.
16767 (sparc32_initialize_trampoline): Likewise.
16768 (sparc64_initialize_trampoline): Likewise.
16769 (sparc_issue_rate): Likewise.
16770 (sparc_register_move_cost): Likewise.
16771 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16772 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16773 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16774 (CPP_CPU_SPEC): Handle M8.
16775 (ASM_CPU_SPEC): Likewise.
16776 (AS_M8_FLAG): Define.
16777 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16778 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16779 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16781 * configure: Regenerate.
16782 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16785 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16787 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16789 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16790 ("*movdi_insn_sp32"): Do not set v3pipe.
16791 ("*movsi_insn"): Likewise.
16792 ("*movdi_insn_sp64"): Likewise.
16793 ("*movsf_insn"): Likewise.
16794 ("*movdf_insn_sp32"): Likewise.
16795 ("*movdf_insn_sp64"): Likewise.
16796 ("*zero_extendsidi2_insn_sp64"): Likewise.
16797 ("*sign_extendsidi2_insn"): Likewise.
16798 ("*mov<VM32:mode>_insn"): Likewise.
16799 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16800 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16801 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16802 ("<vlop:code><VL:mode>3"): Likewise.
16803 ("*not_<vlop:code><VL:mode>3"): Likewise.
16804 ("*nand<VL:mode>_vis"): Likewise.
16805 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16806 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16807 ("one_cmpl<VL:mode>2"): Likewise.
16808 ("faligndata<VM64:mode>_vis"): Likewise.
16809 ("alignaddrsi_vis"): Likewise.
16810 ("alignaddrdi_vis"): Likweise.
16811 ("alignaddrlsi_vis"): Likewise.
16812 ("alignaddrldi_vis"): Likewise.
16813 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16814 ("bmaskdi_vis"): Likewise.
16815 ("bmasksi_vis"): Likewise.
16816 ("bshuffle<VM64:mode>_vis"): Likewise.
16817 ("cmask8<P:mode>_vis"): Likewise.
16818 ("cmask16<P:mode>_vis"): Likewise.
16819 ("cmask32<P:mode>_vis"): Likewise.
16820 ("pdistn<P:mode>_vis"): Likewise.
16821 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16823 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16825 * config/sparc/sparc.md ("subtype"): New insn attribute.
16826 ("*wrgsr_sp64"): Set insn subtype.
16827 ("*rdgsr_sp64"): Likewise.
16828 ("alignaddrsi_vis"): Likewise.
16829 ("alignaddrdi_vis"): Likewise.
16830 ("alignaddrlsi_vis"): Likewise.
16831 ("alignaddrldi_vis"): Likewise.
16832 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16833 ("fexpand_vis"): Likewise.
16834 ("fpmerge_vis"): Likewise.
16835 ("faligndata<VM64:mode>_vis"): Likewise.
16836 ("bshuffle<VM64:mode>_vis"): Likewise.
16837 ("cmask8<P:mode>_vis"): Likewise.
16838 ("cmask16<P:mode>_vis"): Likewise.
16839 ("cmask32<P:mode>_vis"): Likewise.
16840 ("fchksm16_vis"): Likewise.
16841 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16842 ("fmean16_vis"): Likewise.
16843 ("fp<plusminus_insn>64_vis"): Likewise.
16844 ("<plusminus_insn>v8qi3"): Likewise.
16845 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16846 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16847 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16848 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16849 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16850 ("*movqi_insn"): Likewise.
16851 ("*movhi_insn"): Likewise.
16852 ("*movsi_insn"): Likewise.
16853 ("movsi_pic_gotdata_op"): Likewise.
16854 ("*movdi_insn_sp32"): Likewise.
16855 ("*movdi_insn_sp64"): Likewise.
16856 ("movdi_pic_gotdata_op"): Likewise.
16857 ("*movsf_insn"): Likewise.
16858 ("*movdf_insn_sp32"): Likewise.
16859 ("*movdf_insn_sp64"): Likewise.
16860 ("*zero_extendhisi2_insn"): Likewise.
16861 ("*zero_extendqihi2_insn"): Likewise.
16862 ("*zero_extendqisi2_insn"): Likewise.
16863 ("*zero_extendqidi2_insn"): Likewise.
16864 ("*zero_extendhidi2_insn"): Likewise.
16865 ("*zero_extendsidi2_insn_sp64"): Likewise.
16866 ("ldfsr"): Likewise.
16867 ("prefetch_64"): Likewise.
16868 ("prefetch_32"): Likewise.
16869 ("tie_ld32"): Likewise.
16870 ("tie_ld64"): Likewise.
16871 ("*tldo_ldub_sp32"): Likewise.
16872 ("*tldo_ldub1_sp32"): Likewise.
16873 ("*tldo_ldub2_sp32"): Likewise.
16874 ("*tldo_ldub_sp64"): Likewise.
16875 ("*tldo_ldub1_sp64"): Likewise.
16876 ("*tldo_ldub2_sp64"): Likewise.
16877 ("*tldo_ldub3_sp64"): Likewise.
16878 ("*tldo_lduh_sp32"): Likewise.
16879 ("*tldo_lduh1_sp32"): Likewise.
16880 ("*tldo_lduh_sp64"): Likewise.
16881 ("*tldo_lduh1_sp64"): Likewise.
16882 ("*tldo_lduh2_sp64"): Likewise.
16883 ("*tldo_lduw_sp32"): Likewise.
16884 ("*tldo_lduw_sp64"): Likewise.
16885 ("*tldo_lduw1_sp64"): Likewise.
16886 ("*tldo_ldx_sp64"): Likewise.
16887 ("*mov<VM32:mode>_insn"): Likewise.
16888 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16889 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16891 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16893 * config/sparc/sparc.md ("type"): New insn type viscmp.
16894 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16896 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16897 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16898 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16899 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16901 ("n7_vis_logical_11cycle"): Likewise.
16902 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16903 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16904 * config/sparc/niagara.md ("niag_vis"): Likewise.
16905 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16906 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16908 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16910 * config/sparc/sparc.md: New instruction type `bmask'.
16911 (bmaskdi_vis): Use the `bmask' type.
16912 (bmasksi_vis): Likewise.
16913 * config/sparc/ultra3.md (us3_array): Likewise.
16914 * config/sparc/niagara7.md (n7_array): Likewise.
16915 * config/sparc/niagara4.md (n4_array): Likewise.
16916 * config/sparc/niagara2.md (niag2_vis): Likewise.
16917 (niag3_vis): Likewise.
16918 * config/sparc/niagara.md (niag_vis): Likewise.
16920 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16922 * ipa-comdats.c: Remove optimize check from gate.
16923 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16924 for functions not optimized.
16925 (ipa_fn_summary_read): Skip optimize check.
16926 (ipa_fn_summary_write): Likewise.
16927 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16929 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16931 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16932 (check_callers): Give up on caller which is not optimized.
16933 (inline_small_functions): Likewise.
16934 (ipa_inline): Do not give up when not optimizing.
16935 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16936 away unoptimizes cdtors.
16937 (whole_program_function_and_variable_visibility): Do
16938 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16939 * ipa.c (process_references): Do not check optimize.
16940 (symbol_table::remove_unreachable_nodes): Update optimize check.
16941 (set_writeonly_bit): Update optimize check.
16942 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16943 (pass_ipa_single_use::gate): Remove.
16945 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16947 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16948 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16949 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16950 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16951 permute_load, permute_store, adjust_extract, adjust_splat,
16952 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16953 replace_swap_with_copy, dump_swap_insn_table,
16954 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16955 recombine_lvx_pattern, recombine_stvx_pattern,
16956 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16957 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16958 to file rs6000-p8swap.c.
16959 * config/rs6000/rs6000-p8swap.c: New file.
16960 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16961 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16962 and rs6000*-*-* targets.
16964 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16966 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
16968 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16970 * lto-wrapper.c (merge_and_complain): Do not merge
16971 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
16972 fsigned_zeros, ftrapping_math, fwrapv.
16973 (append_compiler_options): Do not track these options.
16974 (append_linker_options): Likewie
16976 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16978 * cgraphunit.c (cgraph_node::finalize_function): When
16979 !flag_toplevel_reorde set no_reorder flag.
16980 (varpool_node::finalize_decl): Likewise.
16981 (symbol_table::compile): Drop no toplevel reorder path.
16983 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16985 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
16986 edges; zero probability is not better than uninitialized.
16988 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16990 * asan.h (asan_sanitize_allocas_p): Declare.
16991 * asan.c (asan_sanitize_allocas_p): New function.
16992 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
16993 (handle_builtin_alloca): Likewise.
16994 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
16995 if !asan_sanitize_allocas_p.
16996 * params.def (asan-instrument-allocas): Add new option.
16997 * params.h (ASAN_PROTECT_ALLOCAS): Define.
16998 * opts.c (common_handle_option): Disable allocas sanitization for
17001 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17003 * asan.c: Include gimple-fold.h.
17004 (get_last_alloca_addr): New function.
17005 (handle_builtin_stackrestore): Likewise.
17006 (handle_builtin_alloca): Likewise.
17007 (asan_emit_allocas_unpoison): Likewise.
17008 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17009 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17010 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17011 (instrument_builtin_call): Pass gimple iterator to
17012 get_mem_refs_of_builtin_call.
17013 (last_alloca_addr): New global.
17014 * asan.h (asan_emit_allocas_unpoison): Declare.
17015 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17016 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17017 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17018 if function calls alloca.
17019 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17020 * gimple-fold.h (replace_call_with_value): Declare.
17021 * internal-fn.c: Include asan.h.
17022 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17023 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17025 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17027 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17028 (C_SELFTEST_FLAGS): New.
17029 (CPP_SELFTEST_FLAGS): New.
17030 (SELFTEST_DEPS): New, from deps of s-selftest.
17031 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17032 (CPP_SELFTEST_DEPS): New.
17033 (selftest): Add dependency on s-selftest-c++.
17034 (s-selftest): Rename to...
17035 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17036 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17037 than SELFTEST_FLAGS.
17038 (selftest-gdb): Rename to...
17039 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17041 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17042 (selftest-valgrind): Rename to...
17043 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17045 (selftest-valgrind): Reintroduce as an alias for
17046 selftest-c-valgrind.
17047 (s-selftest-c++): New.
17048 (selftest-c++-gdb): New.
17049 (selftest-c++-valgrind): New.
17051 2017-07-06 Olivier Hainque <hainque@adacore.com>
17053 * gcc.c (process_command): When deciding if undefined variables
17054 should be ignored when processing specs, accept "gcc -v" as well.
17056 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17058 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17059 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17061 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17063 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17064 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17065 * config/arm/arm-tables.opt: Regenerate.
17066 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17068 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17070 2017-07-06 Carl Love <cel@us.ibm.com>
17072 * ChangeLog: Clean up from mid air collision
17074 2017-07-06 Carl Love <cel@us.ibm.com>
17076 * config/rs6000/rs6000-c.c: Add support for built-in functions
17077 vector signed int vec_subc (vector signed int, vector signed int);
17078 vector signed __int128 vec_subc (vector signed __int128,
17079 vector signed __int128);
17080 vector unsigned __int128 vec_subc (vector unsigned __int128,
17081 vector unsigned __int128);
17082 vector signed int vec_sube (vector signed int, vector signed int,
17083 vector signed int);
17084 vector unsigned int vec_sube (vector unsigned int,
17085 vector unsigned int,
17086 vector unsigned int);
17087 vector signed __int128 vec_sube (vector signed __int128,
17088 vector signed __int128,
17089 vector signed__int128);
17090 vector unsigned __int128 vec_sube (vector unsigned __int128,
17091 vector unsigned __int128,
17092 vector unsigned __int128);
17093 vector signed int vec_subec (vector signed int, vector signed int,
17094 vector signed int);
17095 vector unsigned int vec_subec (vector unsigned int,
17096 vector unsigned int,
17097 vector unsigned int);
17098 vector signed __int128 vec_subec (vector signed __int128,
17099 vector signed __int128,
17100 vector signed__int128);
17101 vector unsigned __int128 vec_subec (vector unsigned __int128,
17102 vector unsigned __int128,
17103 vector unsigned __int128);
17104 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17105 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17106 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17107 BU_ALTIVEC_OVERLOAD_X definitions.
17108 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17109 * doc/extend.texi: Update the built-in documentation file for the new
17110 built-in functions.
17112 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17115 * diagnostic-show-locus.c (layout::layout): Use start and finish
17116 spelling location for the start and finish of each range.
17117 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17118 Add unused aspect param.
17119 * input.c (expand_location_1): Add "aspect" param, and use it
17120 to access the correct part of the location.
17121 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17123 (expand_location_to_spelling_point): Likewise.
17124 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17125 param, and pass it to expand_location_1.
17127 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17129 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17130 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17131 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17132 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17133 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17134 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17135 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17136 _mm_maskz_getmant_ss): New intrinsics.
17137 (__builtin_ia32_getexpss128_mask): Changed to ...
17138 __builtin_ia32_getexpss128_round ... this.
17139 (__builtin_ia32_getexpsd128_mask): Changed to ...
17140 __builtin_ia32_getexpsd128_round ... this.
17141 * config/i386/i386-builtin-types.def
17142 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17143 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17144 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17145 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17146 __builtin_ia32_getmantss_mask_round): New builtins.
17147 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17148 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17149 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17150 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17151 * config/i386/sse.md
17152 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17153 avx512f_sgetexp<mode><mask_scalar_name>
17154 <round_saeonly_scalar_name> ... this.
17155 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
17156 %0, %1, %2<round_saeonly_op3>}): Changed to ...
17157 vgetexp<ssescalarmodesuffix>
17158 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17159 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
17160 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
17161 avx512f_vgetmant<mode><mask_scalar_name>
17162 <round_saeonly_scalar_name> ... this.
17163 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17164 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
17165 vgetmant<ssescalarmodesuffix>
17166 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
17167 %0<mask_scalar_operand4>, %1, %2
17168 <round_saeonly_scalar_mask_op4>, %3} ... this.
17169 * config/i386/subst.md (mask_scalar_operand4,
17170 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
17171 round_saeonly_scalar_nimm_predicate): New subst attributes.
17173 2017-07-06 Julia Koval <julia.koval@intel.com>
17175 * config/i386/i386.c (ix86_erase_embedded_rounding):
17176 Remove code for old rounding pattern.
17178 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
17180 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
17182 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
17184 * doc/sourcebuild.texi (Test Directives, Variants of
17185 dg-require-support): Add documentation for dg-require-stack-check.
17187 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
17189 * config/i386/subst.md (mask_scalar, round_scalar,
17190 round_saeonly_scalar): New meta-templates.
17191 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
17192 round_scalar_mask_operand3, round_scalar_mask_op3,
17193 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
17194 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
17195 round_saeonly_scalar_constraint,
17196 round_saeonly_scalar_prefix): New subst attribute.
17197 * config/i386/sse.md
17198 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
17199 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
17200 <round_scalar_name> ... this.
17201 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
17202 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
17203 <round_scalar_name> ... this.
17204 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
17205 <sse>_vm<code><mode>3<mask_scalar_name>
17206 <round_saeonly_scalar_name> ... this.
17207 (v<plusminus_mnemonic><ssescalarmodesuffix>
17208 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17209 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17210 v<plusminus_mnemonic><ssescalarmodesuffix>
17211 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17212 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17213 (v<multdiv_mnemonic><ssescalarmodesuffix>
17214 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17215 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17216 v<multdiv_mnemonic><ssescalarmodesuffix>
17217 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17218 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17219 (v<maxmin_float><ssescalarmodesuffix>
17220 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17221 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17222 v<maxmin_float><ssescalarmodesuffix>
17223 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17224 %0<mask_scalar_operand3>, %1, %<iptr>2
17225 <round_saeonly_scalar_mask_op3>} ... this.
17227 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17229 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17230 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17232 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17233 Alan Hayward <alan.hayward@arm.com>
17234 David Sherwood <david.sherwood@arm.com>
17236 * combine.c (simplify_if_then_else): Remove "enum" before
17238 * compare-elim.c (can_eliminate_compare): Likewise.
17239 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17241 (aarch64_lookup_simd_builtin_type): Likewise.
17242 (aarch64_simd_builtin_type): Likewise.
17243 (aarch64_init_simd_builtin_types): Likewise.
17244 (aarch64_simd_expand_args): Likewise.
17245 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17247 (aarch64_reverse_mask): Likewise.
17248 (aarch64_simd_emit_reg_reg_move): Likewise.
17249 (aarch64_gen_adjusted_ldpstp): Likewise.
17250 (aarch64_ccmp_mode_to_code): Likewise.
17251 (aarch64_operands_ok_for_ldpstp): Likewise.
17252 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17253 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17255 (aarch64_min_divisions_for_recip_mul): Likewise.
17256 (aarch64_reassociation_width): Likewise.
17257 (aarch64_get_condition_code_1): Likewise.
17258 (aarch64_simd_emit_reg_reg_move): Likewise.
17259 (aarch64_simd_attr_length_rglist): Likewise.
17260 (aarch64_reverse_mask): Likewise.
17261 (aarch64_operands_ok_for_ldpstp): Likewise.
17262 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17263 (aarch64_gen_adjusted_ldpstp): Likewise.
17264 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17266 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17267 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17268 (arm_lookup_simd_builtin_type): Likewise.
17269 (arm_simd_builtin_type): Likewise.
17270 (arm_init_simd_builtin_types): Likewise.
17271 (arm_expand_builtin_args): Likewise.
17272 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17273 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17274 (ft32_setup_incoming_varargs): Likewise.
17275 (ft32_function_arg): Likewise.
17276 (ft32_function_arg_advance): Likewise.
17277 (ft32_pass_by_reference): Likewise.
17278 (ft32_arg_partial_bytes): Likewise.
17279 (ft32_valid_pointer_mode): Likewise.
17280 (ft32_addr_space_pointer_mode): Likewise.
17281 (ft32_addr_space_legitimate_address_p): Likewise.
17282 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17284 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17285 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17286 (iamcu_alignment): Likewise.
17287 (canonicalize_vector_int_perm): Likewise.
17288 (ix86_noce_conversion_profitable_p): Likewise.
17289 (ix86_mpx_bound_mode): Likewise.
17290 (ix86_operands_ok_for_move_multiple): Likewise.
17291 * config/microblaze/microblaze-protos.h
17292 (microblaze_expand_conditional_branch_reg): Likewise.
17293 * config/microblaze/microblaze.c
17294 (microblaze_expand_conditional_branch_reg): Likewise.
17295 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17297 (rs6000_reassociation_width): Likewise.
17298 (rs6000_invalid_binary_op): Likewise.
17299 (fusion_p9_p): Likewise.
17300 (emit_fusion_p9_load): Likewise.
17301 (emit_fusion_p9_store): Likewise.
17302 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17304 (riscv_hard_regno_mode_ok_p): Likewise.
17305 (riscv_address_insns): Likewise.
17306 (riscv_split_symbol): Likewise.
17307 (riscv_legitimize_move): Likewise.
17308 (riscv_function_value): Likewise.
17309 (riscv_hard_regno_nregs): Likewise.
17310 (riscv_expand_builtin): Likewise.
17311 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17312 (riscv_build_integer): Likewise.
17313 (riscv_split_integer): Likewise.
17314 (riscv_legitimate_constant_p): Likewise.
17315 (riscv_cannot_force_const_mem): Likewise.
17316 (riscv_regno_mode_ok_for_base_p): Likewise.
17317 (riscv_valid_base_register_p): Likewise.
17318 (riscv_valid_offset_p): Likewise.
17319 (riscv_valid_lo_sum_p): Likewise.
17320 (riscv_classify_address): Likewise.
17321 (riscv_legitimate_address_p): Likewise.
17322 (riscv_address_insns): Likewise.
17323 (riscv_load_store_insns): Likewise.
17324 (riscv_force_binary): Likewise.
17325 (riscv_split_symbol): Likewise.
17326 (riscv_force_address): Likewise.
17327 (riscv_legitimize_address): Likewise.
17328 (riscv_move_integer): Likewise.
17329 (riscv_legitimize_const_move): Likewise.
17330 (riscv_legitimize_move): Likewise.
17331 (riscv_address_cost): Likewise.
17332 (riscv_subword): Likewise.
17333 (riscv_output_move): Likewise.
17334 (riscv_canonicalize_int_order_test): Likewise.
17335 (riscv_emit_int_order_test): Likewise.
17336 (riscv_function_arg_boundary): Likewise.
17337 (riscv_pass_mode_in_fpr_p): Likewise.
17338 (riscv_pass_fpr_single): Likewise.
17339 (riscv_pass_fpr_pair): Likewise.
17340 (riscv_get_arg_info): Likewise.
17341 (riscv_function_arg): Likewise.
17342 (riscv_function_arg_advance): Likewise.
17343 (riscv_arg_partial_bytes): Likewise.
17344 (riscv_function_value): Likewise.
17345 (riscv_pass_by_reference): Likewise.
17346 (riscv_setup_incoming_varargs): Likewise.
17347 (riscv_print_operand): Likewise.
17348 (riscv_elf_select_rtx_section): Likewise.
17349 (riscv_save_restore_reg): Likewise.
17350 (riscv_for_each_saved_reg): Likewise.
17351 (riscv_register_move_cost): Likewise.
17352 (riscv_hard_regno_mode_ok_p): Likewise.
17353 (riscv_hard_regno_nregs): Likewise.
17354 (riscv_class_max_nregs): Likewise.
17355 (riscv_memory_move_cost): Likewise.
17356 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17357 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17358 (rl78_addr_space_address_mode): Likewise.
17359 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17361 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17362 (rs6000_reassociation_width): Likewise.
17363 (rs6000_invalid_binary_op): Likewise.
17364 (fusion_p9_p): Likewise.
17365 (emit_fusion_p9_load): Likewise.
17366 (emit_fusion_p9_store): Likewise.
17367 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17368 (ok_for_simple_move_operands): Likewise.
17369 (ok_for_simple_move_strict_operands): Likewise.
17370 (ok_for_simple_arith_logic_operands): Likewise.
17371 (visium_legitimize_reload_address): Likewise.
17372 (visium_select_cc_mode): Likewise.
17373 (output_cbranch): Likewise.
17374 (visium_split_double_move): Likewise.
17375 (visium_expand_copysign): Likewise.
17376 (visium_expand_int_cstore): Likewise.
17377 (visium_expand_fp_cstore): Likewise.
17378 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17379 (visium_function_arg): Likewise.
17380 (visium_function_arg_advance): Likewise.
17381 (visium_libcall_value): Likewise.
17382 (visium_setup_incoming_varargs): Likewise.
17383 (visium_legitimate_constant_p): Likewise.
17384 (visium_legitimate_address_p): Likewise.
17385 (visium_legitimize_address): Likewise.
17386 (visium_secondary_reload): Likewise.
17387 (visium_register_move_cost): Likewise.
17388 (visium_memory_move_cost): Likewise.
17389 (prepare_move_operands): Likewise.
17390 (ok_for_simple_move_operands): Likewise.
17391 (ok_for_simple_move_strict_operands): Likewise.
17392 (ok_for_simple_arith_logic_operands): Likewise.
17393 (visium_function_value_1): Likewise.
17394 (rtx_ok_for_offset_p): Likewise.
17395 (visium_legitimize_reload_address): Likewise.
17396 (visium_split_double_move): Likewise.
17397 (visium_expand_copysign): Likewise.
17398 (visium_expand_int_cstore): Likewise.
17399 (visium_expand_fp_cstore): Likewise.
17400 (visium_split_cstore): Likewise.
17401 (visium_select_cc_mode): Likewise.
17402 (visium_split_cbranch): Likewise.
17403 (output_cbranch): Likewise.
17404 (visium_print_operand_address): Likewise.
17405 * expmed.c (flip_storage_order): Likewise.
17406 * expmed.h (emit_cstore): Likewise.
17407 (flip_storage_order): Likewise.
17408 * genrecog.c (validate_pattern): Likewise.
17409 * hsa-gen.c (gen_hsa_addr): Likewise.
17410 * internal-fn.c (expand_arith_overflow): Likewise.
17411 * ira-color.c (allocno_copy_cost_saving): Likewise.
17412 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17413 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17414 (process_invariant_for_inheritance): Likewise.
17415 * lra-eliminations.c (move_plus_up): Likewise.
17416 * omp-low.c (lower_oacc_reductions): Likewise.
17417 * simplify-rtx.c (simplify_subreg): Likewise.
17418 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17419 (TARGET_CHKP_BOUND_MODE): Likewise..
17420 * targhooks.c (default_chkp_bound_mode): Likewise.
17421 (default_setup_incoming_vararg_bounds): Likewise.
17422 * targhooks.h (default_chkp_bound_mode): Likewise.
17423 (default_setup_incoming_vararg_bounds): Likewise.
17424 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17425 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17426 (have_whole_vector_shift): Likewise.
17427 * tree-vect-stmts.c (vectorizable_load): Likewise.
17428 * doc/tm.texi: Regenerate.
17430 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17432 Graceful degrade if Binutils PR21472 is not available.
17435 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17436 .rodata in flash test fails.
17437 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17438 * confgure: Regenerate.
17439 * config.in: Regenerate.
17440 * config/avr/avr.c (avr_asm_named_section)
17441 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17442 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17443 (avr_asm_init_sections): Same.
17445 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17447 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17448 (fma<VH:mode>4_intrinsic): Likewise.
17449 (*fmsub<VCVTF:mode>4): Likewise.
17450 (*fmsub<VH:mode>4_intrinsic): Likewise.
17452 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17455 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17456 Don't depend on "optimize > 0".
17457 (out_movhi_r_mr, out_movqi_mr_r): Same.
17458 (out_movhi_mr_r, out_movqi_r_mr): Same.
17459 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17460 io_address_operand on "optimize > 0".
17462 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17464 * tree-loop-distribution.c: Add general explanantion on the pass.
17465 (generate_loops_for_partition): Mark distributed loop.
17466 (pg_add_dependence_edges): New parameter. Handle alias data
17467 dependence specially and record it in the parameter if asked.
17468 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17469 (init_partition_graph_vertices, add_partition_graph_edge): New.
17470 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17471 (free_partition_graph_vdata, build_partition_graph): New.
17472 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17473 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17474 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17475 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17476 (version_for_distribution_p, finalize_partitions): New.
17477 (distribute_loop): Handle alias data dependence specially. Factor
17478 out loop fusion code as functions and call these functions.
17480 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17482 * tree-loop-distribution.c (classify_partition): New parameter and
17483 better handle reduction statement.
17484 (rdg_build_partitions): Revise comment.
17485 (distribute_loop): Compute statements in all partitions and pass it
17486 to classify_partition.
17488 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17490 * tree-loop-distribution.c (enum partition_type): New.
17491 (struct partition): New field type.
17492 (partition_merge_into): Add parameter. Update partition type.
17493 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17494 (build_rdg_partition_for_vertex): Compute partition type.
17495 (rdg_build_partitions): Dump partition type.
17496 (distribute_loop): Update calls to partition_merge_into.
17498 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17500 * tree-loop-distribution.c (struct ddr_hasher): New.
17501 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17503 (classify_partition): Call get_data_dependence.
17504 (pg_add_dependence_edges): Ditto.
17505 (distribute_loop): Release data dependence hash table.
17507 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17509 * tree-loop-distribution.c (ref_base_address): Delete.
17510 (similar_memory_accesses): Rename ...
17511 (share_memory_accesses): ... to this. Check if partitions access
17512 the same memory reference.
17513 (distribute_loop): Call share_memory_accesses.
17515 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17517 * tree-loop-distribution.c (struct partition): New field recording
17518 its data reference.
17519 (partition_alloc, partition_free): Init and release data refs.
17520 (partition_merge_into): Merge data refs.
17521 (build_rdg_partition_for_vertex): Collect data refs for partition.
17522 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17524 (distribute_loop): Remve data refs from vertice data of partition
17527 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17529 * tree-loop-distribution.c (params.h): Include header file.
17530 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17531 (datarefs_vec): New global var.
17532 (create_rdg_vertices): Use datarefs_vec directly.
17533 (free_rdg): Don't free data references.
17534 (build_rdg): Update use. Don't free data references.
17535 (distribute_loop): Compute global variable for data references.
17536 Bail out if there are too many data references.
17538 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17540 * tree-loop-distribution.c (loop_nest): New global var.
17541 (build_rdg): Use loop directly, rather than loop nest.
17542 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17544 (distribute_loop): Compute global variable loop nest. Update use.
17546 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17548 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17549 (partition_merge_into): New parameter. Dump reason for fusion.
17550 (distribute_loop): Update use of partition_merge_into.
17552 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17554 * tree-loop-distribution.c (bb_top_order_index): New.
17555 (bb_top_order_index_size, bb_top_order_cmp): New.
17556 (stmts_from_loop): Use topological order.
17557 (pass_loop_distribution::execute): Compute and release topological
17558 order for basic blocks.
17560 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17562 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17565 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17567 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17568 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17569 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17570 * internal-fn.def (LOOP_DIST_ALIAS): New.
17571 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17572 (fold_loop_internal_call): ... this.
17573 (vect_loop_dist_alias_call): New function.
17574 (set_uid_loop_bbs): Call fold_loop_internal_call.
17575 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17578 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17581 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17582 Require dead FLAGS_REG at the beginning of a peephole.
17584 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17587 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17588 arguments in the call to __builtin_ia32_sbb_u32.
17589 (_subborrow_u64): Swap _X and _Y arguments in the call to
17590 __builtin_ia32_sbb_u64.
17592 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17595 * tree-vrp.c (compare_assert_loc): Turn into a function template
17596 with stable template parameter. Only test if a->e is NULL,
17597 !a->e == !b->e has been verified already. Use e == NULL or
17598 e != NULL instead of e or ! e tests. If stable is true, don't use
17599 iterative_hash_expr, on the other side allow a or b or both NULL
17600 and sort the NULLs last.
17601 (process_assert_insertions): Sort using compare_assert_loc<false>
17602 instead of compare_assert_loc, later sort using
17603 compare_assert_loc<true> before calling process_assert_insertions_for
17604 in a loop. Use break instead of continue once seen NULL pointer.
17606 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17608 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17609 Cortex-R7 and Cortex-R8 processors.
17611 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17613 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17614 uninitialized while src is not.
17616 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17618 * common/config/arm/arm-common.c: Adjust include path for
17620 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17621 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17622 (arm-cpu-data.h): Likewise.
17623 (arm-cpu-cdata.h): Likewise.
17624 * config/arm/arm-cpu.h: Delete.
17625 * config/arm/arm-cpu-cdata.h: Delete.
17626 * config/arm/arm-cpu-data.h: Delete.
17628 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17630 * config/arm/arm-cpus.in (cortex-a55): New.
17631 (cortex-a75): Likewise.
17632 (cortex-a75.cortex-a55): Likewise.
17633 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17635 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17636 * config/arm/arm-cpu-cdata.h: Regenerate.
17637 * config/arm/arm-cpu-data.h: Regenerate.
17638 * config/arm/arm-cpu.h: Regenerate.
17639 * config/arm/arm-tables.opt: Regenerate.
17640 * config/arm/arm-tune.md: Regenerate.
17642 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17644 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17646 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17648 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17651 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17653 PR tree-optimization/81292
17654 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
17655 full_string_p, also call adjust_related_strinfos if the adjustment
17656 is simple, otherwise invalidate related strinfos.
17658 2017-07-04 Martin Liska <mliska@suse.cz>
17661 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
17662 newly created variable as DECL_IGNORED_P.
17664 2017-07-04 Martin Liska <mliska@suse.cz>
17667 * ipa-inline.c (inline_small_functions):
17668 Use xstrdup_for_dump.
17670 2017-07-04 Tom de Vries <tom@codesourcery.com>
17672 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17674 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17677 * config/darwin.c (darwin_function_switched_text_sections):
17680 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17682 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17684 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17686 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17688 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17690 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17691 min_profitable_iters, and th as inclusive lower bounds.
17692 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17693 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17694 for min_profitable_iters and min_profitable_estimate.
17695 (vect_transform_loop): Treat th as an inclusive lower bound.
17696 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17698 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17701 * config/darwin.c (darwin_function_switched_text_sections):
17702 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17703 in two pieces, and suppress the use of buf.
17705 2017-07-03 Nathan Sidwell <nathan@acm.org>
17707 * hash-table.h (hash_table_mod1): Fix indentation.
17709 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17711 PR middle-end/81290
17712 * predict.c (force_edge_cold): Be more careful about propagation
17714 * profile-count.h (profile_probability::guessed,
17715 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17717 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17719 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17721 * doc/invoke.texi (rcpc architecture extension): Document it.
17723 2017-07-03 Richard Biener <rguenther@suse.de>
17725 PR tree-optimization/60510
17726 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17727 the scalar reduction PHI and use it.
17728 (vectorizable_reduction): Properly guard the single_defuse_cycle
17729 path for non-SLP reduction chains where we cannot use it.
17730 Rework reduc_def/index and vector type deduction. Rework
17731 vector operand gathering during reduction op code-gen.
17732 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17733 chains dissolve the chain and leave it to non-SLP reduction
17736 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17738 * tree-data-ref.h (dr_alignment): Declare.
17739 * tree-data-ref.c (dr_alignment): New function.
17740 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17741 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17743 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17745 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17747 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17748 and base_misalignment fields.
17749 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17750 * tree-data-ref.c: Include builtins.h.
17751 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17752 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17753 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17754 * tree-vect-data-refs.c: Include tree-cfg.h.
17755 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17756 fields instead of calculating an alignment here.
17757 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17758 innermost_loop_behavior fields.
17760 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17762 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17764 (DR_STEP_ALIGNMENT): New macro.
17765 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17766 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17767 (create_data_ref): Print it.
17768 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17769 to tell whether the step preserves vector (mis)alignment.
17770 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17771 Move the check for an integer step and generalise to all INTEGER_CST.
17772 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17773 Print the outer step alignment.
17775 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17777 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17778 with offset_alignment.
17779 (DR_ALIGNED_TO): Delete.
17780 (DR_OFFSET_ALIGNMENT): New macro.
17781 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17782 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17783 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17784 (create_data_ref): Likewise.
17785 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17786 (vect_analyze_data_refs): Likewise.
17787 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17788 creating dummy innermost behavior.
17790 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17792 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17793 with a "innermost_loop_behavior *" and refeence tree.
17794 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17795 (create_data_ref): Update call accordingly.
17796 * tree-predcom.c (find_looparound_phi): Likewise.
17798 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17800 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17801 fields with dr_wrt_vec_loop.
17802 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17803 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17804 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17805 (vect_dr_behavior): New function.
17806 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17807 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17808 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17809 track whether the step preserves the misalignment.
17810 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17811 Use vect_dr_behavior.
17812 (vect_setup_realignment): Update call accordingly.
17813 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17814 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17815 call to vect_create_addr_base_for_vector_ref.
17816 (vect_create_cond_for_align_checks): Likewise.
17817 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17818 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17819 (vect_recog_mask_conversion_pattern): Likewise.
17820 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17821 (new_stmt_vec_info): Remove redundant zeroing.
17823 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17825 * common/config/arm/arm-common.c (arm_be8_option): New function.
17826 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17827 (ISA_ARMv6): Add isa_bit_be8.
17828 * config/arm/arm.h (arm_be8_option): Add prototype.
17829 (BE8_SPEC_FUNCTION): New define.
17830 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17831 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17832 (mlittle-endian): Similarly.
17833 (mbe8, mbe32): New options.
17834 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17835 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17837 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17839 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17841 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17843 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17844 (cleanup_tree_cfg_bb): Use it.
17845 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17847 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17849 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17852 * loop-doloop.c (add_test): Update profile.
17854 2017-07-03 Martin Liska <mliska@suse.cz>
17857 * sanopt.c (rewrite_usage_of_param): New function.
17858 (sanitize_rewrite_addressable_params): Likewise.
17859 (pass_sanopt::execute): Call rewrite_usage_of_param.
17861 2017-07-03 Richard Biener <rguenther@suse.de>
17863 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17864 back to using VIEW_CONVERT_EXPR.
17866 2017-07-03 Martin Liska <mliska@suse.cz>
17869 * doc/extend.texi: Document when a resolver function is
17870 generated for target_clones.
17872 2017-07-03 Martin Liska <mliska@suse.cz>
17874 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17875 and shadow memory of auto variables which are subject of
17876 use-after-scope sanitization.
17877 (asan_expand_mark_ifn): Add do set only when is_poison.
17879 2016-07-03 Richard Biener <rguenther@suse.de>
17881 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17883 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17884 (vectorizable_reduction): Perform reduction PHI creation when
17885 visiting a reduction PHI and adjust and simplify code generation
17886 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17887 (vect_transform_loop): Visit reduction PHIs.
17888 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17889 defs into the SLP tree.
17890 (vect_build_slp_tree): Reduction defs terminate the recursion.
17891 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17893 (vect_get_vec_defs_for_stmt_copy): Export.
17894 (vect_get_vec_defs): Likewise.
17895 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17897 (vect_get_vec_defs_for_stmt_copy): Declare.
17898 (vect_get_vec_defs): Likewise.
17900 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17902 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17903 parameter with a "loop" parameter and use it instead of the
17904 loop containing DR_STMT. Don't check simple_iv when doing
17905 BB analysis. Describe the two analysis modes in the comment.
17907 2017-07-03 Tom de Vries <tom@codesourcery.com>
17909 PR tree-optimization/69468
17910 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17911 (find_same_succ_bb): Handle ignore_edge_flags.
17913 2017-07-03 Tom de Vries <tom@codesourcery.com>
17915 PR tree-optimization/81192
17916 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17918 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17920 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17922 2017-07-03 Tom de Vries <tom@codesourcery.com>
17924 PR tree-optimization/81192
17925 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17926 BB_SAME_SUCC (bb) == NULL.
17928 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17930 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17933 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17935 * dumpfile.c: Include profile-count.h
17936 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17938 (insert_cond_bb): Update profile.
17939 * tree-cfg.h (insert_cond_bb): Update prototype.
17940 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17941 * tree-dump.c: Do not include tree-cfg.
17943 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17945 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17947 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17949 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17952 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17954 * tree-complex.c (expand_complex_div_wide): update profile.
17956 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17957 Alan Hayward <alan.hayward@arm.com>
17958 David Sherwood <david.sherwood@arm.com>
17960 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17961 (CORETYPES_H): New define.
17962 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17963 (insn-modes-inline.h, s-modes-inline-h): New rules.
17964 (generated_files): Add insn-modes-inline.h.
17965 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17966 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
17967 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
17968 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
17969 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
17970 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
17971 (build/gencodes.o, build/genconditions.o): Likewise.
17972 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
17973 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
17974 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
17975 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
17976 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
17977 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
17978 * coretypes.h: Include everything up to real.h for generators.
17979 Include insn-modes.h first. Include wide-int-print.h after
17980 wide-int.h. Include insn-modes-inline.h and then machmode.h.
17981 * machmode.h: Don't include insn-modes.h here.
17982 * function-tests.c: Remove includes of signop.h, machmode.h,
17983 double-int.h and wide-int.h.
17985 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
17987 * optc-save-gen.awk: Likewise.
17988 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
17989 * godump.c: Remove include of wide-int-print.h.
17990 * pretty-print.h: Likewise.
17991 * wide-int-print.cc: Likewise.
17992 * wide-int.cc: Likewise.
17993 * hash-map-tests.c: Remove include of signop.h.
17994 * hash-set-tests.c: Likewise.
17995 * rtl-tests.c: Likewise.
17996 * mkconfig.sh: Remove include of machmode.h.
17997 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
17999 (emit_insn_modes_inline_h): ...this new function. Emit the code
18000 into an insn-modes-inline.h header file, adding appropriate
18001 include guards and end comments.
18002 (emit_insn_modes_c_header): Remove include of machmode.h.
18003 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18005 (main): Handle -i flag and call emit_insn_modes_inline_h when
18008 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18010 * tree-ssa-strlen.c (strinfo): Rename the length field to
18011 nonzero_chars. Add a full_string_p field.
18012 (compare_nonzero_chars, zero_length_string_p): New functions.
18013 (get_addr_stridx): Add an offset_out parameter.
18014 Use compare_nonzero_chars.
18015 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18016 (new_strinfo): Update after above changes to strinfo.
18017 (set_endptr_and_length): Set full_string_p.
18018 (get_string_length): Update after above changes to strinfo.
18019 (unshare_strinfo): Update call to new_strinfo.
18020 (maybe_invalidate): Likewise.
18021 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18022 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18023 as a uhwi instead of an shwi. Update after above changes to
18024 strinfo and new_strinfo.
18025 (zero_length_string): Assert that chainsi contains full strings.
18026 Use zero_length_string_p. Update call to new_strinfo.
18027 (adjust_related_strinfos): Update after above changes to strinfo.
18028 Copy full_string_p from origsi.
18029 (adjust_last_stmt): Use zero_length_string_p.
18030 (handle_builtin_strlen): Update after above changes to strinfo and
18031 new_strinfo. Install the lhs as the string length if the previous
18032 entry didn't describe a full string.
18033 (handle_builtin_strchr): Update after above changes to strinfo
18035 (handle_builtin_strcpy): Likewise.
18036 (handle_builtin_strcat): Likewise.
18037 (handle_builtin_malloc): Likewise.
18038 (handle_pointer_plus): Likewise.
18039 (handle_builtin_memcpy): Likewise. Track nonzero characters
18040 that aren't necessarily followed by a nul terminator.
18041 (handle_char_store): Likewise.
18043 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18045 PR tree-optimization/80769
18046 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18047 for malloc and calloc. Document the new invariant that all related
18048 strinfos have delayed lengths or none do.
18049 (verify_related_strinfos): Move earlier in file.
18050 (set_endptr_and_length): New function, split out from...
18051 (get_string_length): ...here. Also set the lengths of related
18053 (zero_length_string): Assert that chainsi has known (rather than
18055 (adjust_related_strinfos): Likewise.
18057 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18059 PR tree-optimization/81136
18060 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18061 assert that two references with the same misalignment have the same
18062 compile-time misalignment if those compile-time misalignments
18065 2017-07-01 Andi Kleen <ak@linux.intel.com>
18067 * print-tree.c (print_node): Print all attributes.
18069 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18071 * cfg.c (scale_bbs_frequencies): New function.
18072 * cfg.h (scale_bbs_frequencies): Declare it.
18073 * cfgloopanal.c (single_likely_exit): Cleanup.
18074 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18076 (scale_loop_profile): Likewise.
18077 (loop_version): Likewise.
18078 (create_empty_loop_on_edge): Update.
18079 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18080 scale_loop_frequencies, scale_loop_profile, loopify,
18081 loop_version): Update prototypes.
18082 * modulo-sched.c (sms_schedule): Update.
18083 * predict.c (unlikely_executed_edge_p): Also check probability.
18084 (probably_never_executed_edge_p): Fix typo.
18085 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18086 * tree-parloops.c (gen_parallel_loop): Update.
18087 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18088 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18089 * tree-ssa-loop-split.c (split_loop): Update.
18090 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18091 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18092 (vect_loop_versioning): Update.
18093 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18095 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18097 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18099 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18101 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18102 to keep profile consistent.
18104 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18106 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18107 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18108 * profile-count.h (max_safe_multiplier): Make unsigned.
18109 (profile_count::guessed_zero): New.
18111 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18113 * bb-reorder.c (fix_up_crossing_landing_pad,
18114 fix_crossing_conditional_branches): Use make_single_succ_edge
18115 to keep profile consistent.
18117 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18119 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18122 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18125 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18126 the right scopes, make sure cond_jump isn't preserved between multiple
18127 iterations. Search for fallthru edge whenever there are 3+ edges and
18128 use find_fallthru_edge for it.
18130 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18132 Patch by Alexander Monakov <amonakov@ispras.ru>
18133 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18134 probabilities consistently.
18136 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18138 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18139 emit_cmp_and_jump_insns.
18141 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18144 * tree-inline.c (expand_call_inline): Combine profile statuses.
18146 2017-06-30 Andrew Pinski <apinski@cavium.com>
18148 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18149 fold_stmt returned true.
18151 2017-06-30 Nathan Sidwell <nathan@acm.org>
18153 * ggc.h (empty_string): Delete.
18154 * cfgexpand.c (expand_asm_stmt): Use plain "".
18155 * optabs.c (expand_asm_memory_barrier): Likewise.
18156 * stringpool.c (empty_string): Delete.
18157 (digit_vector, digit_string): Delete.
18158 (ggc_alloc_string): Use plain "", don't optimize single digit
18159 strings. Use ggc_alloc_atomic.
18161 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
18163 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
18164 comparison set and one other set, use the cost of the non-comparison
18167 2017-06-30 Nathan Sidwell <nathan@acm.org>
18169 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
18172 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
18174 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
18175 loops. Remove now unneeded calls to gimple_switch_set_label() that
18176 just set removed labels to NULL_TREE.
18178 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
18180 * tree-ssanames.c (set_range_info_raw): Abstract from ...
18181 (set_range_info): ...here. Only call set_range_info_raw if domain
18183 (set_nonzero_bits): Call set_range_info_raw.
18184 * tree-ssanames.h (set_range_info_raw): New.
18186 2017-06-30 Jakub Jelinek <jakub@redhat.com>
18189 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
18190 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
18191 of nonimmediate_operand and <store_mask_constraint> instead of m
18192 for the input operand. For V8FI iterator, always split if input
18193 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
18194 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
18195 <store_mask_predicate> instead of register_operand and
18196 <store_mask_constraint> instead of v for the input operand. Make
18197 sure both operands aren't MEMs for if not <mask_applied>.
18199 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
18201 * lto-wrapper.c (copy_file) Close both file descriptors before
18204 2017-06-30 Martin Liska <mliska@suse.cz>
18207 * multiple_target.c (create_dispatcher_calls): Make ifunc
18208 also for function that don't have calls or are not referenced.
18210 2017-06-30 Richard Biener <rguenther@suse.de>
18212 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
18213 analyze the first scalar stmt. Move vector type computation
18214 for the BB case here from ...
18215 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
18216 live operation processing in the SLP case properly.
18218 2017-06-30 Richard Biener <rguenther@suse.de>
18220 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18222 2017-06-30 Martin Liska <mliska@suse.cz>
18225 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18226 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18228 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18230 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18233 2017-06-30 Richard Biener <rguenther@suse.de>
18235 PR tree-optimization/81249
18236 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18237 condition reduction result to original scalar type.
18239 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18241 * profile-count.h (enum profile_quality): Fix typos and whitespace
18244 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18246 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18247 type for branch probabilities.
18249 2017-06-29 Julian Brown <julian@codesourcery.com>
18250 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18252 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18253 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18254 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18255 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18257 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18259 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18260 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18261 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18262 CC usage from generic code to here.
18263 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18264 CC usage into the target macros.
18266 2017-06-29 Maya Rashish <coypu@sdf.org>
18268 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18271 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18273 * arm/arm-builtins.c: Include profile-count.h
18274 * except.c (sjlj_emit_function_enter): Use
18275 profile_probability::unlikely.
18277 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18279 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18280 and tocrel_offset be pointer args rather than implicitly using
18282 (legitimate_constant_pool_address_p, rs6000_emit_move,
18283 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18284 tocrel_offset and use in toc_relative_expr_p call.
18285 (print_operand, print_operand_address): Use static tocrel_base_oac
18286 and tocrel_offset_oac.
18287 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18290 2017-06-29 Maya Rashish <coypu@sdf.org>
18292 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18294 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18296 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18297 objects, take into account only the alignment of 'op0' and 'mode1' if
18300 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18302 * ccmp.c (ccmp_tree_comparison_p): New function.
18303 (ccmp_candidate_p): Update to use above function.
18304 (get_compare_parts): New function.
18305 (expand_ccmp_next): Update to use new functions.
18306 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18308 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18309 take mode as argument.
18310 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18311 * expr.c (expand_expr_real_1): Pass mode as argument.
18313 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18315 * combine.c (combine_instructions): Print insns to dump_file, together
18318 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18320 * asan.c (asan_emit_stack_protection): Update.
18321 (create_cond_insert_point): Update.
18322 * auto-profile.c (afdo_propagate_circuit): Update.
18323 * basic-block.h (struct edge_def): Turn probability to
18324 profile_probability.
18325 (EDGE_FREQUENCY): Update.
18326 * bb-reorder.c (find_traces_1_round): Update.
18327 (better_edge_p): Update.
18328 (sanitize_hot_paths): Update.
18329 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18330 (make_single_succ_edge): Update.
18331 (check_bb_profile): Update.
18332 (dump_edge_info): Update.
18333 (update_bb_profile_for_threading): Update.
18334 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18335 probabilitycount to 0.
18336 * cfgbuild.c (compute_outgoing_frequencies): Update.
18337 * cfgcleanup.c (try_forward_edges): Update.
18338 (outgoing_edges_match): Update.
18339 (try_crossjump_to_edge): Update.
18340 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18341 (expand_gimple_tailcall): Update.
18342 (construct_init_block): Use make_single_succ_edge.
18343 (construct_exit_block): Use make_single_succ_edge.
18344 * cfghooks.c (verify_flow_info): Update.
18345 (redirect_edge_succ_nodup): Update.
18346 (split_edge): Update.
18347 (account_profile_record): Update.
18348 * cfgloopanal.c (single_likely_exit): Update.
18349 * cfgloopmanip.c (scale_loop_profile): Update.
18350 (set_zero_probability): Remove.
18351 (duplicate_loop_to_header_edge): Update.
18352 * cfgloopmanip.h (loop_version): Update prototype.
18353 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18354 (force_nonfallthru_and_redirect): Update.
18355 (update_br_prob_note): Update.
18356 (rtl_verify_edges): Update.
18357 (purge_dead_edges): Update.
18358 (rtl_lv_add_condition_to_bb): Update.
18359 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18360 * cgraphunit.c (init_lowered_empty_function): Update.
18361 (cgraph_node::expand_thunk): Update.
18362 * cilk-common.c: Include profile-count.h
18363 * dojump.c (inv): Remove.
18364 (jumpifnot): Update.
18365 (jumpifnot_1): Update.
18366 (do_jump_1): Update.
18368 (do_jump_by_parts_greater_rtx): Update.
18369 (do_compare_rtx_and_jump): Update.
18370 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18371 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18372 * dwarf2cfi.c: Include profile-count.h
18373 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18374 (sjlj_emit_dispatch_table): Likewise.
18375 * explow.c: Include profile-count.h
18376 * expmed.c (emit_store_flag_force): Update.
18377 (do_cmp_and_jump): Update.
18378 * expr.c (compare_by_pieces_d::generate): Update.
18379 (compare_by_pieces_d::finish_mode): Update.
18380 (emit_block_move_via_loop): Update.
18381 (store_expr_with_bounds): Update.
18382 (store_constructor): Update.
18383 (expand_expr_real_2): Update.
18384 (expand_expr_real_1): Update.
18385 * expr.h (try_casesi, try_tablejump): Update prototypes.
18386 * gimple-pretty-print.c (dump_probability): Update.
18387 (dump_profile): New.
18388 (dump_gimple_label): Update.
18389 (dump_gimple_bb_header): Update.
18390 * graph.c (draw_cfg_node_succ_edges): Update.
18391 * hsa-gen.c (convert_switch_statements): Update.
18392 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18393 (find_if_case_1): Update.
18394 (find_if_case_2): Update.
18395 * internal-fn.c (expand_arith_overflow_result_store): Update.
18396 (expand_addsub_overflow): Update.
18397 (expand_neg_overflow): Update.
18398 (expand_mul_overflow): Update.
18399 (expand_vector_ubsan_overflow): Update.
18400 * ipa-cp.c (good_cloning_opportunity_p): Update.
18401 * ipa-split.c (split_function): Use make_single_succ_edge.
18402 * ipa-utils.c (ipa_merge_profiles): Update.
18403 * loop-doloop.c (add_test): Update.
18404 (doloop_modify): Update.
18405 * loop-unroll.c (compare_and_jump_seq): Update.
18406 (unroll_loop_runtime_iterations): Update.
18407 * lra-constraints.c (lra_inheritance): Update.
18408 * lto-streamer-in.c (input_cfg): Update.
18409 * lto-streamer-out.c (output_cfg): Update.
18410 * mcf.c (adjust_cfg_counts): Update.
18411 * modulo-sched.c (sms_schedule): Update.
18412 * omp-expand.c (expand_omp_for_init_counts): Update.
18413 (extract_omp_for_update_vars): Update.
18414 (expand_omp_ordered_sink): Update.
18415 (expand_omp_for_ordered_loops): Update.
18416 (expand_omp_for_generic): Update.
18417 (expand_omp_for_static_nochunk): Update.
18418 (expand_omp_for_static_chunk): Update.
18419 (expand_cilk_for): Update.
18420 (expand_omp_simd): Update.
18421 (expand_omp_taskloop_for_outer): Update.
18422 (expand_omp_taskloop_for_inner): Update.
18423 * omp-simd-clone.c (simd_clone_adjust): Update.
18424 * optabs.c (expand_doubleword_shift): Update.
18425 (expand_abs): Update.
18426 (emit_cmp_and_jump_insn_1): Update.
18427 (expand_compare_and_swap_loop): Update.
18428 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18429 * predict.c (predictable_edge_p): Update.
18430 (edge_probability_reliable_p): Update.
18431 (set_even_probabilities): Update.
18432 (combine_predictions_for_insn): Update.
18433 (combine_predictions_for_bb): Update.
18434 (propagate_freq): Update.
18435 (estimate_bb_frequencies): Update.
18436 (force_edge_cold): Update.
18437 * profile-count.c (profile_count::dump): Add missing space into dump.
18438 (profile_count::debug): Add newline.
18439 (profile_count::differs_from_p): Explicitly convert to unsigned.
18440 (profile_count::stream_in): Update.
18441 (profile_probability::dump): New member function.
18442 (profile_probability::debug): New member function.
18443 (profile_probability::differs_from_p): New member function.
18444 (profile_probability::differs_lot_from_p): New member function.
18445 (profile_probability::stream_in): New member function.
18446 (profile_probability::stream_out): New member function.
18447 * profile-count.h (profile_count_quality): Rename to ...
18448 (profile_quality): ... this one.
18449 (profile_probability): New.
18450 (profile_count): Update.
18451 * profile.c (compute_branch_probabilities): Update.
18452 * recog.c (peep2_attempt): Update.
18453 * sched-ebb.c (schedule_ebbs): Update.
18454 * sched-rgn.c (find_single_block_region): Update.
18455 (compute_dom_prob_ps): Update.
18456 (schedule_region): Update.
18457 * sel-sched-ir.c (compute_succs_info): Update.
18458 * stmt.c (struct case_node): Update.
18459 (do_jump_if_equal): Update.
18460 (get_outgoing_edge_probs): Update.
18461 (conditional_probability): Update.
18462 (emit_case_dispatch_table): Update.
18463 (expand_case): Update.
18464 (expand_sjlj_dispatch_table): Update.
18465 (emit_case_nodes): Update.
18466 * targhooks.c: Update.
18467 * tracer.c (better_p): Update.
18468 (find_best_successor): Update.
18469 * trans-mem.c (expand_transaction): Update.
18470 * tree-call-cdce.c: Update.
18471 * tree-cfg.c (gimple_split_edge): Upate.
18472 (move_sese_region_to_fn): Upate.
18473 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18474 * tree-eh.c (lower_resx): Upate.
18475 (cleanup_empty_eh_move_lp): Upate.
18476 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18477 * tree-inline.c (copy_edges_for_bb): Update.
18478 (copy_cfg_body): Update.
18479 * tree-parloops.c (gen_parallel_loop): Update.
18480 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18481 (gimple_gen_time_profiler): Update.
18482 * tree-ssa-dce.c (remove_dead_stmt): Update.
18483 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18484 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18485 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18486 (unloop_loops): Update.
18487 (try_peel_loop): Update.
18488 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18489 * tree-ssa-loop-split.c (connect_loops): Update.
18490 (split_loop): Update.
18491 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18492 (hoist_guard): Update.
18493 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18494 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18495 (value_replacement): Update.
18496 * tree-ssa-reassoc.c (branch_fixup): Update.
18497 * tree-ssa-tail-merge.c (replace_block_by): Update.
18498 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18499 (create_edge_and_update_destination_phis): Update.
18500 (compute_path_counts): Update.
18501 (recompute_probabilities): Update.
18502 (update_joiner_offpath_counts): Update.
18503 (freqs_to_counts_path): Update.
18504 (duplicate_thread_path): Update.
18505 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18506 (struct switch_conv_info): Update.
18507 (gen_inbound_check): Update.
18508 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18509 (vect_do_peeling): Update.
18510 (vect_loop_versioning): Update.
18511 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18512 (optimize_mask_stores): Update.
18513 * ubsan.c (ubsan_expand_null_ifn): Update.
18514 * value-prof.c (gimple_divmod_fixed_value): Update.
18515 (gimple_divmod_fixed_value_transform): Update.
18516 (gimple_mod_pow2): Update.
18517 (gimple_mod_pow2_value_transform): Update.
18518 (gimple_mod_subtract): Update.
18519 (gimple_mod_subtract_transform): Update.
18520 (gimple_ic): Update.
18521 (gimple_stringop_fixed_value): Update.
18522 (gimple_stringops_transform): Update.
18523 * value-prof.h: Update.
18525 2017-06-29 Carl Love <cel@us.ibm.com>
18527 * config/rs6000/rs6000-c.c: Add support for built-in functions
18528 vector signed int vec_signed (vector float);
18529 vector signed long long vec_signed (vector double);
18530 vector signed int vec_signed2 (vector double, vector double);
18531 vector signed int vec_signede (vector double);
18532 vector signed int vec_signedo (vector double);
18533 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18534 instruction generator.
18535 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18536 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18537 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18539 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18540 vunsignede_v2df): Add define_expands.
18541 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18542 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18543 VEC_UNSIGNEDO): Add definitions.
18544 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18545 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18546 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18547 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18548 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18549 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18550 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18551 vec_unsignede, vec_unsignedo): Add builtin defines.
18552 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18554 * doc/extend.texi: Update the built-in documentation file for the
18555 new built-in functions.
18557 2017-06-29 Richard Biener <rguenther@suse.de>
18559 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18560 reduction chains to LOOP_VINFO_REDUCTIONS.
18561 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18562 SLP reductions after processing reduction chains.
18564 2017-06-29 Nathan Sidwell <nathan@acm.org>
18566 * builtins.c (fold_builtin_FUNCTION): Use
18567 lang_hooks.decl_printable_name.
18569 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18571 PR middle-end/81194
18572 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18573 with only one label.
18574 * stmt.c (expand_case): Assert NCASES is greater than one.
18576 2017-06-29 Richard Biener <rguenther@suse.de>
18578 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18580 (group_case_labels): Likewise.
18581 (find_taken_edge): Push sanity checking on val to workers...
18582 (find_taken_edge_cond_expr): ... here
18583 (find_taken_edge_switch_expr): ... and here, handle cases
18584 with just a default label.
18585 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18586 (group_case_labels): Likewise.
18587 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18588 group_case_labels does anything cleanup the CFG again.
18590 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18592 PR tree-optimization/81196
18593 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18594 exit condition comparing two IVs.
18596 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18598 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18599 profile to the dummy entry at the end of the list of architectures.
18600 * config/arm/arm-cpu-cdata.h: Regenerated.
18602 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18603 Michael Collison <michael.collison@arm.com>
18606 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18608 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18609 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18610 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18611 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18612 of shift when the shift amount is masked with constant equal to
18613 the size of the mode.
18614 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18617 2017-06-29 Martin Liska <mliska@suse.cz>
18619 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18621 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18624 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18627 (edge_growth_cache_entry::edge_growth_cache_entry): New
18629 (reset_edge_growth_cache): Update to use constructor.
18631 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18633 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18634 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18635 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18637 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18639 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18640 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18642 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18644 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18645 (*-linux-uclibc*): Add t-uclibc tmake_file.
18646 * config/t-musl: New.
18647 * config/t-uclibc: New.
18649 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18651 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18653 (gen_comm_data): Emit architectural setting of arch_prof.
18654 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
18656 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
18657 (armv8-m.base, armv8-m.main): Likewise.
18658 * arm-protos.h (arm_build_target): Add profile field.
18659 (arch_option): Likewise.
18660 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
18662 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
18663 arm_active_target.profile.
18665 2017-06-28 Richard Biener <rguenther@suse.de>
18667 PR middle-end/81227
18668 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
18669 TYPE_OVERFLOW_WRAPS.
18670 * match.pd (negate_expr_p): Likewise.
18671 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18672 fold_build2, not fold_binary.
18674 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18676 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18677 Convert memory address to Pmode.
18678 (aarch64_print_operand): Assert MEM operands are always Pmode.
18680 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18683 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18684 Remove redundant if.
18685 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18686 * config/arm/aarch-common-protos.h
18687 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18688 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18690 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18693 * multiple_target.c (create_dispatcher_calls): Set the default
18694 clone to be static, not public.
18696 2017-06-28 Richard Biener <rguenther@suse.de>
18698 * tree-vect-loop.c (vectorizable_reduction): Move special
18699 cond reduction IV var creation ...
18700 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18701 parameter. Use STMT_VINFO_VECTYPE.
18702 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18705 2017-06-28 Martin Liska <mliska@suse.cz>
18708 * ipa-visibility.c (non_local_p): Handle visibility.
18710 2017-06-28 Martin Liska <mliska@suse.cz>
18713 * common.opt: Add IntegerRange to various options.
18714 * opt-functions.awk (integer_range_info): New function.
18715 * optc-gen.awk: Add integer_range_info to cl_options struct.
18716 * opts-common.c (decode_cmdline_option): Handle
18717 CL_ERR_INT_RANGE_ARG.
18718 (cmdline_handle_error): Likewise.
18719 * opts.c (print_filtered_help): Show valid interval in
18720 when --help is provided.
18721 * opts.h (struct cl_option): Add range_min and range_max fields.
18722 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18724 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18726 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18727 (x * C EQ/NE y * C): New transformation.
18729 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18731 * genmultilib (combination_space): Accept '+' in option names.
18733 2017-06-28 Martin Liska <mliska@suse.cz>
18736 * asan.c (instrument_derefs): Bail out inner references
18737 that are hard register variables.
18739 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18742 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18743 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18745 2017-06-28 Richard Biener <rguenther@suse.de>
18747 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18748 (vect_get_slp_defs): Adjust.
18749 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18751 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18753 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18754 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18755 (vectorizable_reduction): Adjust.
18756 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18758 (vect_get_slp_defs): Likewise.
18759 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18760 (vectorizable_bswap): Adjust.
18761 (vectorizable_call): Likewise.
18762 (vectorizable_conversion): Likewise.
18763 (vectorizable_assignment): Likewise.
18764 (vectorizable_shift): Likewise.
18765 (vectorizable_operation): Likewise.
18766 (vectorizable_store): Likewise.
18767 (vectorizable_condition): Likewise.
18768 (vectorizable_comparison): Likewise.
18770 2017-06-28 Michael Collison <michael.collison@arm.com>
18773 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18775 (arm_gen_movmemqi): Removed unused variable 'i'.
18776 Convert 'for' loop into 'while' loop.
18777 (arm_expand_prologue): Remove last unnecessary set of insn.
18778 (thumb_pop): Remove unused variable 'pushed_words'.
18779 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18781 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18783 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18784 * config/s390/s390-protos.h: Add prototype of
18785 s390_rel_address_ok_p.
18786 * config/s390/s390.c (s390_got_symbol): New function.
18787 (s390_rel_address_ok_p): New function.
18788 (legitimize_pic_address): Use s390_rel_address_ok_p.
18789 (s390_load_got): Use s390_got_symbol.
18790 (s390_option_override): Issue error if
18791 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18792 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18794 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18796 2017-06-27 Andrew Pinski <apinski@cavium.com>
18798 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18799 (X * copysign (1.0, X)): New pattern.
18800 (X * copysign (1.0, -X)): New pattern.
18801 (copysign (-1.0, CST)): New pattern.
18803 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18805 * genmultilib (combination_space): Remove variable.
18806 Validate reuse rules against regular expression for any sequence
18807 of multilib options in any order.
18809 2017-06-27 Michael Collison <michael.collison@arm.com>
18811 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18812 call aarch64_split_simd_combine.
18813 * (aarch64_combine_internal<mode>): Delete pattern.
18814 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18815 Allow register and subreg operands.
18817 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18819 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18820 specific need, just fallback on defaults.
18821 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18823 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18824 Olivier Hainque <hainque@adacore.com>
18826 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18828 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18829 targets. Pick a default if no particular attempt applied.
18830 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18833 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18835 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18836 (x86_64-wrs-vxworks7): Likewise.
18838 2017-06-27 Marek Polacek <polacek@redhat.com>
18841 * ubsan.c (instrument_null): Check get_base_address's result for null.
18843 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18845 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18847 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18849 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18850 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18851 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18852 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18853 New function types.
18854 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18855 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18856 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18857 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18858 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18859 BUILT_IN_FEUPDATEENV): New builtins.
18860 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18861 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18862 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18863 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18865 (builtin_structptr_types): Adjust size.
18866 * tree.c (builtin_structptr_types): Add four entries.
18868 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18869 Olivier Hainque <hainque@adacore.com>
18871 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18872 (TLS_SYM): New local macro, forcing reference to __tls__ on
18873 link command lines for VxWorks 7 RTPs, triggering initialization
18875 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18876 OS features TLS support, true for RTPs on VxWorks 7.
18877 * config/vxworks.c (vxworks_override_options): Setup emutls
18880 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18882 * predict.c (test_prediction_value_range): Use -1U instead of -1
18883 to avoid narrowing conversion warning.
18884 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18885 to avoid narrowing conversion warning.
18886 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18888 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18890 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18892 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18893 64bit configurations.
18894 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18895 (SIZE_TYPE): Likewise.
18896 * config/vxworks.c (vxworks_emutls_var_fields): Use
18897 long_unsigned_type_node instead of unsigned_type_node as the offset
18898 field type, which is "pointer" mode in emutls.c.
18900 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18903 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18905 PR middle-end/81207
18906 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18907 gimple_vuse copying separately from gimple_vdef copying.
18909 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18911 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18914 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18915 Olivier Hainque <hainque@adacore.com>
18917 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18918 for all vxworks7 targets.
18919 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18920 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18921 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18922 variations for VX6/VX7 and 32/64bits later on in ...
18923 (VXWORKS_LIB_SPEC): Leverage new macros.
18924 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18925 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18927 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18929 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18930 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18932 2017-06-26 Carl Love <cel@us.ibm.com>
18934 * config/rs6000/rs6000-c.c: Add support for built-in functions
18935 vector bool char vec_reve (vector bool char);
18936 vector signed char vec_reve (vector signed char);
18937 vector unsigned char vec_reve (vector unsigned char);
18938 vector bool int vec_reve (vector bool int);
18939 vector signed int vec_reve (vector signed int);
18940 vector unsigned int vec_reve (vector unsigned int);
18941 vector bool long long vec_reve (vector bool long long);
18942 vector signed long long vec_reve (vector signed long long);
18943 vector unsigned long long vec_reve (vector unsigned long long);
18944 vector bool short vec_reve (vector bool short);
18945 vector signed short vec_reve (vector signed short);
18946 vector double vec_reve (vector double);
18947 vector float vec_reve (vector float);
18948 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18949 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18950 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18951 (altivec_vreve): New pattern.
18952 * config/rs6000/altivec.h (vec_reve): New define.
18953 * doc/extend.texi (vec_rev): Update the built-in documentation file
18954 for the new built-in functions.
18956 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18958 PR tree-optimization/71815
18959 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18961 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18963 (slsr_process_phi): Likewise.
18964 (replace_uncond_cands_and_profitable_phis): Don't replace a
18965 multiply candidate with a stride of 1 (copy or cast).
18966 (phi_incr_cost): Call uses_consumed_by_stmt rather than
18968 (lowest_cost_path): Likewise.
18969 (total_savings): Likewise.
18971 2017-06-26 Richard Biener <rguenther@suse.de>
18974 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
18975 Use def_builtin_pure for all gather builtins.
18977 2017-06-26 Richard Biener <rguenther@suse.de>
18979 PR tree-optimization/81203
18980 * tree-tailcall.c (find_tail_calls): Do not move stmts into
18981 non-dominating BBs.
18983 2017-06-26 Marek Polacek <polacek@redhat.com>
18986 * doc/invoke.texi: Document -Wmultistatement-macros.
18988 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
18990 * doc/sourcebuild.texi (ARM-specific attributes): Document new
18991 arm_neon_ok_no_float_abi effective target.
18993 2017-06-26 Richard Biener <rguenther@suse.de>
18995 PR tree-optimization/80928
18996 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
18997 (copy_bbs): Set BB_DUPLICATED flag early.
18998 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19000 (execute_on_shrinking_pred): Likewise.
19001 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19002 BB_DUPLICATED blocks.
19003 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19004 iterate over all PHIs considering removal of *gsi.
19006 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19008 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19011 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19013 * config/rs6000/rs6000-string.c: (expand_block_clear,
19014 do_load_for_compare, select_block_compare_mode,
19015 compute_current_alignment, expand_block_compare,
19016 expand_strncmp_align_check, expand_strn_compare,
19017 expand_block_move, rs6000_output_load_multiple)
19018 Move functions related to string/block move/compare
19019 to a separate file.
19020 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19021 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19022 for this function which is now used in two files.
19023 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19024 * config.gcc: Add rs6000-string.o to extra_objs for
19025 targets powerpc*-*-* and rs6000*-*-*.
19027 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19030 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19031 32-bit, since indexed is not valid for DImode.
19032 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19033 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19034 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19035 (define_peephole2 for Altivec d-form store): Likewise.
19038 * multiple_target.c (create_dispatcher_calls): Only create the
19039 dispatcher call if the function is the default clone of a
19040 versioned function.
19042 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19044 PR middle-end/80902
19045 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19046 a call, force the call to not be a tail call.
19048 2017-06-23 Jeff Law <law@redhat.com>
19050 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19053 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19055 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19056 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19057 handling for early expansion of vector shifts (sl,sr,sra,rl).
19058 (builtin_function_type): Add vector shift right instructions
19059 to the unsigned argument list.
19061 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19063 rtl-optimizatoin/79286
19064 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19065 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19066 trap. PIC register plus a const unspec without offset can never trap.
19068 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19070 * tree.h (builtin_structptr_type): New type.
19071 (builtin_structptr_types): Declare new array.
19072 * tree.c (builtin_structptr_types): New array.
19073 (free_lang_data, build_common_tree_nodes): Use it.
19075 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19078 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19081 2017-06-22 Matt Turner <mattst88@gmail.com>
19083 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19084 Lake models to skylake case. Assume skylake for unknown
19085 models with clflushopt.
19087 2017-06-22 Jeff Law <law@redhat.com>
19089 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19090 frame sizes that do not satisfy aarch64_uimm12_shift.
19092 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19094 * profile-count.h (apply_probability,
19095 apply_scale, probability_in): Fix checks for zero.
19097 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19099 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19100 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19102 2016-06-22 Richard Biener <rguenther@suse.de>
19104 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19105 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19106 REDUC_MAX_EXPR support.
19107 (vectorizable_reduction): Likewise.
19108 (vect_create_epilog_for_reduction): Likewise.
19110 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19112 * match.pd (A / (1 << B) -> A >> B): New.
19113 * generic-match-head.c: Include optabs-tree.h.
19114 * gimple-match-head.c: Likewise.
19115 * optabs-tree.h (target_supports_op_p): New.
19116 * optabs-tree.c (target_supports_op_p): New.
19118 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19120 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19121 $gcc_cv_ld --help output.
19122 (gcc_cv_ld_demangle): Likewise.
19123 (gcc_cv_ld_eh_frame_hdr): Likewise.
19124 (gcc_cv_ld_pie): Likewise.
19125 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19126 (gcc_cv_ld_buildid): Likewise.
19127 (gcc_cv_ld_sysroot): Likewise.
19128 (ld_bndplt_support): Likewise.
19129 (ld_pushpopstate_support): Likewise.
19130 * configure: Regenerate.
19131 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19133 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19136 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19137 operands indexes to avoid gap between operands and match_dups.
19139 2017-06-21 Andrew Pinski <apinski@cavium.com>
19141 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19142 Increment Arith_shift and Arith_shift_reg by 1.
19143 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19145 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19146 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19147 (aarch64_strip_extend): Add new argument and test for it.
19148 (aarch64_cheap_mult_shift_p): New function.
19149 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19150 add a cost if it is true.
19151 Update calls to aarch64_strip_extend.
19152 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19154 2017-06-21 Andrew Pinski <apinski@cavium.com>
19156 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
19158 (thunderxt88): Likewise.
19159 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
19160 (thunderx_prefetch_tune): New variable.
19161 (thunderx2t99_prefetch_tune): Update for the correct values.
19162 (thunderxt88_tunings): New variable.
19163 (thunderx_tunings): Use thunderx_prefetch_tune instead of
19164 generic_prefetch_tune.
19165 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
19167 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19169 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
19170 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
19171 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
19172 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
19173 (aarch64_atomic_cas<mode>, GPI): Likewise.
19175 2017-06-21 Martin Liska <mliska@suse.cz>
19177 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
19178 statements on cold and hot labels.
19179 * predict.c (tree_estimate_probability_bb): Remove the
19180 prediction from this place.
19182 2017-06-21 Martin Liska <mliska@suse.cz>
19184 PR tree-optimization/79489
19185 * gimplify.c (maybe_add_early_return_predict_stmt): New
19187 (gimplify_return_expr): Call the function.
19188 * predict.c (tree_estimate_probability_bb): Remove handling
19190 * predict.def: Update comment about early return predictor.
19191 * gimple-predict.h (is_gimple_predict): New function.
19192 * predict.def: Change default value of early return to 66.
19193 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
19195 * passes.def: Put pass_strip_predict_hints to the beginning of
19198 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
19200 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
19201 FUNCTION_DECL declarations.
19202 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
19204 (dwaf2out_decl): Likewise.
19205 * godump.c (go_early_global_decl): Skip call to the real debug hook
19206 for FUNCTION_DECL declarations.
19207 * passes.c (rest_of_decl_compilation): Skip call to the
19208 early_global_decl debug hook for FUNCTION_DECL declarations, unless
19209 -fdump-go-spec is passed.
19211 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19213 * config/i386/i386.c (struct builtin_isa): New field pure_p.
19214 Reorder for compactness.
19215 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
19216 (def_builtin_pure, def_builtin_pure2): New functions.
19217 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19219 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19221 * match.pd (nop_convert): New predicate.
19222 ((A +- CST1) +- CST2): Allow some NOP conversions.
19224 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19227 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19228 with ctors/dtors if GOVD_SHARED is set.
19230 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19232 * config/aarch64/aarch64.md (movti_aarch64):
19233 Emit mov rather than orr.
19234 (movtf_aarch64): Likewise.
19235 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19236 Emit mov rather than orr.
19238 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19240 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19241 Swap alternatives, make integer dup more expensive.
19243 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19245 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19246 Return true for non-tls symbols.
19248 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19250 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19251 (cortex-a75): Likewise.
19252 (cortex-a75.cortex-a55): Likewise.
19253 * config/aarch64/aarch64-tune.md: Regenerate.
19254 * doc/invoke.texi (-mtune): Document new values for -mtune.
19256 2017-06-21 Tom de Vries <tom@codesourcery.com>
19258 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19259 stack_size feature.
19260 (Effective-Target Keywords, Other attributes): Suggest using
19261 dg-add-options stack_size feature to get stack limit in stack_size
19262 effective target documentation.
19264 2017-06-21 Julian Brown <julian@codesourcery.com>
19265 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19267 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19268 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19269 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19271 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19272 attribute type list for neon_multiply.
19273 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19274 attribute type list for neon_multiply.
19275 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19276 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19277 attribute type list for neon_multiply.
19278 * config/arm/types.md (crypto_pmull): Add.
19279 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19280 attribute type list.
19282 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19284 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19287 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19289 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19290 to make sure not to dereference a NULL cost_classes_ptr pointer.
19292 2017-06-20 Carl Love <cel@us.ibm.com>
19294 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19295 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19296 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19297 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19298 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19299 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19300 VMULOSW): New enum "unspec" values.
19301 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19302 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19303 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19304 altivec_vmulosw): New patterns.
19305 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19306 VMULOSW): Add definitions.
19308 2017-06-20 Julia Koval <julia.koval@intel.com>
19310 * config/i386/i386.c: Fix rounding expand for new pattern.
19311 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19313 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19315 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19316 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19318 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19320 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19323 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19325 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19326 architecture, then by implementer ID.
19327 * config/aarch64/aarch64-tune.md: Regenerate.
19329 2017-06-20 Richard Biener <rguenther@suse.de>
19331 PR middle-end/81097
19332 * fold-const.c (split_tree): Fold to type before negating.
19334 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19336 * diagnostic-show-locus.c
19337 (selftest::test_fixit_deletion_affecting_newline): New function.
19338 (selftest::diagnostic_show_locus_c_tests): Call it.
19340 2017-06-20 Andreas Schwab <schwab@suse.de>
19343 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19346 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19348 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19349 __ARM_FEATURE_COPROC according to support.
19351 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19353 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19354 Rewritten to avoid overflow for > 32-bit pointers.
19357 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19358 by removing enum keyword.
19359 (ubsan_type_descriptor): Likewise. Formatting fix.
19362 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19363 splitter): Require TARGET_SSE2 in the condition.
19365 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19368 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19369 for doing vector set of SFmode on ISA 3.0.
19370 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19371 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19373 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19374 SFmode value into a V4SF variable that was extracted from another
19375 V4SF variable without converting the element to double precision
19376 and back to single precision vector format.
19377 (vsx_insert_extract_v4sf_p9_2): Likewise.
19379 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19381 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19382 in UWHI to avoid undefined overflow.
19385 * ubsan.h (enum ubsan_encode_value_phase): New.
19386 (ubsan_encode_value): Change second argument to
19387 enum ubsan_encode_value_phase with default value of
19388 UBSAN_ENCODE_VALUE_GENERIC.
19389 * ubsan.c (ubsan_encode_value): Change second argument to
19390 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19391 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19392 create_tmp_var_raw instead of create_tmp_var and use a
19394 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19395 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19396 ubsan_encode_value callers.
19399 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19400 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19401 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19403 2017-06-19 Richard Biener <rguenther@suse.de>
19405 PR middle-end/81118
19406 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19407 estimates if we changed anything.
19409 2017-06-19 Richard Biener <rguenther@suse.de>
19411 PR tree-optimization/80887
19412 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19413 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19414 simplified lookups, then reset mprts_hook.
19415 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19417 (try_to_simplify): Likewise.
19419 2017-06-19 Martin Liska <mliska@suse.cz>
19422 * gimplify.c (gimplify_switch_expr):
19423 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19425 2017-06-19 Martin Liska <mliska@suse.cz>
19427 * doc/install.texi: Document that PGO runs in 4 stages.
19429 2017-06-19 Martin Liska <mliska@suse.cz>
19432 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19433 to dispatcher function name.
19434 * multiple_target.c (replace_function_decl): New function.
19435 (create_dispatcher_calls): Redirect both edges and references.
19437 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19439 * profile-count.c (profile_count::dump): Dump quality.
19440 (profile_count::differs_from_p): Update for unsigned val.
19441 * profile-count.h (profile_count_quality): New enum.
19442 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19444 2017-06-19 Richard Biener <rguenther@suse.de>
19446 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19447 struct function as arg.
19448 (estimate_numbers_of_iterations): Export overload with loop arg.
19449 (free_numbers_of_iterations_estimates_loop): Use an overload of
19450 free_numbers_of_iterations_estimates instead.
19451 * tree-cfg.c (remove_bb): Adjust.
19452 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19453 * tree-parloops.c (gen_parallel_loop): Likewise.
19454 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19456 (tree_unroll_loops_completely): Likewise.
19457 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19458 Use an overload instead and export.
19459 (estimated_loop_iterations): Adjust.
19460 (max_loop_iterations): Likewise.
19461 (likely_max_loop_iterations): Likewise.
19462 (estimate_numbers_of_iterations): Take struct function as arg
19464 (loop_exits_before_overflow): Adjust.
19465 (free_numbers_of_iterations_estimates_loop): Use an overload.
19466 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19467 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19469 2017-06-19 Richard Biener <rguenther@suse.de>
19472 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19473 RANGE_EXPR conservatively.
19475 2017-06-16 Carl Love <cel@us.ibm.com>
19477 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19478 definitions for vec_float, vec_float2, vec_floato,
19479 vec_floate built-ins.
19480 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19481 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19483 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19484 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19485 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19486 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19487 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19488 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19489 vec_floato): Add builtin defines.
19490 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19491 Update the built-in documentation file for the new built-in
19494 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19496 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19497 (mthumb): Mark as the negative of -marm.
19499 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19501 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19503 (ARM Options, -mtune): Document that this accepts the same
19504 extension options as -mcpu.
19505 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19507 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19509 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19510 permitted extensions.
19512 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19514 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19515 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19516 (armv8-m.main): Add option +nodsp.
19517 * config/arm/arm-cpu-cdata.h: Regenerated.
19519 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19521 * config/arm/t-fuchsia: New file.
19522 * config.gcc (arm*-*-fuchsia*): Use it.
19524 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19526 * config/arm/t-symbian: Rewrite for new option infrastructure.
19528 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19530 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19531 (MULTILIB_REQUIRED): Likewise.
19533 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19535 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19536 (MULTILIB_RESUE): Likewise.
19537 (MULTILIB_MATCHES): Likewise.
19538 (MULTLIB_REQUIRED): Likewise.
19540 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19542 * config/arm/t-rtems: Rewrite for new option framework.
19544 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19546 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19547 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19548 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19549 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19550 * config/arm/t-multilib: ... here.
19551 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19552 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19553 armv7-a and armv8*-a when A-profile libraries have not been built.
19554 * config/arm/t-rmprofile: Rewrite.
19556 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19558 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19559 with a backslash. Remove the backslash after substituting unescaped
19561 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19563 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19565 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19566 the driver to pass through the default setting of -mfloat-abi.
19567 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19569 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19570 (all_feat_combs): New rule.
19571 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19573 * config/arm/t-aprofile: Rewrite.
19575 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19577 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19578 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19579 fpu is not specified by the user/command-line.
19580 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19581 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19582 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19583 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19584 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19585 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19587 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19589 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19590 * config/arm/t-arm-elf: Rewritten.
19592 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19594 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19595 have some floating-point instructions.
19596 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19597 (TARGET_MAYBE_HARD_FLOAT): New macro.
19598 * config/arm/arm-builtins.c (arm_init_builtins): Use
19599 TARGET_MAYBE_HARD_FLOAT.
19600 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19602 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19604 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19605 (configargs.h): Include it.
19606 (arm_print_hint_for_fpu_option): New function.
19607 (arm_parse_fpu_option): New function.
19608 (candidate_extension): New class.
19609 (arm_canon_for_multilib): New function.
19610 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19611 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19612 (ARCH_CANONICAL_SPECS): New macro.
19613 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19615 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19617 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19618 are set after handling multilib fragments. Set target_cpu_default2
19621 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19623 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19625 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19626 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19628 (check_arch): Likewise.
19629 * config/arm/arm.c (arm_configure_build_target): Handle
19630 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19631 options in the default.
19633 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19635 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19636 when an option is an alias of another.
19637 * config/arm/parsecpu.awk (optalias): New parser token.
19638 (gen_comm_data): Mark non-alias options as such. Emit entries
19639 for extension aliases.
19640 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19641 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19642 (armv6kz, armv6zk, armv6t2): Likewise.
19643 (armv7): Make vfpv3-d16 an alias.
19644 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19646 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19647 Sort in canonical order.
19648 (armv8-a): Sort in canonical order.
19649 (armv8.1-a, armv8.2-a): Likewise.
19650 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19652 (cortex-a9): Sort in canonical order.
19653 * config/arm/arm.c (selftests.h): Include it.
19654 (arm_test_cpu_arch_data): New function.
19655 (arm_run_self_tests): New function.
19656 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
19657 (targetm): Move declaration to the end of the file.
19658 * arm-cpu-cdata.h: Regenerated.
19660 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19662 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
19663 call to target_mode_check describing the type of option passed.
19664 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
19665 (arm_target_thumb_only): Use arm_parse_arch_option_name or
19666 arm_parse_cpu_option_name to match parameters against list of
19668 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
19669 arm_arch_core_flags data structure.
19670 * config/arm/arm-cpu_cdata.h: Regenerated.
19672 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19674 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19676 (arm_print_hint_for_cpu_option): Likewise.
19677 (arm_print_hint_for_arch_option): Likewise.
19678 (arm_parse_cpu_option_name): Likewise.
19679 (arm_parse_arch_option_name): Likewise.
19680 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19681 of entries in the all_fpus list.
19682 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19683 (arm_parse_cpu_option_name): Declare.
19684 (arm_parse_arch_option_name): Declare.
19685 (arm_parse_option_features): Declare.
19686 (arm_intialize_isa): Declare.
19687 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19689 (gen_comm_data): ... here. Make definitions non-static.
19690 * config/arm/arm-cpu-data.h: Regenerated.
19691 * config/arm/arm-cpu-cdata.h: Regenerated.
19693 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19695 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19696 (cpu_arch_extension): New structure.
19697 (cpu_arch_option, arch_option, cpu_option): New structures.
19698 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19699 architecture types.
19700 (gen_data): Generate new format data tables.
19701 * config/arm/arm.c (cpu_tune): New structure.
19702 (cpu_option, processors): Delete.
19703 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19704 (arm_print_hint_for_cpu_option): ... this and ...
19705 (arm_print_hint_for_arch_option): ... this.
19706 (arm_parse_arch_cpu_name): Delete. Replace with ...
19707 (arm_parse_cpu_option_name): ... this and ...
19708 (arm_parse_arch_option_name): ... this.
19709 (arm_unrecognized_feature): Change type of target parameter to
19711 (arm_parse_arch_cpu_features): Delete. Replace with ...
19712 (arm_parse_option_features): ... this.
19713 (arm_configure_build_target): Rework to use new configuration data
19715 (arm_print_tune_info): Rework for new configuration data tables.
19716 * config/arm/arm-cpu-data.h: Regenerated.
19717 * config/arm/arm-cpu.h: Regenerated.
19719 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19721 * Makefile.in (OBJS): Move sbitmap.o from here ...
19722 (OBJS-libcommon): ... to here.
19724 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19726 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19727 (ISA_ALL_CRYPTO): New macro.
19728 (ISA_ALL_SIMD): New macro
19729 (ISA_ALL_FP): New macro.
19730 * config/arm/arm.c (fpu_bitlist): Update initializer.
19731 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19733 (arm9e): Add fpu. Add option for nofp
19734 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19735 (arm926ej-s, arm1026ej-s): Likewise.
19736 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19737 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19738 neon-fp16, neon-vfpv4, nofp and nosimd.
19739 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19740 (cortex-a8): Add fpu. Add option for nofp.
19741 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19742 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19743 (cortex-r4f): Add fpu.
19744 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19745 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19747 (cortex-r8): Likewise.
19748 (cortex-m4): Add fpu. Add option for nofp.
19749 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19750 (cortex-a17.cortex-a7): Likewise.
19751 (cortex-a32): Add fpu. Add options for crypto and nofp.
19752 (cortex-a35, cortex-a53): Likewise.
19753 (cortex-a57): Add fpu. Add option for crypto.
19754 (cortex-a72, cortex-a73): Likewise.
19755 (exynos-m1): Likewise.
19756 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19757 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19758 (cortex-m33): Add fpu. Add option for nofp.
19759 * config/arm/arm-cpu-cdata.h: Regenerated
19760 * config/arm/arm-cpu-data.h: Regenerated.
19762 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19764 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19765 (armv5te, armv5tej): Likewise.
19766 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19767 (armv7): Add options fp and vfpv3-d16.
19768 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19769 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19771 (armv7ve): Likewise.
19772 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19773 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19774 (armv8-a): Add nocrypto option.
19775 (armv8.1-a, armv8.2-a): Likewise.
19776 (armv8-m.main): add options fp, fp.dp and nofp.
19778 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19780 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19782 (armv8-a+crc): Delete.
19783 (armv8.1-a): Add options simd, crypto and nofp.
19784 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19785 (armv8.2-a+fp16): Delete.
19786 (armv8-m.main): Add option dsp.
19787 (armv8-m.main+dsp): Delete.
19788 (cortex-a8): Add fpu. Add nofp option.
19789 (cortex-a9): Add fpu. Add nofp and nosimd options.
19790 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19791 link to main cpu and architecture data structures.
19792 (gen_comm_data): Only put isa attributes from the main architecture
19794 (option): New statement for architecture and CPU entries.
19795 * arm.c (struct cpu_option): New structure.
19796 (struct processors): Add entry for options.
19797 (arm_unrecognized_feature): New function.
19798 (arm_parse_arch_cpu_name): Ignore any characters after the first
19800 (arm_parse_arch_cpu_feature): New function.
19801 (arm_configure_build_target): Separate out any CPU and architecture
19802 features and parse separately. Don't error out if -mfpu=auto is
19803 used with only an architecture string.
19804 (arm_print_asm_arch_directives): New function.
19805 (arm_file_start): Call it.
19806 * config/arm/arm-cpu-cdata.h: Regenerated.
19807 * config/arm/arm-cpu-data.h: Likewise.
19808 * config/arm/arm-tables.opt: Likewise.
19810 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19812 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19813 assembler when it is not -mfpu=auto.
19815 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19817 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19818 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19819 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19820 (ASM_CPU_SPEC): Rewrite.
19821 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19822 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19823 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19824 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19825 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19826 copied string is NUL-terminated. Also strip any characters prefixed
19828 (arm_rewrite_selected_arch): New function.
19829 (arm_rewrite_march): New function.
19831 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19833 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19834 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19835 (march, mcpu, mtune): Convert to string-based options.
19836 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19837 (arm_parse_arch_cpu_name): New function.
19838 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19839 identify selected architecture or CPU.
19840 (arm_option_save): New function.
19841 (TARGET_OPTION_SAVE): Redefine.
19842 (arm_option_restore): Restore string options.
19843 (arm_option_print): Print string options.
19845 2017-06-16 Martin Sebor <msebor@redhat.com>
19847 PR tree-optimization/80933
19848 PR tree-optimization/80934
19849 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19850 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19851 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19852 (gimple_fold_builtin): Call them.
19854 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19856 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19857 as unlikely; update profile.
19859 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19861 * predict.c (force_edge_cold): Handle declaring edges impossible
19864 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19866 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19868 (try_unroll_loop_completely): Fix reporting.
19870 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19872 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19874 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19877 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19878 if given a non-constant argument for an intrinsic which requires a
19881 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19883 * profile.c (compare_freqs): New function.
19884 (branch_prob): Sort edge list.
19885 (find_spanning_tree): Assume that the list is priority sorted.
19887 2017-06-16 Richard Biener <rguenther@suse.de>
19889 PR tree-optimization/81090
19890 * passes.def (pass_record_bounds): Remove.
19891 * tree-pass.h (make_pass_record_bounds): Likewise.
19892 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19893 make_pass_record_bounds): Likewise.
19894 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19895 not free niter estimates at the beginning but at the end.
19896 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19898 2017-06-16 Richard Biener <rguenther@suse.de>
19900 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19901 initializer to workaround ICE in host GCC 4.8.
19903 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19905 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19907 (clone_inlined_nodes): Update.
19909 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19911 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19912 prefetch settings, and enable prefetching by default at -O3.
19914 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19916 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19917 Set flag_prefetch_loop_arrays according to tuning data.
19919 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19921 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19922 New tune structure.
19923 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19924 [Unrelated to main purpose of the patch] Place the pointer field last
19925 to enable type checking errors when tune structure are wrongly merged.
19926 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19927 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19928 (thunderx2t99_prefetch_tune): New tune constants.
19929 (tune_params *_tunings): Update all tunings (no functional change).
19930 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19931 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19932 from tunings structures.
19934 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19937 * ubsan.c (instrument_null): Add T argument, use it instead
19938 of computing it based on IS_LHS.
19939 (instrument_object_size): Likewise.
19940 (pass_ubsan::execute): Adjust instrument_null and
19941 instrument_object_size callers to pass gimple_get_lhs or
19942 gimple_assign_rhs1 result to it. Use instrument_null instead of
19943 calling get_base_address and instrument_mem_ref. Handle
19944 aggregate call arguments for object-size sanitization.
19946 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19948 PR tree-optimization/81089
19949 * tree-vrp.c (is_masked_range_test): Validate operands of
19952 2017-06-15 Martin Sebor <msebor@redhat.com>
19955 * dumpfile.c (dump_register): Avoid calling memset to initialize
19956 a class with a default ctor.
19957 * gcc.c (struct compiler): Remove const qualification.
19958 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19959 * hash-table.h: Ditto.
19960 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19962 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19963 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19965 * params.h (struct param_info): Make struct members non-const.
19966 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
19967 with default initialization.
19968 * vec.h (vec_copy_construct, vec_default_construct): New helper
19970 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
19971 with vec_copy_construct.
19972 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
19973 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
19974 * doc/invoke.texi (-Wclass-memaccess): Document.
19976 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19978 * emit-rtl.h (is_leaf): Update comment about local
19979 register allocator.
19981 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19984 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
19985 for a variable to have a section before checking if the section has a
19987 Set section to.persistent if persistent attribute is set.
19988 Warn if .persistent attribute is used on an automatic variable.
19990 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
19992 PR rtl-optimization/80474
19993 * reorg.c (update_block): Do not ignore instructions in a delay slot.
19995 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
19997 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20000 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20002 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20003 (casesi): Emit bounds checking as RTL.
20004 (casesi_internal_mips16_<mode>): Remove bounds checking.
20006 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20008 * config/xtensa/xtensa.c (xtensa_option_override): Append
20009 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20010 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20011 xtensa_doloop_hooks): Define unconditionally.
20012 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20014 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20015 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20016 for it in xtensa_option_override.
20017 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20018 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20020 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20022 * doc/cppopts.texi: Document '-' special value to -MF.
20024 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20026 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20027 (cortex_a53_fconst): Likewise.
20028 (cortex_a53_fpmul): Likewise.
20029 (cortex_a53_f_load_64): Likewise.
20030 (cortex_a53_f_load_many): Likewise.
20031 (cortex_a53_advsimd_alu): Likewise.
20032 (cortex_a53_advsimd_alu_q): Likewise.
20033 (cortex_a53_advsimd_mul): Likewise.
20034 (cortex_a53_advsimd_mul_q): Likewise.
20035 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20036 Add missing fmul, r2f_cvt and fconst cases.
20038 2017-06-14 Richard Biener <rguenther@suse.de>
20040 PR middle-end/81088
20041 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20043 (fold_binary_loc): When associating do not treat pre-existing
20044 TREE_OVERFLOW on literal constants as a reason to allow
20045 TREE_OVERFLOW on associated literal constants.
20047 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20049 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20050 (MASK_FEATURES): New macro.
20051 * config/sparc/sparc.c (sparc_option_override): Remove the special
20052 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20054 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20056 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20057 a division of 0 if non-call exceptions are enabled.
20059 2017-06-14 Andrew Pinski <apinski@cavium.com>
20060 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20063 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20064 Improve vector initialization code gen for only variable case.
20066 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20068 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20070 2017-06-14 Richard Biener <rguenther@suse.de>
20072 PR tree-optimization/81083
20073 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20076 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20078 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20079 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20080 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20081 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20082 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20083 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20085 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20087 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20088 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20090 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20092 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20094 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20096 * config/rs6000/t-rtems: Don't handle SPE.
20098 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20100 * config/rs6000/t-linux: Don't handle SPE.
20102 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20104 * config/rs6000/eabispe.h: Delete file.
20106 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20108 * config/rs6000/t-spe: Delete file.
20110 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20112 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20113 (rs6000_legitimate_offset_address_p): Return false for anything in
20114 V2SImode or V2SFmode.
20116 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20118 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20119 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20120 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20122 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20123 (rs6000_legitimate_offset_address_p): Ditto.
20124 (rs6000_emit_move): Ditto.
20125 (rs6000_init_builtins): Remove V4HI_type_node.
20127 2017-06-13 Martin Liska <mliska@suse.cz>
20130 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20131 (gate_asan): Likewise.
20132 * asan.h (asan_no_sanitize_address_p): Remove the function.
20133 (sanitize_flags_p): New function.
20134 * builtins.def: Fix coding style.
20135 * common.opt: Use renamed enum value.
20136 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20137 * doc/extend.texi: Document no_sanitize attribute.
20138 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20139 to SANITIZE_UNDEFINED_NONDEFAULT.
20140 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20141 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20142 Use sanitize_flags_p.
20143 * gimplify.c (gimplify_function_tree): Likewise.
20144 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20145 * opts.c (parse_no_sanitize_attribute): New function.
20146 (common_handle_option): Use renamed enum value.
20147 * opts.h (parse_no_sanitize_attribute): Declare.
20148 * tree.c (sanitize_flags_p): New function.
20149 * tree.h: Declared here.
20150 * tsan.c: Use sanitize_flags_p.
20151 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20152 (instrument_mem_ref): Likewise.
20153 (instrument_bool_enum_load): Likewise.
20154 (do_ubsan_in_current_function): Remove the function.
20155 (pass_ubsan::execute): Use sanitize_flags_p.
20156 * ubsan.h: Remove do_ubsan_in_current_function
20157 * tree-cfg.c (print_no_sanitize_attr_value): New function.
20158 (dump_function_to_file): Use it here.
20160 2017-06-13 Martin Jambor <mjambor@suse.cz>
20162 PR tree-optimization/80803
20163 PR tree-optimization/81063
20164 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
20165 (propagate_subaccesses_across_link): Enqueue subtree whenever
20166 necessary instead of relying on the caller.
20168 2017-06-13 Martin Jambor <mjambor@suse.cz>
20170 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
20171 that have a first_link.
20172 (sort_and_splice_var_accesses): Do not check first_link before
20174 (subtree_mark_written_and_enqueue): Likewise.
20175 (propagate_all_subaccesses): Likewise and do not stop at first
20176 parent with a first_link.
20178 2017-06-13 Martin Jambor <mjambor@suse.cz>
20180 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
20183 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20185 * match.pd: New pattern.
20187 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20189 * tree-vrp.c (is_masked_range_test): New function.
20190 (register_edge_assert_for): Determine ranges for
20193 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20195 PR tree-optimization/67328
20196 * fold-const.c (maskable_range_p): New function.
20197 (build_range_check): Generate bittests if possible.
20199 2017-06-13 Martin Liska <mliska@suse.cz>
20201 * gimple-pretty-print.c (dump_probability): Add new argument.
20202 (dump_edge_probability): Dump both probability and count.
20203 (dump_gimple_label): Likewise.
20204 (dump_gimple_bb_header): Likewise.
20206 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
20209 * config/avr/avr-devices.c: Fix indentation.
20210 * config/avr/gen-avr-mmcu-specs.c: Dito.
20212 2017-06-13 Richard Biener <rguenther@suse.de>
20214 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
20215 instead get vector type from stmt_info.
20216 (vectorizable_reduction): Adjust. Remove dead code.
20218 2017-06-13 Richard Biener <rguenther@suse.de>
20220 PR middle-end/81065
20221 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20222 case of C * (x * C2 + C3).
20223 (fold_addr_of_array_ref_difference): Properly fold index difference.
20225 2017-06-12 David S. Miller <davem@davemloft.net>
20228 * config/sparc/sparc.md (return expander): Emit frame blockage if
20229 function uses alloca.
20231 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20233 * combine.c (make_field_assignment): Check len rather than the mode
20234 precision when calling force_to_mode.
20236 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20238 Support multilibs and devices that see flash in RAM address range.
20241 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20242 (avr_mcu_t) <flash_pm_offset>: New field.
20243 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20244 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20245 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20246 (AVR_TINY_PM_OFFSET): Remove macro.
20247 * config/avr/avr.opt (-mshort-calls): New option.
20248 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20249 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20250 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20251 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20252 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20253 instead of avr_arch->have_jmp_call.
20254 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20255 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20256 avr_arch->flash_pm_offset to define.
20257 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20258 new field flash_pm_offset. Add entry for avrxmega3.
20259 (avr_texinfo): Add entry for avrxmega3.
20260 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20261 attiny212, attiny214,
20262 attiny412, attiny414, attiny416, attiny417,
20263 attiny814, attiny816, attiny817,
20264 attiny1614, attiny1616, attiny1617,
20265 attiny3214, attiny3216, attiny3217.
20266 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20267 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20268 (avr_print_operand_address) [AVR_TINY]: Same.
20269 (avr_asm_init_sections) <readonly_data_section>: Only patch
20270 callback if avr_arch->flash_pm_offset = 0.
20271 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20272 for rodata if avr_arch->flash_pm_offset != 0.
20273 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20274 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20275 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20276 (m_options): Append opt_rcall.
20277 (m_dirnames): Append dir_rcall.
20278 * config/avr/t-multilib: Regenerate.
20280 * configure.ac [target=avr]: Check whether avrxmega3 default
20281 linker description file works as needed.
20282 * configure: Regenerate.
20283 * doc/avr-mmcu.texi: Regenerate.
20284 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20285 <__AVR_ARCH__>: Document avrxmega3 and 103.
20286 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20287 <__AVR_SHORT_CALLS__>: Document it.
20288 <__AVR_PM_BASE_ADDRESS__>: Document it.
20289 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20290 (AVR Variable Attributes) <progmem>: Document this is
20291 not needed for avrxmega3.
20292 (AVR Named Address Spaces) <__flash>: Dito.
20294 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20296 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20298 2017-06-12 Doug Rupp <rupp@adacore.com>
20300 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20301 Append vxworks-stdint.h to the tm_file list.
20302 * config/vxworks-stdint.h: New file.
20304 2017-06-12 Martin Liska <mliska@suse.cz>
20306 PR tree-optimization/81041
20307 * tree-profile.c (gimple_gen_ic_func_profiler):
20308 Create an extra BB in profile-generate
20309 (gimple_gen_time_profiler): Likewise.
20311 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20313 PR tree-optimization/81003
20314 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20315 (update_range_test): Use it instead of force_gimple_operand_gsi.
20317 2017-06-12 Richard Biener <rguenther@suse.de>
20319 PR tree-optimization/81053
20320 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20321 with backedge value not defined in loop. Simplify def stmt
20324 2017-06-11 Tom de Vries <tom@codesourcery.com>
20327 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20329 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20330 nvptx_cannot_force_const_mem.
20332 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20334 * opts.c (finish_options): Move test for flag_split_stack after
20335 it has been initialized.
20337 2017-06-11 Jason Merrill <jason@redhat.com>
20339 * tree.h (id_equal): New.
20340 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20341 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20342 instead of strcmp of IDENTIFIER_POINTER.
20344 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20346 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20347 (mark_all_inlined_calls_cdtor): Fix formating.
20348 (inline_transform): Rescale profile before inlining.
20350 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20352 * cgraph.h (cgraph_edge::clone): Update prototype.
20353 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20354 (cgraph_node::create_clone): Update.
20355 (cgraph_node::create_version_clone): Update.
20356 * tree-inline.c (copy_bb): Update.
20357 (expand_call_inline): Update.
20359 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20361 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20362 factored out from ...
20363 (rs6000_emit_prologue): ... here.
20365 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20367 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20368 factored out from ...
20369 (rs6000_emit_prologue): ... here.
20371 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20373 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20375 (handle_missing_profiles): Fix computation of tp_first_run.
20376 (counts_to_freqs): Do not touch freqs when count is 0.
20378 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20380 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20383 2017-06-10 Tom de Vries <tom@codesourcery.com>
20385 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20386 attributes): Document signal effective target.
20388 2017-06-10 Tom de Vries <tom@codesourcery.com>
20390 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20391 Document effective target stack_size.
20393 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20395 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20396 to the edit_context if they can be auto-applied.
20398 2017-06-9 Ian Lance Taylor <iant@golang.org>
20400 * opts.c (finish_options): If -fsplit-stack, disable implicit
20401 -forder-blocks-and-partition.
20402 * doc/invoke.texi (Optimize Options): Document that when using
20403 -fsplit-stack -forder-blocks-and-partition is not implicitly
20406 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20408 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20409 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20410 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20411 * builtins.def (abort, trap, unreachable): Declare cold.
20412 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20413 * tree-core.h (ECF_COLD): New.
20414 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20415 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20417 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20419 * predict.c (unlikely_executed_stmt_p): Cleanup.
20421 2017-06-09 Richard Biener <rguenther@suse.de>
20423 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20424 model if the ref is always written to.
20426 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20428 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20430 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20432 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20435 2017-06-09 Tom de Vries <tom@codesourcery.com>
20438 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20439 "target cannot support label values" when encountering LABEL_REF.
20441 2017-06-09 Martin Liska <mliska@suse.cz>
20443 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20444 (gimple_gen_ic_func_profiler): Emit direct comparison
20445 of __gcov_indirect_call_callee with NULL.
20446 (gimple_gen_time_profiler): Change probability from
20447 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20449 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20451 * profile.c (edge_gcov_counts): Turn to pointer.
20452 (compute_branch_probabilities, compute_branch_probabilities): Update.
20453 (branch_prob): Do not clear edge_gcov_count.
20454 * profile.h (edge_gcov_counts): Turn to pointer.
20455 (edge_gcov_count): Update.
20457 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20459 * gimple.h (gimple_check_failed): Mark cold.
20461 2017-06-09 Richard Biener <rguenther@suse.de>
20463 PR tree-optimization/66623
20464 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20465 refactor check_reduction into two parts, properly computing
20466 whether we have to check reduction validity for outer loop
20469 2017-06-09 Richard Biener <rguenther@suse.de>
20471 PR tree-optimization/79483
20472 * graphite-scop-detection.c (order): New global.
20473 (get_order): Compute bb to order mapping that satisfies code
20474 generation constraints.
20475 (cmp_pbbs): New helper.
20476 (build_scops): Start domwalk at entry block, sort generated
20479 2017-06-09 Richard Biener <rguenther@suse.de>
20481 PR middle-end/81007
20482 * ipa-polymorphic-call.c
20483 (ipa_polymorphic_call_context::restrict_to_inner_class):
20484 Skip FIELD_DECLs with error_mark_node type.
20485 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20488 2017-06-09 Martin Liska <mliska@suse.cz>
20490 * predict.c (struct branch_predictor): New struct.
20491 (test_prediction_value_range): New test.
20492 (predict_c_tests): New function.
20493 * selftest-run-tests.c (selftest::run_tests): Run the function.
20494 * selftest.h: Declare new tests.
20496 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20499 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20500 gen_add3_insn did not fail.
20501 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20502 r0, construct that number in a temporary reg and add that reg to r0.
20503 If asked to put the result in r0 as well, fail.
20505 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20507 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20508 for early expansion of vec_eqv.
20510 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20512 PR middle-end/81005
20513 * ubsan.c (instrument_null): Avoid pointless code temporary.
20514 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20516 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20520 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20523 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20526 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20528 * predict.c (unlikely_executed_edge_p): Move ahead.
20529 (probably_never_executed_edge_p): Use it.
20531 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20533 PR middle-end/79988
20534 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20535 gimple_call_builtin_p call.
20537 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20539 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20540 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20541 rtl_check_failed_type2, rtl_check_failed_code1,
20542 rtl_check_failed_code2, rtl_check_failed_code_mode,
20543 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20544 rtvec_check_failed_bounds, rtl_check_failed_flag,
20545 _fatal_insn_not_found, _fatal_insn): Likewise.
20546 * tree.h (tree_contains_struct_check_failed,
20547 tree_check_failed, tree_not_check_failed,
20548 tree_class_check_failed, tree_range_check_failed,
20549 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20550 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20551 tree_operand_check_failed, omp_clause_check_failed,
20552 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20555 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20557 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20558 flag_branch_probabilities.
20559 * ipa-inline.c (edge_badness): Likewise.
20560 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20561 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20562 * predict.c (maybe_hot_frequency_p): Likewise.
20563 (probably_never_executed): Likewise.
20564 * sched-ebb.c (schedule_ebbs): Likewise.
20565 * sched-rgn.c (find_single_block_region): Likewise.
20566 * tracer.c (tail_duplicate): Likewise.
20568 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20570 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20571 longer requires x_flag_profile_use.
20573 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20575 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20576 instead of flag_reorder_blocks_and_partition.
20577 * dbxout.c (dbxout_function_end): Likewise.
20578 * dwarf2out.c (gen_subprogram_die): Likewise.
20579 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20580 * hw-doloop.c (reorg_loops): Likewise.
20581 * varasm.c (assemble_start_function,
20582 assemble_end_function): Likewise.
20583 (decide_function_section): Do not check for
20584 flag_reorder_blocks_and_partition.
20586 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20588 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20590 (chkp_get_hard_register_fake_addr_expr): Ditto.
20591 (chkp_build_addr_expr): Add check for hard reg case.
20592 (chkp_parse_array_and_component_ref): Ditto.
20593 (chkp_find_bounds_1): Ditto.
20594 (chkp_process_stmt): Don't generate bounds store for
20597 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20599 * predict.c (maybe_hot_bb_p): Do not check profile status.
20600 (maybe_hot_edge_p): Likewise.
20601 (probably_never_executed): Check for zero counts even if profile
20603 (unlikely_executed_edge_p): New function.
20604 (unlikely_executed_stmt_p): New function.
20605 (unlikely_executed_bb_p): New function.
20606 (set_even_probabilities): Use unlikely predicates.
20607 (combine_predictions_for_bb): Likewise.
20608 (predict_paths_for_bb): Likewise.
20609 (predict_paths_leading_to_edge): Likewise.
20610 (determine_unlikely_bbs): New function.
20611 (estimate_bb_frequencies): Use it.
20612 (compute_function_frequency): Use zero counts even if profile is
20614 * profile-count.h: Fix typo.
20616 2017-08-08 Julia Koval <julia.koval@intel.com>
20618 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20619 _mm512_mask_cvtsepi16_storeu_epi8,
20620 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20621 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20622 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20623 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20624 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20625 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20626 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20627 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20628 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20629 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20630 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20631 __builtin_ia32_pmovuswb256mem_mask,
20632 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20633 __builtin_ia32_pmovwb512mem_mask): New builtins.
20635 2017-08-08 Julia Koval <julia.koval@intel.com>
20637 PR target/73350,80862
20638 * config/i386/subst.md (round): Fix round pattern.
20639 * config/i386/i386.c (ix86_erase_embedded_rounding):
20640 Fix erasing rounding for the fixed pattern.
20642 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20644 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20646 2017-06-08 Martin Liska <mliska@suse.cz>
20648 PR gcov-profile/80911
20649 * gcov.c (block_info::block_info): New constructor.
20651 2017-06-07 Carl Love <cel@us.ibm.com>
20653 * config/rs6000/rs6000-c: The return type of the following
20654 built-in functions was implemented as int not long long. Fix sign
20655 of return value for the unsigned version of vec_mulo and vec_mule.
20656 vector unsigned long long vec_bperm (vector unsigned long long,
20657 vector unsigned char)
20658 vector signed long long vec_mule (vector signed int,
20660 vector unsigned long long vec_mule (vector unsigned int,
20661 vector unsigned int)
20662 vector signed long long vec_mulo (vector signed int,
20664 vector unsigned long long vec_mulo (vector unsigned int,
20665 vector unsigned int)
20666 * doc/extend.texi: Fix the documentation for the built-in
20669 2017-06-07 Carl Love <cel@us.ibm.com>
20672 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20675 2017-06-07 Carl Love <cel@us.ibm.com>
20677 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20678 support, Generate doublehv for signed int/float for BE case only.
20680 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20682 * doc/invoke.texi (mcx16): Rewrite.
20684 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20686 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20687 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20688 *mov<mode>_softfloat, and an anonymous splitter): Use
20689 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20691 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20693 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20695 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20696 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20697 (rs6000_debug_reg_global): Adjust.
20698 (rs6000_init_hard_regno_mode_ok): Adjust.
20699 (rs6000_dbx_register_number): Adjust.
20700 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20701 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20702 Remove SPE_ACC and SPEFSCR.
20703 (REG_ALLOC_ORDER): Ditto.
20704 (FRAME_POINTER_REGNUM): Change to 111.
20705 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20706 (REG_CLASS_NAMES): Ditto.
20707 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20708 (REGISTER_NAMES): Ditto.
20709 (ADDITIONAL_REG_NAMES): Ditto.
20710 (rs6000_reg_names): Ditto.
20711 * config/rs6000/rs6000.md: Renumber some register number
20714 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20716 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20718 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20719 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20721 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20722 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20723 Delete the SPE high registers.
20724 (REG_ALLOC_ORDER): Ditto.
20725 (enum reg_class): Remove SPE_HIGH_REGS.
20726 (REG_CLASS_NAMES): Ditto.
20727 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20728 (REGISTER_NAMES): Ditto.
20729 (rs6000_reg_names): Ditto.
20730 * doc/tm.texi.in: Remove SPE as example.
20731 * doc/tm.texi: Regenerate.
20733 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20735 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20736 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20737 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20738 * config/rs6000/rs6000.md (type): Remove "brinc".
20740 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20742 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20743 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20744 * config/rs6000/linuxspe.h: Delete file.
20745 * config/rs6000/rs6000.md: Don't include spe.md.
20746 * config/rs6000/spe.h: Delete file.
20747 * config/rs6000/spe.md: Delete file.
20748 * config/rs6000/t-rs6000: Remove spe.md.
20750 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20752 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20753 (reg_or_none500mem_operand): Delete.
20754 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20755 instead of reg_or_none500mem_operand.
20757 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20759 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20760 handling of SPE flags.
20761 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20763 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20765 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20767 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20768 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20769 paired_divv2sf3): Similar.
20770 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20771 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20772 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20774 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20775 Rename the paired_* instruction patterns.
20776 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20778 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20779 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20780 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20781 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20782 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20783 PAIRED_VECTOR_MODE.
20784 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20785 (spe_func_has_64bit_regs_p): Delete.
20786 (spe_expand_predicate_builtin): Delete.
20787 (spe_expand_evsel_builtin): Delete.
20788 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20789 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20790 (invalid_e500_subreg): Delete.
20791 (rs6000_legitimize_address): Always force_reg op2 as well, for
20792 paired single memory accesses.
20793 (rs6000_member_type_forces_blk): Delete.
20794 (rs6000_spe_function_arg): Delete.
20795 (rs6000_expand_unop_builtin): Delete SPE handling.
20796 (rs6000_expand_binop_builtin): Ditto.
20797 (spe_expand_stv_builtin): Delete.
20798 (bdesc_2arg_spe): Delete.
20799 (spe_expand_builtin): Delete.
20800 (spe_expand_predicate_builtin): Delete.
20801 (spe_expand_evsel_builtin): Delete.
20802 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20803 (spe_init_builtins): Delete.
20804 (spe_func_has_64bit_regs_p): Delete.
20805 (savres_routine_name): Delete "info" parameter. Adjust callers.
20806 (rs6000_emit_stack_reset): Ditto.
20807 (rs6000_dwarf_register_span): Delete.
20808 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20809 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20810 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20812 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20814 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20815 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20816 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20817 mulv2sf3, divv2sf3): Delete expanders.
20819 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20821 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20823 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20825 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20826 * config/rs6000/rs6000.c: Ditto.
20828 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20830 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20831 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20832 comparison_operator.
20834 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20836 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20837 * config/rs6000/rs6000.opt: Ditto.
20838 * config/rs6000/t-rtems: Ditto.
20840 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20842 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20843 TARGET_E500_SINGLE by 0, simplify.
20844 * config/rs6000/rs6000.c: Ditto.
20845 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20846 (spe_build_register_parallel): Delete.
20847 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20848 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20849 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20850 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20851 (E500_CONVERT): Delete.
20852 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20854 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20856 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20857 * config/rs6000/dfp.md: Ditto.
20858 (negdd2, *negdd2_fpr): Merge.
20859 (absdd2, *absdd2_fpr): Merge.
20860 (negtd2, *negtd2_fpr): Merge.
20861 (abstd2, *abstd2_fpr): Merge.
20862 * config/rs6000/e500.h: Delete file.
20863 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20864 TARGET_FPRS by 1 and simplify.
20865 * config/rs6000/rs6000-c.c: Ditto.
20866 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20867 TARGET_DF_SPE by 0.
20868 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20870 * config/rs6000/rs6000.md: Ditto.
20871 (floatdidf2, *floatdidf2_fpr): Merge.
20872 (move_from_CR_gt_bit): Delete.
20873 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20874 (E500_CR_IOR_COMPARE): Delete.
20875 (All patterns that require !TARGET_FPRS): Delete.
20876 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20878 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20880 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20882 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20884 * graphds.c (add_edge): Intitialize edge's attached data.
20885 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20886 pointer parameter. Call pointed function on each edge during
20887 graph traversing. Skip traversing the edge when the function
20889 (graphds_dfs, graphds_scc): Ditto.
20890 (for_each_edge): New parameter. Pass the new parameter to callback
20892 * graphds.h (skip_edge_callback): New function pointer type.
20893 (graphds_dfs, graphds_scc): New function pointer parameter.
20894 (graphds_edge_callback, for_each_edge): New parameter.
20896 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20898 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20899 out code checking if runtime alias check is possible to below ...
20900 Call the new function.
20901 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20902 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20904 2017-06-07 Marek Polacek <polacek@redhat.com>
20907 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20908 TYPE_OVERFLOW_WRAPS check.
20910 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20912 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20913 if versioning is required.
20914 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20915 peeling with the check for versioning.
20917 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20919 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20920 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20921 Set true to new parameter if new ssa variable is defined.
20922 (vect_gen_vector_loop_niters): Refactor. Set range information
20923 for the new vector loop bound variable.
20924 (vect_do_peeling): Ditto.
20926 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20928 * tree-affine.c (ssa.h): Include header file.
20929 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20930 has wrapping overflow behavior.
20932 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20934 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20936 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20938 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20939 (tree_to_aff_combination): ... here.
20941 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20943 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20944 reg_pressure model function.
20945 (ivopts_global_cost_for_size): Delete.
20946 (determine_set_costs, iv_ca_recount_cost): Call new model function
20947 ivopts_estimate_reg_pressure.
20949 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20951 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20952 expensive than udiv. Remove floating point cases from mod.
20954 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20956 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20957 Increase idiv cost.
20959 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20961 * config/aarch64/aarch64.md
20962 (copysignsf3): Fix mask generation.
20964 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20966 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
20968 (class dump_manager): Add register_dumps method.
20969 * dumpfile.c: Include langhooks.h.
20970 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
20971 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
20972 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
20973 (dump_manager::dump_register): Start with 512 entries instead of 32.
20974 (dump_manager::register_dumps): New method.
20975 * toplev.c (general_init): Instead of invoking register_dumps
20976 langhook, invoke register_dumps method on the dump manager.
20977 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
20980 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
20982 * doc/md.texi: Clarify the restrictions on a define_insn condition.
20983 Say that # requires an associated define_split to exist, and that
20984 the define_split must be suitable for use after register allocation.
20986 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20988 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
20989 (compute_outgoing_frequencies): Also initialize zero counts.
20990 (find_many_sub_basic_blocks): Do not produce uninitialized profile
20991 around loops; preserve more of profile when nothing changes.
20993 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
20995 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
20997 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
20998 * config/arm/arm-cpu-cdata.h: Regenerate.
20999 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21000 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21001 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21002 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21003 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21005 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21007 * config/arm/t-rmprofile: Likewise.
21008 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21010 2017-06-06 David S. Miller <davem@davemloft.net>
21013 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21014 blockage if function uses alloca.
21016 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21018 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21019 New "uid" fields to hold pretty-print IDs of group and ref.
21020 Memory references are now identified as <group_id>:<ref_id>
21021 instead of using [random] addresses.
21022 (dump_mem_details): Simplify, no functional change.
21023 (dump_mem_ref): Simplify and make output more concise.
21024 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21025 (find_or_create_group): Initialize group uid.
21026 (record_ref): Initialize ref uid. Improve debug output.
21027 (prune_group_by_reuse, should_issue_prefetch_p,)
21028 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21029 (mark_nontemporal_store, determine_loop_nest_reuse):
21030 Improve debug output.
21032 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21034 * dbgcnt.def (prefetch): New debug counter.
21035 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21036 (schedule_prefetches): Stop issueing prefetches if debug counter
21039 2017-06-06 Tom de Vries <tom@codesourcery.com>
21041 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21042 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21044 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21046 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21047 Use aarch64_reg_or_zero predicate for operand 4.
21048 (aarch64_compare_and_swap<mode> define_insn_and_split):
21049 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21050 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21052 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21054 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21055 (arm_compute_save_core_reg_mask): This.
21056 (thumb1_compute_save_reg_mask): Rename into ...
21057 (thumb1_compute_save_core_reg_mask): This.
21058 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21059 (arm_compute_frame_layout): Likewise.
21061 2017-06-06 Richard Biener <rguenther@suse.de>
21063 PR tree-optimization/80974
21064 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21065 keep or clear leaders SSA info.
21067 2017-06-06 Tom de Vries <tom@codesourcery.com>
21069 * config/nvptx/nvptx.c (split_mode_p): New function.
21070 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21072 2017-06-06 Tom de Vries <tom@codesourcery.com>
21074 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21076 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21079 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21082 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21084 * shrink-wrap.c (handle_simple_exit): Update profile.
21085 (try_shrink_wrapping): Upate profile.
21087 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21089 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21090 (tree_guess_outgoing_edge_probabilities): New.
21091 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21092 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21094 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21096 * ipa-split.c (split_function): Initialize return bb profile.
21098 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21100 * profile.c (compute_branch_probabilities): Also initialize
21101 EXIT_BLOCK profile.
21103 2017-06-06 Richard Biener <rguenther@suse.de>
21105 PR tree-optimization/80928
21106 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21107 (vect_analyze_loop_operations): Properly guard analysis for
21109 (vect_transform_loop): Likewise.
21110 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21111 (vect_model_induction_cost): Do not cost for pure SLP.
21112 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21113 of induction in inner loop vectorization.
21114 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21115 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21116 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21118 (vect_analyze_slp_cost_1): Cost induction.
21119 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21120 (vect_get_slp_vect_defs): Likewise.
21121 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21122 (vect_transform_stmt): Handle SLP reductions.
21123 * tree-vectorizer.h (vectorizable_induction): Adjust.
21125 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21127 * config/rs6000/rs6000.c (make_resolver_func): Update
21128 init_lowered_empty_function call.
21130 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21132 * doc/invoke.texi: Document the -fprofile-abs-path option.
21133 * common.opt (fprofile-abs-path): New option.
21134 * gcov-io.h (gcov_write_filename): Declare.
21135 * gcov-io.c (gcov_write_filename): New function.
21136 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21137 * profile.c (output_location): Likewise.
21139 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21141 * shring-wrap.c: Revert accidental commit.
21143 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21145 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21147 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21149 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21151 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21152 profile in callgraph edge.
21153 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21154 (apply_scale): Likewise.
21155 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
21156 Also scale profile when inlining function with zero profile.
21157 (initialize_cfun): Update exit block profile even when it is zero.
21158 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
21159 when profile is read.
21161 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21163 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
21164 (CLONE_*): New constants to define the processors we can generate
21165 code for with the target_clone attribute.
21166 (rs6000_clone_map): New array to identify which clone processors
21167 the current program is running on.
21168 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
21169 target_clone attribute.
21170 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
21171 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
21172 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
21173 (cpu_expand_builtin): Add support for target_clone attribute.
21174 (rs6000_valid_attribute_p): Allow "default" attribute.
21175 (get_decl_name): New debug function to simplify printing the
21176 current function name in debugging statements.
21177 (rs6000_clone_priority): New functions to support the target_clone
21178 attribute, and be able to generate code to switch between ISA 2.05
21179 through ISA 3.0 (power6 through power9).
21180 (rs6000_compare_version_priority): Likewise.
21181 (rs6000_get_function_versions_dispatcher): Likewise.
21182 (make_resolver_func): Likewise.
21183 (add_condition_to_bb): Likewise.
21184 (dispatch_function_versions): Likewise.
21185 (rs6000_generate_version_dispatcher_body): Likewise.
21186 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
21187 (fusion_gpr_load_p): Fix a spacing issue.
21188 * doc/extend.texi (Common Function Attributes): Document that the
21189 PowerPC supports the target_clone attribute.
21191 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21193 * config/arm/arm.h: explain F symbol found in description of ARM
21194 register allocation in its legend.
21196 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21198 * config/mips/frame-header-opt.c: Include profile-count.h.
21199 * config/riscv/riscv.c: Include profile-count.h
21201 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21203 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
21205 (sm_set_flag_if_changed): Add bbs field.
21206 (execute_sm_if_changed_flag_set): Pass BBS.
21207 (execute_sm): Update.
21209 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21211 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
21214 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21216 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
21217 (peephole2): New peephole2 to emit the above.
21218 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21220 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21222 * config/aarch64/aarch64.c (define_peephole2 above
21223 *sub_<shift>_<mode>): New peephole.
21225 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21227 * config/i386/i386.c (make_resolver_func): Update.
21228 * Makefile.in: Add profile-count.h and profile-count.o
21229 * auto-profile.c (afdo_indirect_call): Update to new API.
21230 (afdo_set_bb_count): Update.
21231 (afdo_propagate_edge): Update.
21232 (afdo_propagate_circuit): Update.
21233 (afdo_calculate_branch_prob): Update.
21234 (afdo_annotate_cfg): Update.
21235 * basic-block.h: Include profile-count.h
21236 (struct edge_def): Turn count to profile_count.
21237 (struct basic_block_def): Likewie.
21238 (REG_BR_PROB_BASE): Move to profile-count.h
21239 (RDIV): Move to profile-count.h
21240 * bb-reorder.c (max_entry_count): Turn to profile_count.
21241 (find_traces): Update.
21242 (rotate_loop):Update.
21243 (connect_traces):Update.
21244 (sanitize_hot_paths):Update.
21245 * bt-load.c (migrate_btr_defs): Update.
21246 * cfg.c (RDIV): Remove.
21247 (init_flow): Use alloc_block.
21248 (alloc_block): Uninitialize count.
21249 (unchecked_make_edge): Uninitialize count.
21250 (check_bb_profile): Update.
21251 (dump_edge_info): Update.
21252 (dump_bb_info): Update.
21253 (update_bb_profile_for_threading): Update.
21254 (scale_bbs_frequencies_int): Update.
21255 (scale_bbs_frequencies_gcov_type): Update.
21256 (scale_bbs_frequencies_profile_count): New.
21257 * cfg.h (update_bb_profile_for_threading): Update.
21258 (scale_bbs_frequencies_profile_count): Declare.
21259 * cfgbuild.c (compute_outgoing_frequencies): Update.
21260 (find_many_sub_basic_blocks): Update.
21261 * cfgcleanup.c (try_forward_edges): Update.
21262 (try_crossjump_to_edge): Update.
21263 * cfgexpand.c (expand_gimple_tailcall): Update.
21264 (construct_exit_block): Update.
21265 * cfghooks.c (verify_flow_info): Update.
21266 (dump_bb_for_graph): Update.
21267 (split_edge): Update.
21268 (make_forwarder_block): Update.
21269 (duplicate_block): Update.
21270 (account_profile_record): Update.
21271 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21272 (get_estimated_loop_iterations): Update.
21273 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21274 (single_likely_exit): Update.
21275 * cfgloopmanip.c (scale_loop_profile): Update.
21277 (set_zero_probability): Update.
21278 (lv_adjust_loop_entry_edge): Update.
21279 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21280 (purge_dead_edges): Update.
21281 (rtl_account_profile_record): Update.
21282 * cgraph.c (cgraph_node::create): Uninitialize count.
21283 (symbol_table::create_edge): Uninitialize count.
21284 (cgraph_update_edges_for_call_stmt_node): Update.
21285 (cgraph_edge::dump_edge_flags): Update.
21286 (cgraph_node::dump): Update.
21287 (cgraph_edge::maybe_hot_p): Update.
21288 * cgraph.h: Include profile-count.h
21289 (create_clone), create_edge, create_indirect_edge): Update.
21290 (cgraph_node): Turn count to profile_count.
21291 (cgraph_edge0: Likewise.
21292 (make_speculative, clone): Update.
21293 (create_edge): Update.
21294 (init_lowered_empty_function): Update.
21295 * cgraphclones.c (cgraph_edge::clone): Update.
21296 (duplicate_thunk_for_node): Update.
21297 (cgraph_node::create_clone): Update.
21298 * cgraphunit.c (cgraph_node::analyze): Update.
21299 (cgraph_node::expand_thunk): Update.
21300 * final.c (dump_basic_block_info): Update.
21301 * gimple-streamer-in.c (input_bb): Update.
21302 * gimple-streamer-out.c (output_bb): Update.
21303 * graphite.c (print_global_statistics): Update.
21304 (print_graphite_scop_statistics): Update.
21305 * hsa-brig.c: Include basic-block.h.
21306 * hsa-dump.c: Include basic-block.h.
21307 * hsa-gen.c (T sum_slice): Update.
21308 (convert_switch_statements):Update.
21309 * hsa-regalloc.c: Include basic-block.h.
21310 * ipa-chkp.c (chkp_produce_thunks): Update.
21311 * ipa-cp.c (struct caller_statistics): Update.
21312 (init_caller_stats): Update.
21313 (gather_caller_stats): Update.
21314 (ipcp_cloning_candidate_p): Update.
21315 (good_cloning_opportunity_p): Update.
21316 (get_info_about_necessary_edges): Update.
21317 (dump_profile_updates): Update.
21318 (update_profiling_info): Update.
21319 (update_specialized_profile): Update.
21320 (perhaps_add_new_callers): Update.
21321 (decide_about_value): Update.
21322 (ipa_cp_c_finalize): Update.
21323 * ipa-devirt.c (struct odr_type_warn_count): Update.
21324 (struct decl_warn_count): Update.
21325 (struct final_warning_record): Update.
21326 (possible_polymorphic_call_targets): Update.
21327 (ipa_devirt): Update.
21328 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21329 * ipa-icf.c (sem_function::merge): Update.
21330 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21331 * ipa-inline.c (compute_uninlined_call_time): Update.
21332 (compute_inlined_call_time): Update.
21333 (want_inline_small_function_p): Update.
21334 (want_inline_self_recursive_call_p): Update.
21335 (edge_badness): Update.
21336 (lookup_recursive_calls): Update.
21337 (recursive_inlining): Update.
21338 (inline_small_functions): Update.
21339 (dump_overall_stats): Update.
21340 (dump_inline_stats): Update.
21341 * ipa-profile.c (ipa_profile_generate_summary): Update.
21342 (ipa_propagate_frequency): Update.
21343 (ipa_profile): Update.
21344 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21345 * ipa-utils.c (ipa_merge_profiles): Update.
21346 * loop-doloop.c (doloop_modify): Update.
21347 * loop-unroll.c (report_unroll): Update.
21348 (unroll_loop_runtime_iterations): Update.
21349 * lto-cgraph.c (lto_output_edge): Update.
21350 (lto_output_node): Update.
21351 (input_node): Update.
21352 (input_edge): Update.
21353 (merge_profile_summaries): Update.
21354 * lto-streamer-in.c (input_cfg): Update.
21355 * lto-streamer-out.c (output_cfg): Update.
21356 * mcf.c (create_fixup_graph): Update.
21357 (adjust_cfg_counts): Update.
21358 (sum_edge_counts): Update.
21359 * modulo-sched.c (sms_schedule): Update.
21360 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21361 * predict.c (maybe_hot_count_p): Update.
21362 (probably_never_executed): Update.
21363 (dump_prediction): Update.
21364 (combine_predictions_for_bb): Update.
21365 (propagate_freq): Update.
21366 (handle_missing_profiles): Update.
21367 (counts_to_freqs): Update.
21368 (rebuild_frequencies): Update.
21369 (force_edge_cold): Update.
21370 * predict.h: Include profile-count.h
21371 (maybe_hot_count_p, counts_to_freqs): UPdate.
21372 * print-rtl-function.c: Do not include cfg.h
21373 * print-rtl.c: Include basic-block.h
21374 * profile-count.c: New file.
21375 * profile-count.h: New file.
21376 * profile.c (is_edge_inconsistent): Update.
21377 (correct_negative_edge_counts): Update.
21378 (is_inconsistent): Update.
21379 (set_bb_counts): Update.
21380 (read_profile_edge_counts): Update.
21381 (compute_frequency_overlap): Update.
21382 (compute_branch_probabilities): Update; Initialize and deinitialize
21384 (branch_prob): Update.
21385 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21386 (edge_gcov_count): New.
21387 (bb_gcov_count): New.
21388 * shrink-wrap.c (try_shrink_wrapping): Update.
21389 * tracer.c (better_p): Update.
21390 * trans-mem.c (expand_transaction): Update.
21391 (ipa_tm_insert_irr_call): Update.
21392 (ipa_tm_insert_gettmclone_call): Update.
21393 * tree-call-cdce.c: Update.
21394 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21395 (gimple_duplicate_sese_tail): Update.
21396 (gimple_account_profile_record): Update.
21397 (execute_fixup_cfg): Update.
21398 * tree-inline.c (copy_bb): Update.
21399 (copy_edges_for_bb): Update.
21400 (initialize_cfun): Update.
21401 (freqs_to_counts): Update.
21402 (copy_cfg_body): Update.
21403 (expand_call_inline): Update.
21404 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21405 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21406 (try_unroll_loop_completely): Update.
21407 (try_peel_loop): Update.
21408 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21409 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21410 * tree-ssa-loop-split.c (connect_loops): Update.
21411 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21412 * tree-ssa-reassoc.c (branch_fixup): Update.
21413 * tree-ssa-tail-merge.c (replace_block_by): Update.
21414 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21415 (compute_path_counts): Update.
21416 (update_profile): Update.
21417 (recompute_probabilities): Update.
21418 (update_joiner_offpath_counts): Update.
21419 (estimated_freqs_path): Update.
21420 (freqs_to_counts_path): Update.
21421 (clear_counts_path): Update.
21422 (ssa_fix_duplicate_block_edges): Update.
21423 (duplicate_thread_path): Update.
21424 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21425 (struct switch_conv_info): Update.
21426 * tree-tailcall.c (decrease_profile): Update.
21427 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21428 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21429 * value-prof.c (check_counter): Update.
21430 (gimple_divmod_fixed_value): Update.
21431 (gimple_mod_pow2): Update.
21432 (gimple_mod_subtract): Update.
21433 (gimple_ic_transform): Update.
21434 (gimple_stringop_fixed_value): Update.
21435 * value-prof.h (gimple_ic): Update.
21437 2017-06-02 Carl Love <cel@us.ibm.com>
21439 * config/rs6000/rs6000-c: Add support for built-in functions
21440 vector double vec_doublee (vector signed int);
21441 vector double vec_doublee (vector unsigned int);
21442 vector double vec_doublee (vector float);
21443 vector double vec_doubleh (vector signed int);
21444 vector double vec_doubleh (vector unsigned int);
21445 vector double vec_doubleh (vector float);
21446 vector double vec_doublel (vector signed int);
21447 vector double vec_doublel (vector unsigned int);
21448 vector double vec_doublel (vector float);
21449 vector double vec_doubleo (vector signed int);
21450 vector double vec_doubleo (vector unsigned int);
21451 vector double vec_doubleo (vector float);.
21452 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21453 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21455 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21456 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21457 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21459 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21460 vec_doublel, vec_doubleh.
21461 * doc/extend.texi: Update the built-in documentation file for the
21462 new built-in functions.
21464 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21467 * ipa-inline-analysis.c (free_growth_caches): Set
21468 edge_removal_hook_holder to NULL after removing it.
21470 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21472 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21473 comparision with zero.
21475 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21476 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21477 for early expansion of vec_min and vec_max builtins.
21478 (builtin_function_type): Add min/max unsigned variants to those
21479 identified as having unsigned arguments.
21481 2017-06-02 Olivier Hainque <hainque@adacore.com>
21483 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21485 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21487 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21488 Use VALL_F16 iterator rather than VALL.
21490 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21492 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21493 Emit CBNZ inside loop when doing a strong exchange and comparing
21494 against zero. Generate the CC flags after the loop.
21496 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21498 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21499 (dl_section_ref): New.
21500 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21501 On AIX, append an expression to subtract the size of the
21502 section length to dl_section_ref.
21504 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21506 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21507 for early expansion of vector absolute builtins.
21509 2017-06-02 Richard Biener <rguenther@suse.de>
21511 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21512 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21514 2017-06-02 Richard Biener <rguenther@suse.de>
21516 PR tree-optimization/80948
21517 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21518 stmt order as well.
21520 2017-06-02 Richard Biener <rguenther@suse.de>
21522 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21524 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21525 for inductions as relevant.
21527 2017-06-02 Richard Biener <rguenther@suse.de>
21529 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21530 (vectorizable_induction): ... this. Remove dead code.
21532 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21534 * builtins. (expand_builtin_alloca): Remove second parameter and
21535 infer its value from the first parameter instead.
21536 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21538 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21540 PR rtl-optimization/80903
21541 * loop-doloop.c (add_test): Unshare sequence.
21543 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21545 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21547 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21549 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21551 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21552 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21553 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21554 (xlogue_layout::get_stub_rtx): Make static.
21555 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21556 (xlogue_layout::compute_stub_managed_regs): Rename to...
21557 (xlogue_layout::count_stub_managed_regs): ...this.
21558 (xlogue_layout::is_stub_managed_reg): New function.
21559 (xlogue_layout::m_stub_names): Rename to...
21560 (xlogue_layout::s_stub_names): ...this, make static.
21561 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21562 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21563 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21564 xlogue_layout::s_stub_names): Instantiate statics.
21565 (stub_managed_regs): Remove.
21566 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21567 (disable_call_ms2sysv_xlogues): Rename to...
21568 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21569 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21571 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21572 change after reload_completed.
21573 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21575 (ix86_expand_prologue): Likewise.
21576 (ix86_expand_epilogue): Likewise.
21577 (ix86_expand_split_stack_prologue): Likewise.
21578 (ix86_compute_frame_layout): Remove frame parameter ...
21579 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21580 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21582 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21583 (ix86_frame): Move from here ...
21584 * config/i386/i386.h (ix86_frame): ... to here.
21585 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21586 complete ix86_frame data structure instead. Remove some_ld_name.
21588 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21590 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21591 symbols that hold a DECL_VALUE_EXPR.
21593 2017-06-01 Martin Jambor <mjambor@suse.cz>
21595 PR tree-optimization/80898
21596 * tree-sra.c (process_subtree_disqualification): Removed.
21597 (disqualify_candidate): Do not acll
21598 process_subtree_disqualification.
21599 (subtree_mark_written_and_enqueue): New function.
21600 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21601 RHS has been disqualified and re-queue LHS if necessary. Apart
21602 from that, ignore disqualified RHS.
21604 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21606 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21607 address fetch for z10 or later.
21609 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21611 * config/arc/arc.md (tst_movb): Add guard when splitting.
21613 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21615 * config/arc/arc.c (arc_can_eliminate): Test against
21616 arc_frame_pointer_needed.
21618 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21620 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21621 to prevent store reordering.
21622 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21623 (type): Add block type.
21624 (stack_tie): Define special instruction to be used in
21627 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21629 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21630 constraint. It is not valid for the pattern.
21631 (noncommutative_binary_comparison): Likewise.
21633 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21635 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21638 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21640 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21641 be used by the reg-alloc.
21643 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21645 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21646 reg-alloc when having mul64 or mul32x16 instructions.
21647 (mulsidi3): Likewise.
21648 (umulsidi3): Likewise.
21649 (mulsi32x16): New pattern.
21650 (mulsi64): Likewise.
21651 (mulsidi64): Likewise.
21652 (umulsidi64): Likewise.
21653 (MUL32x16_REG): Define.
21654 (mul64_600): Use MUL32x16_REG.
21655 (mac64_600): Likewise.
21656 (umul64_600): Likewise.
21657 (umac64_600): Likewise.
21659 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21661 * config/arc/arc.md (mulsi3_700): Make it commutative.
21663 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
21665 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
21667 (*sign_extendsidi2_insn): Likewise for movstosw.
21669 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21671 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21672 the type of the input discriminant value. Convert the
21673 discriminant value of signedness vary.
21675 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21677 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21678 Add to -Wall section.
21680 2017-06-01 Richard Biener <rguenther@suse.de>
21682 PR middle-end/66313
21683 * fold-const.c (fold_plusminus_mult_expr): If the factored
21684 factor may be zero use a wrapping type for the inner operation.
21685 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21686 and handle moved defs.
21687 (process_assignment): Properly guard the unary op case. Return a
21688 tri-state indicating that moving the stmt before the call may allow
21689 to continue. Pass through to_move.
21690 (find_tail_calls): Handle moving unrelated defs before
21693 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21696 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21697 splitter result in the canonical way.
21699 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21701 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21702 also for 32bit target. Update insn attributes.
21703 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21705 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21707 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21708 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21709 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21710 (_mm_maskz_min_ss): New intrinsics.
21712 2017-05-31 Martin Liska <mliska@suse.cz>
21714 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21715 Change comment style to one we normally use.
21716 (vectorizable_reduction): Likewise.
21717 (vectorizable_induction): Likewise.
21718 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21719 (vectorizable_call): Likewise.
21720 (vectorizable_simd_clone_call): Likewise.
21721 (vectorizable_conversion): Likewise.
21722 (vectorizable_assignment): Likewise.
21723 (vectorizable_shift): Likewise.
21724 (vectorizable_operation): Likewise.
21725 (vectorizable_store): Likewise.
21726 (vectorizable_load): Likewise.
21727 * tree-vectorizer.h: Likewise.
21729 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21731 * passes.c (emergency_dump_function): New.
21732 * tree-pass.h (emergency_dump_function): Declare.
21733 * plugin.c (plugins_internal_error_function): Remove.
21734 * plugin.h (plugins_internal_error_function): Remove declaration.
21735 * toplev.c (internal_error_function): New static function. Use it...
21736 (general_init): ...here.
21738 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21740 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21741 (arc_rtx_costs): Add costs for new patterns.
21742 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21743 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21745 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21747 * tree-ssa-strlen.c (get_next_strinfo): New function.
21748 (get_stridx_plus_constant): Use it.
21749 (zero_length_string): Likewise.
21750 (adjust_related_strinfos): Likewise.
21751 (adjust_last_stmt): Likewise.
21753 2017-05-31 Richard Biener <rguenther@suse.de>
21756 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21757 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21759 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21761 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21762 loop_vinfo argument and use of dependence distance vectors.
21763 Check instead whether the two references differ only in their
21764 initial value and assume that they have the same alignment if the
21765 difference is a multiple of the vector alignment.
21766 (vect_analyze_data_refs_alignment): Update call accordingly.
21768 2017-05-31 Martin Liska <mliska@suse.cz>
21771 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21773 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21775 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21776 (create_intersect_range_checks): Move from ...
21777 * tree-data-ref.c (create_intersect_range_checks_index)
21778 (create_intersect_range_checks): ... to here.
21779 (create_runtime_alias_checks): New function factored from ...
21780 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21781 here. Call above function.
21782 * tree-data-ref.h (create_runtime_alias_checks): New function.
21784 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21786 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21787 segment length for dr_b and compute it in wide_int.
21789 2017-05-31 Richard Biener <rguenther@suse.de>
21791 PR tree-optimization/80906
21792 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21793 and pass through iv_map.
21794 (copy_bb_and_scalar_dependences): Adjust.
21795 (translate_pending_phi_nodes): Likewise.
21796 (copy_loop_close_phi_args): Handle code-generating IVs instead
21799 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21801 * diagnostic-color.c (color_dict): Add "type-diff".
21802 (parse_gcc_colors): Update comment.
21803 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21804 -fdiagnostics-show-template-tree and -fno-elide-type.
21805 (GCC_COLORS): Add type-diff to example.
21807 (-fdiagnostics-show-template-tree): New.
21808 (-fno-elide-type): New.
21809 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21810 the pp_format_decoder callback. Call any m_format_postprocessor's
21812 (pretty_printer::pretty_printer): Initialize
21813 m_format_postprocessor.
21814 (pretty_printer::~pretty_printer): Delete any
21815 m_format_postprocessor.
21816 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21817 (class format_postprocessor): New class.
21818 (struct pretty_printer::format_decoder): Document the new parameters.
21819 (struct pretty_printer::m_format_postprocessor): New field.
21820 * tree-diagnostic.c (default_tree_printer): Update for new
21821 bool and const char ** params.
21822 * tree-diagnostic.h (default_tree_printer): Likewise.
21824 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21826 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21827 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21828 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21829 rs6000_gen_cell_microcode code.
21830 (rs6000_final_prescan_insn): Delete.
21831 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21832 "warn-cell-microcode" entries.
21833 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21834 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21835 throughout. Change cc_reg_not_micro_cr0_operand to
21836 cc_reg_not_cr0_operand throughout.
21837 (*extendhi<mode>2_noload): Delete.
21838 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21839 (mwarn-cell-microcode): Delete.
21840 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21841 -mgen-cell-microcode and -mwarn-cell-microcode.
21843 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21846 * config/i386/constraints.md (Yd): New constraint.
21848 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21849 and (?Yd, r) alternatives. Update insn attributes.
21850 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21851 and (?*Yd, r) alternatives. Update insn attributes.
21852 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21854 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21856 * gimplify.c (gimplify_modify_expr): Don't create a
21857 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21860 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21862 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21864 2017-05-30 Richard Biener <rguenther@suse.de>
21866 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21867 and reduc_def fields.
21868 (STMT_VINFO_REDUC_TYPE): New define.
21869 (STMT_VINFO_REDUC_DEF): Likewise.
21870 (vect_force_simple_reduction): Adjust prototype.
21871 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21872 (vect_is_simple_reduction): Remove check_reduction argument.
21873 (vect_force_simple_reduction): Adjust and set
21874 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21875 (vectorizable_reduction): Do not re-do reduction analysis
21876 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21877 * tree-parloops.c (gather_scalar_reductions): Adjust.
21879 2017-05-30 Richard Biener <rguenther@suse.de>
21881 PR middle-end/80901
21882 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21885 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21887 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21888 Introduce unknown_misalignment parameter and remove vf.
21889 (vect_peeling_hash_get_lowest_cost):
21890 Pass unknown_misalignment parameter.
21891 (vect_enhance_data_refs_alignment):
21892 Fix unsupportable data ref treatment.
21894 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21896 * tree-vect-data-refs.c (vect_get_data_access_cost):
21897 Workaround for SLP handling.
21898 (vect_enhance_data_refs_alignment):
21899 Compute costs for doing no peeling at all, compare to the best
21900 peeling costs so far and avoid peeling if cheaper.
21902 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21904 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21905 Return peeling info and set costs to zero for unlimited cost
21907 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21908 with unknown misalignment. Compute and costs for unknown
21909 misalignment, compare them to the costs for known misalignment
21910 and choose the cheapest for peeling.
21912 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21914 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21915 (vect_get_peeling_costs_all_drs): Create function.
21916 (vect_peeling_hash_get_lowest_cost):
21917 Use vect_get_peeling_costs_all_drs.
21918 (vect_peeling_supportable): Create function.
21919 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21921 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21923 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21924 DR_HAS_NEGATIVE_STEP.
21925 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21926 (vect_enhance_data_refs_alignment): Use.
21927 (vect_duplicate_ssa_name_ptr_info): Use.
21928 * tree-vectorizer.h (dr_misalignment): Use.
21929 (known_alignment_for_access_p): Use.
21931 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21934 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21936 (has_section_name): New function.
21938 2017-05-30 Martin Liska <mliska@suse.cz>
21941 * auto-profile.c (get_function_decl_from_block): Fix
21944 2017-05-30 Richard Biener <rguenther@suse.de>
21946 PR middle-end/80876
21947 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21949 2017-05-30 Martin Liska <mliska@suse.cz>
21951 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21952 * dumpfile.h (struct dump_file_info): Remove ctors.
21954 2017-05-30 Martin Liska <mliska@suse.cz>
21956 * predict.def: Fix GNU coding style.
21958 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21960 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21961 Mark 'to' argument with ATTRIBUTE_UNUSED.
21963 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21965 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21966 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
21967 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
21970 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
21972 * doc/install.texi (Options specification): Restore entry of
21973 --enable-sjlj-exceptions.
21975 2017-05-27 Michael Eager <eager@eagercon.com>
21978 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
21980 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
21982 * config/microblaze/microblaze.h
21983 (FIXED_REGISTERS): Update in macro.
21984 (CALL_USED_REGISTERS): Update in macro.
21986 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
21988 * doc/install.texi: Add links to macOS binary distributions.
21990 2017-05-27 Jakub Jelinek <jakub@redhat.com>
21994 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21996 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21998 2017-05-26 Martin Liska <mliska@suse.cz>
22000 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22002 2017-05-26 Martin Liska <mliska@suse.cz>
22004 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22005 always leading ';; '.
22006 (dump_bb_info): Likewise.
22007 (brief_dump_cfg): Likewise.
22008 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22009 * dumpfile.c: Remove usage of TDF_VERBOSE.
22010 * dumpfile.h (enum dump_kind): Likewise.
22011 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22012 * print-tree.c (debug_verbose): Remove.
22013 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22014 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22015 * tree-diagnostic.c (default_tree_printer): Replace
22016 TDF_DIAGNOSTIC with TDF_SLIM.
22018 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22020 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22021 in parameter loop, rather than loop_vinfo.
22022 (create_intersect_range_checks): Ditto.
22023 (vect_create_cond_for_alias_checks): Update call to above functions.
22025 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22027 PR tree-optimization/80815
22028 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22029 for merging runtime alias checks. Handle negative DR_STEPs.
22031 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22033 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22035 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22036 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22037 out code pruning runtime alias checks.
22038 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22039 factored out from above.
22040 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22042 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22044 (prune_runtime_alias_test_list): New decalaration.
22046 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22048 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22049 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22050 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22051 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22052 (operator==, comp_dr_with_seg_len_pair): Ditto.
22053 (vect_prune_runtime_alias_test_list): Ditto.
22055 2017-05-26 Martin Liska <mliska@suse.cz>
22058 * params.def: Bound partial-inlining-entry-probability param.
22060 2017-05-26 Marek Polacek <polacek@redhat.com>
22063 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22066 2017-05-26 Richard Biener <rguenther@suse.de>
22068 PR tree-optimization/80842
22069 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22072 2017-05-26 Richard Biener <rguenther@suse.de>
22074 PR tree-optimization/80844
22075 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22077 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22079 * doc/md.texi (Machine Constraints): Update x86 family
22080 machine constraints section to match 'config/i386/constraints.md'.
22082 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22084 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22086 2017-05-25 Nathan Sidwell <nathan@acm.org>
22088 * doc/invoke.texi (--enable-languages): Update documentation.
22090 2017-05-25 Martin Liska <mliska@suse.cz>
22092 * dumpfile.c: Add TDF_FOLDING.
22093 * dumpfile.h (enum dump_kind): Likewise.
22094 * genmatch.c (dt_simplify::gen_1): Use it.
22096 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22098 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22100 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22102 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22103 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22105 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22107 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22108 * match.pd (X == C): Rewrite it here.
22109 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22110 with_certain_nonzero_bits2): New predicates.
22111 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22113 2017-05-24 Nathan Sidwell <nathan@acm.org>
22115 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22118 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22121 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22123 * config/powerpcspe: New port. Files are copied from the rs6000
22124 port, with "rs6000" in filenames replaced by "powerpcspe".
22126 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22128 PR rtl-optimization/80754
22129 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22131 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22133 * config/sparc/sparc.md (length): Return the correct value for -mflat
22134 sibcalls to match output_sibcall.
22136 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22140 * config/rs6000/rs6000.c (struct machine_function): Add new field
22142 (rs6000_get_separate_components): Init that field, use it.
22143 (rs6000_components_for_bb): Use the field.
22145 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22147 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22149 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22151 PR middle-end/80823
22152 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22154 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22157 * config/s390/s390.c (s390_check_qrst_address): Check incoming
22158 address against address_operand predicate.
22159 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
22161 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
22163 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
22164 parameters passed indirectly.
22166 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
22168 * config/i386/i386.md (*movdi_internal): Remove SSE4
22169 alternative 18 (?r, *v). Update insn attributes.
22170 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
22171 Update insn attributes.
22172 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
22173 Update insn attributes.
22174 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
22175 alternative 1 (r, v). Remove isa attribute.
22176 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
22177 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
22178 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
22180 2017-05-23 Tom de Vries <tom@codesourcery.com>
22182 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
22185 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22187 * cgraphunit.c (symbol_table::process_new_functions): Update.
22188 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
22189 (inline_generate_summary): Rename to ...
22190 (ipa_fn_summary_generate): ... this one.
22191 (inline_read_summary): Rename to ...
22192 (ipa_fn_summary_read): ... this one.
22193 (inline_write_summary): Rename to ...
22194 (ipa_fn_summary_write): ... this one.
22195 (inline_free_summary): Rename to ...
22196 (ipa_free_fn_summary): ... this one.
22197 (pass_data_local_fn_summary, pass_local_fn_summary,
22198 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
22199 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
22200 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
22201 make_pass_ipa_fn_summary): New.
22202 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
22203 inline_write_summary, inline_free_summary): Remove.
22204 (ipa_free_fn_summary) : New.
22205 * ipa-inline.c (ipa_inline): Update.
22206 (pass_ipa_inline): Do not generate summaries.
22207 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
22209 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
22210 and add pass_ipa_fn_summary.
22211 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
22213 (make_pass_inline_parameters): Remove.
22215 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
22217 * omp-low.c (struct omp_context): Remove "default_kind" member.
22220 * omp-offload.c (execute_oacc_device_lower): Remove the
22221 parallelism dimensions function attributes for unparallelized
22222 OpenACC kernels constructs.
22224 2017-05-23 Martin Liska <mliska@suse.cz>
22226 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22228 (cgraph_edge::make_speculative): Likewise.
22229 (cgraph_edge::resolve_speculation): Likewise.
22230 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22231 (cgraph_node::dump): Likewise.
22232 * cgraph.h: Likewise.
22233 * cgraphunit.c (analyze_functions): Likewise.
22234 (symbol_table::compile): Likewise.
22235 * ipa-cp.c (print_all_lattices): Likewise.
22236 (determine_versionability): Likewise.
22237 (initialize_node_lattices): Likewise.
22238 (ipcp_verify_propagated_values): Likewise.
22239 (estimate_local_effects): Likewise.
22240 (update_profiling_info): Likewise.
22241 (create_specialized_node): Likewise.
22242 (perhaps_add_new_callers): Likewise.
22243 (decide_about_value): Likewise.
22244 (decide_whether_version_node): Likewise.
22245 (identify_dead_nodes): Likewise.
22246 (ipcp_store_bits_results): Likewise.
22247 * ipa-devirt.c (dump_targets): Likewise.
22248 (ipa_devirt): Likewise.
22249 * ipa-icf.c (sem_item::dump): Likewise.
22250 (sem_function::equals): Likewise.
22251 (sem_variable::equals): Likewise.
22252 (sem_item_optimizer::read_section): Likewise.
22253 (sem_item_optimizer::execute): Likewise.
22254 (congruence_class::dump): Likewise.
22255 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22256 (dump_inline_summary): Likewise.
22257 (estimate_node_size_and_time): Likewise.
22258 (inline_analyze_function): Likewise.
22259 * ipa-inline-transform.c (inline_call): Likewise.
22260 * ipa-inline.c (report_inline_failed_reason): Likewise.
22261 (want_early_inline_function_p): Likewise.
22262 (edge_badness): Likewise.
22263 (update_edge_key): Likewise.
22264 (inline_small_functions): Likewise.
22265 * ipa-profile.c (ipa_profile): Likewise.
22266 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22267 (ipa_make_edge_direct_to_target): Likewise.
22268 (remove_described_reference): Likewise.
22269 (ipa_impossible_devirt_target): Likewise.
22270 (propagate_controlled_uses): Likewise.
22271 (ipa_print_node_params): Likewise.
22272 (ipcp_transform_function): Likewise.
22273 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22274 (propagate_pure_const): Likewise.
22275 * ipa-reference.c (generate_summary): Likewise.
22276 (read_write_all_from_decl): Likewise.
22277 (propagate): Likewise.
22278 (ipa_reference_read_optimization_summary): Likewise.
22279 * ipa-utils.c (ipa_merge_profiles): Likewise.
22280 * ipa.c (walk_polymorphic_call_targets): Likewise.
22281 (symbol_table::remove_unreachable_nodes): Likewise.
22282 (ipa_single_use): Likewise.
22283 * passes.c (execute_todo): Likewise.
22284 * predict.c (drop_profile): Likewise.
22285 * symtab.c (symtab_node::get_dump_name): New function.
22286 (symtab_node::dump_name): Likewise.
22287 (symtab_node::dump_asm_name): Likewise.
22288 (symtab_node::dump_references): Likewise.
22289 (symtab_node::dump_referring): Likewise.
22290 (symtab_node::dump_base): Likewise.
22291 (symtab_node::debug_symtab): Likewise.
22292 * tree-sra.c (convert_callers_for_node): Likewise.
22293 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22294 * value-prof.c (init_node_map): Likewise.
22296 2017-05-23 Martin Liska <mliska@suse.cz>
22298 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22299 and symtab_node::debug_symtab to symbol_table::debug.
22300 * cgraphunit.c (analyze_functions): Use the renamed function.
22301 (symbol_table::compile): Likewise.
22302 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22303 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22304 * passes.c (execute_todo): Likewise.
22305 * symtab.c (symbol_table::dump): New function.
22306 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22308 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22310 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22311 that nonconst implies exec.
22313 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22315 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22316 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22317 (inline_edge_summary_vec): Turn into ...
22318 (ipa_call_summaries): ... this one.
22319 (redirect_to_unreachable, edge_set_predicate,
22320 evaluate_properties_for_edge, inline_summary_alloc,
22321 reset_ipa_call_summary, reset_inline_summary,
22322 inline_summary_t::duplicate): Update.
22323 (inline_edge_duplication_hook): Turn to ...
22324 (ipa_call_summary_t::duplicate): ... this one.
22325 (inline_edge_removal_hook): Turn to ...
22326 (ipa_call_summary_t::remove): ... this one.
22327 (dump_inline_edge_summary): Turn to ...
22328 (dump_ipa_call_summary): ... this one.
22329 (estimate_function_body_sizes): Update.
22330 (inline_update_callee_summaries): Update.
22331 (remap_edge_change_prob): Update.
22332 (remap_edge_summaries): Update.
22333 (inline_merge_summary): Update.
22334 (do_estimate_edge_time): Update.
22335 (inline_generate_summary): Update.
22336 (inline_read_section): Update.
22337 (inline_read_summary): Update.
22338 (inline_free_summary): Update.
22339 * ipa-inline.c (can_inline_edge_p): Update.
22340 (compute_inlined_call_time): Update.
22341 (want_inline_small_function_p): Update.
22342 (edge_badness): Update.
22343 (early_inliner): Update.
22344 * ipa-inline.h (inline_edge_summary): Turn to ...
22345 (ipa_call_summary): ... this one.
22346 (ipa_call_summary_t): New class.
22347 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22348 (ipa_call_summaries): New.
22349 (inline_edge_summary): Remove.
22350 (estimate_edge_growth): Update.
22351 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22352 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22353 * ipa-split.c (execute_split_functions): Update.
22354 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22356 2017-05-23 Tom de Vries <tom@codesourcery.com>
22358 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22359 attributes): Document rdrand effective target.
22361 2017-05-23 Tom de Vries <tom@codesourcery.com>
22363 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22364 attributes): Sort alphabetically.
22366 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22368 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22370 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22373 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22374 V2DF/V2DI splat into two separate patterns, one that handles
22375 registers, and the other that only handles memory. Drop support
22376 for splatting from a GPR on ISA 2.07 and then splitting the
22377 splat into direct move and splat.
22378 (vsx_splat_<mode>_reg): Likewise.
22379 (vsx_splat_<mode>_mem): Likewise.
22381 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22383 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22385 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22387 PR middle-end/80809
22388 * omp-low.c (finish_taskreg_remap): New function.
22389 (finish_taskreg_scan): If unit size of ctx->record_type
22390 is non-constant, unshare the size expression and replace
22391 decls in it with possible outer var refs.
22393 PR middle-end/80809
22394 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22395 GOVD_SHARED rather than GOVD_PRIVATE with it.
22396 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22397 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22399 PR middle-end/80853
22400 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22401 as last argument to build_outer_var_ref for pointer bases of array
22402 section reductions.
22404 2017-05-19 Martin Sebor <msebor@redhat.com>
22406 * print-tree.c (print_node): Print DECL_READ_P flag.
22408 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22410 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22411 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22412 * cgraph.c: Likewise.
22413 * cgraphunit.c: Likewise.
22414 * gengtype.c: Likewise.
22415 * ipa-cp.c: Likewise.
22416 * ipa-devirt.c: Likewise.
22417 * ipa-icf.c: Likewise.
22418 * ipa-predicate.c: Likewise.
22419 * ipa-profile.c: Likewise.
22420 * ipa-prop.c: Likewise.
22421 * ipa-split.c: Likewise.
22423 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22424 edge_predicate_pool, dump_inline_hints,
22425 inline_summary::account_size_time, redirect_to_unreachable,
22426 edge_set_predicate, set_hint_predicate,
22427 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22428 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22429 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22430 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22431 ipa_call_summary_t::remove, initialize_growth_caches,
22432 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22433 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22434 mark_modified, unmodified_parm_1, unmodified_parm,
22435 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22436 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22437 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22438 will_be_nonconstant_predicate, record_modified_bb_info,
22439 get_minimal_bb, record_modified, param_change_prob,
22440 phi_result_unknown_predicate, predicate_for_phi_result,
22441 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22442 estimate_function_body_sizes, compute_inline_parameters,
22443 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22444 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22445 inline_update_callee_summaries, remap_edge_change_prob,
22446 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22447 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22448 inline_analyze_function, inline_summary_t::insert,
22449 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22450 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22451 inline_free_summary): Move to ipa-fnsummary.h
22452 (predicate_t): Remove.
22453 * ipa-fnsummary.c: New file.
22454 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22455 (enum inline_hints_vals, inline_hints, agg_position_info,
22456 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22457 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22458 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22459 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22460 inline_read_summary, inline_write_summary, inline_free_summary,
22461 inline_analyze_function, initialize_inline_failed,
22462 inline_merge_summary, inline_update_overall_summary,
22463 compute_inline_parameters): Move to ipa-fnsummary.h
22464 * ipa-fnsummary.h: New file.
22465 * ipa-inline-transform.h: Include ipa-inline.h.
22466 * ipa-inline.c: LIkewise.
22468 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22470 * ipa-inline.c (edge_badness): Use inlined_time instead of
22471 inline_summaries->get.
22473 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22475 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22477 2017-05-22 Nathan Sidwell <nathan@acm.org>
22479 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22480 (fdump-lang): Document 'raw' option.
22481 * dumpfile.h (TDI_tu): Delete.
22482 * dumpfile.c (dump_files): Remove translation-unit.
22483 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22485 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22487 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22488 command option from $(AWK) call.
22489 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22490 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22491 [FORMAT]: Remove handling of variable.
22492 * config/avr/t-multilib: Regenerate.
22494 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22496 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22498 (dump_inline_summary): Do not print self_time.
22499 (estimate_function_body_sizes): Do not set self_time.
22500 (compute_inline_parameters): Likewise.
22501 (inline_read_section, inline_write_summary): Do not stream self_time.
22502 * ipa-inline.h (inline_summary): Drop self_time.
22504 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22506 * ipa-inline-analysis.c (account_size_time): Rename to ...
22507 (inline_summary::account_size_time): ... this one.
22508 (reset_ipa_call_summary): Turn to ...
22509 (ipa_call_summary::reset): ... this one.
22510 (reset_inline_summary): Turn to ...
22511 (inline_summary::reset): ... this one.
22512 (inline_summary_t::remove): Update.
22513 (inline_summary_t::duplicate): Update.
22514 (ipa_call_summary_t::remove): Update.
22515 (dump_inline_summary): Update.
22516 (estimate_function_body_sizes): Update.
22517 (compute_inline_parameters): Update.
22518 (estimate_node_size_and_time): Update.
22519 (inline_merge_summary): Update.
22520 (inline_update_overall_summary): Update.
22521 (inline_read_section): Update.
22522 (inline_write_summary): Update.
22523 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22524 add account_size_time and reset member functions.
22525 (ipa_call_summary): Add reset function.
22526 * ipa-predicate.h (predicate::operator &): Constify.
22528 2017-05-22 Richard Biener <rguenther@suse.de>
22530 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22532 2017-05-19 Jason Merrill <jason@redhat.com>
22534 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22536 2017-05-19 Marek Polacek <polacek@redhat.com>
22539 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22540 TYPE_OVERFLOW_WRAPS checks.
22542 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22544 * tree-core.h (enum omp_clause_default_kind): Add
22545 "OMP_CLAUSE_DEFAULT_PRESENT".
22546 * tree-pretty-print.c (dump_omp_clause): Handle it.
22547 * gimplify.c (enum gimplify_omp_var_data): Add
22548 "GOVD_MAP_FORCE_PRESENT".
22549 (gimplify_adjust_omp_clauses_1): Map it to
22550 "GOMP_MAP_FORCE_PRESENT".
22551 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22553 * gimplify.c (oacc_default_clause): Clarify.
22555 2017-05-19 Nathan Sidwell <nathan@acm.org>
22557 LANG_HOOK_REGISTER_DUMPS
22558 * toplev.c (general_init): Call register dump lang hook.
22559 * doc/invoke.texi: Document -fdump-lang option family.
22560 * dumpfile.c (dump_files): Remove class dump here.
22561 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22562 * dumpfile.h (tree_dump_index): Remove TDI_class.
22563 * langhooks-def.h (lhd_register_dumps): Declare.
22564 (LANG_HOOKS_REGISTER_DUMPS): Define.
22565 (LANG_HOOKS_INITIALIZER): Add it.
22566 * langhooks.c (lhd_register_dumps): Define.
22567 * langhooks.h (struct lang_hooks): Add register_dumps.
22569 2017-05-19 Nathan Sidwell <nathan@acm.org>
22571 * context.h (context::set_passes): New.
22572 * context.c (context::context): Do not create pass manager.
22573 * toplev.c (general_init): Create pass manager here.
22575 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22577 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22578 use this splitter if two add or or instructions would also work for
22579 the constant we want to generate.
22581 2017-05-19 Richard Biener <rguenther@suse.de>
22584 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22585 predicate evaluation.
22587 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22589 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22591 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22594 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22596 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22597 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22598 (inline_edge_summary_vec): Turn into ...
22599 (ipa_call_summaries): ... this one.
22600 (redirect_to_unreachable, edge_set_predicate,
22601 evaluate_properties_for_edge, inline_summary_alloc,
22602 reset_ipa_call_summary, reset_inline_summary,
22603 inline_summary_t::duplicate): Update.
22604 (inline_edge_duplication_hook): Turn to ...
22605 (ipa_call_summary_t::duplicate): ... this one.
22606 (inline_edge_removal_hook): Turn to ...
22607 (ipa_call_summary_t::remove): ... this one.
22608 (dump_inline_edge_summary): Turn to ...
22609 (dump_ipa_call_summary): ... this one.
22610 (estimate_function_body_sizes): Update.
22611 (inline_update_callee_summaries): Update.
22612 (remap_edge_change_prob): Update.
22613 (remap_edge_summaries): Update.
22614 (inline_merge_summary): Update.
22615 (do_estimate_edge_time): Update.
22616 (inline_generate_summary): Update.
22617 (inline_read_section): Update.
22618 (inline_read_summary): Update.
22619 (inline_free_summary): Update.
22620 * ipa-inline.c (can_inline_edge_p): Update.
22621 (compute_inlined_call_time): Update.
22622 (want_inline_small_function_p): Update.
22623 (edge_badness): Update.
22624 (early_inliner): Update.
22625 * ipa-inline.h (inline_edge_summary): Turn to ...
22626 (ipa_call_summary): ... this one.
22627 (ipa_call_summary_t): New class.
22628 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22629 (ipa_call_summaries): New.
22630 (inline_edge_summary): Remove.
22631 (estimate_edge_growth): Update.
22632 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22633 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22634 * ipa-split.c (execute_split_functions): Update.
22635 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22637 2017-05-19 Richard Biener <rguenther@suse.de>
22639 PR middle-end/80764
22640 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22642 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22644 * config/rs6000/rs6000.c (struct machine_function): Add field
22645 fpr_is_wrapped_separately.
22646 (rs6000_get_separate_components): Use 64 components. Handle the
22647 new FPR components.
22648 (rs6000_components_for_bb): Handle the FPR components.
22649 (rs6000_emit_prologue_components): Handle the FPR components.
22650 (rs6000_emit_epilogue_components): Handle the FPR components.
22651 (rs6000_set_handled_components): Handle the FPR components.
22652 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22653 that are already separately shrink-wrapped.
22654 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
22655 that are already separately shrink-wrapped.
22657 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22660 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
22663 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
22664 (define_peephole2 for Altivec d-form load): Add peepholes to catch
22665 cases where the register allocator uses a move and an offsettable
22666 memory operation to/from a FPR register on ISA 2.06/2.07.
22667 (define_peephole2 for Altivec d-form store): Likewise.
22669 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22672 * config/i386/mmx.md (*mov<mode>_internal): Enable
22673 alternatives 11, 12, 13 and 14 also for 32bit targets.
22674 Remove alternatives 15, 16, 17 and 18.
22675 * config/i386/sse.md (vec_concatv2di): Change
22676 alternative (!x, *y) to (x, ?!*Yn).
22678 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22680 * dumpfile.h (enum dump_kind): Remove stray comma.
22682 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22684 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22685 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22686 predicate::num_conditions
22687 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22688 (CHANGED): turn into predicate::changed.
22689 (agg_position_info): Move to ipa-predicate.h
22690 (add_condition, predicate::add_clause, predicate::operator &=,
22691 predicate::or_with, predicate::evaluate, predicate::probability,
22692 dump_condition, dump_clause, predicate::dump,
22693 predicate::remap_after_duplication, predicate::remap_after_inlining,
22694 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22695 (evaluate_conditions_for_known_args): Update.
22696 (set_cond_stmt_execution_predicate): Update.
22697 * ipa-inline.h: Include ipa-predicate.h
22698 (condition, inline_param_summary, conditions, agg_position_info,
22699 predicate): Move to ipa-predicate.h
22700 * ipa-predicate.c: New file.
22701 * ipa-predicate.h: New file.
22703 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22705 * final.c (leaf_function_p): Check we are not in a sequence.
22707 2017-05-18 Martin Liska <mliska@suse.cz>
22709 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22710 * dumpfile.c (dump_register): Use new enum dump_kind.
22711 (get_dump_file_name): Likewise.
22712 (dump_enable_all): Likewise.
22713 (dump_switch_p_1): Likewise.
22714 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22715 * dumpfile.h (enum dump_kind): New enum type.
22716 (struct dump_file_info): Create constructor and
22717 format fields and comments.
22718 * passes.c (pass_manager::register_one_dump_file):
22720 * statistics.c (statistics_early_init): Likewise.
22721 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22722 TDF_TREE with TDF_SLIM.
22723 (gather_memory_references_ref): Likewise.
22725 2017-05-18 Martin Liska <mliska@suse.cz>
22727 * vec.h (struct vnull): Use it.
22729 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22731 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22732 (true_predicate, false_predicate, true_predicate_p,
22733 false_predicate_p): Remove.
22734 (single_cond_predicate, not_inlined_predicate): Turn to member function
22736 (add_condition): Update.
22737 (add_clause): Turn to...
22738 (predicate::add_clause): ... this one; update; allow passing NULL
22740 (and_predicates): Turn to ...
22741 (predicate::operator &=): ... this one.
22742 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22743 (or_predicates): Turn to ...
22744 (predicate::or_with): ... this one.
22745 (evaluate_predicate): Turn to ...
22746 (predicate::evaluate): ... this one.
22747 (predicate_probability): Turn to ...
22748 (predicate::probability): ... this one.
22749 (dump_condition): Update.
22750 (dump_predicate): Turn to ...
22751 (predicate::dump): ... this one.
22752 (account_size_time): Update.
22753 (edge_set_predicate): Update.
22754 (set_hint_predicate): UPdate.
22755 (evaluate_conditions_for_known_args): Update.
22756 (evaluate_properties_for_edge): Update.
22757 (remap_predicate_after_duplication): Turn to...
22758 (predicate::remap_after_duplication): ... this one.
22759 (remap_hint_predicate_after_duplication): Update.
22760 (inline_summary_t::duplicate): UPdate.
22761 (dump_inline_edge_summary): Update.
22762 (dump_inline_summary): Update.
22763 (set_cond_stmt_execution_predicate): Update.
22764 (set_switch_stmt_execution_predicate): Update.
22765 (compute_bb_predicates): Update.
22766 (will_be_nonconstant_expr_predicate): Update.
22767 (will_be_nonconstant_predicate): Update.
22768 (phi_result_unknown_predicate): Update.
22769 (predicate_for_phi_result): Update.
22770 (array_index_predicate): Update.
22771 (estimate_function_body_sizes): Update.
22772 (estimate_node_size_and_time): Update.
22773 (estimate_ipcp_clone_size_and_time): Update.
22774 (remap_predicate): Rename to ...
22775 (predicate::remap_after_inlining): ... this one.
22776 (remap_hint_predicate): Update.
22777 (inline_merge_summary): Update.
22778 (inline_update_overall_summary): Update.
22779 (estimate_size_after_inlining): Update.
22780 (read_predicate): Rename to ...
22781 (predicate::stream_in): ... this one.
22782 (read_inline_edge_summary): Update.
22783 (write_predicate): Rename to ...
22784 (predicate::stream_out): ... this one.
22785 (write_inline_edge_summary): Update.
22786 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22787 (clause_t): Turn to uint32_t
22788 (predicate): Turn to class; implement constructor and operators
22790 (size_time_entry): Update.
22791 (inline_summary): Update.
22792 (inline_edge_summary): Update.
22794 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22796 * fold-const.c (fold_binary_loc): Move transformation...
22797 * match.pd (C - X CMP X): ... here.
22799 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22801 * config/sparc/sparc.c (sparc_option_override): Set function
22802 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22803 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22805 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22807 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22809 2017-05-18 Marek Polacek <polacek@redhat.com>
22812 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22813 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22814 gimple_assign_load_p.
22816 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22818 PR middle-end/80692
22819 * real.c (do_compare): Give decimal_do_compare preference over
22820 comparing just the signs.
22822 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22824 * doc/md.texi (Canonicalization of Instructions): Describe the
22825 canonical form of instructions that inherently set a condition
22828 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22830 PR middle-end/80775
22831 * tree-cfg.c: Move deletion of unreachable case statements to after
22832 the merging of consecutive case labels.
22834 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22836 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22837 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22838 restoring of callee-saved registers.
22840 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22842 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22843 * config/visium/visium.c (single_set_and_flags): Likewise.
22844 * config/visium/visium.md (Substitutions): Likewise.
22846 2017-05-17 Martin Liska <mliska@suse.cz>
22848 * cfg.c: Introduce dump_flags_t type and
22849 use it instead of int type.
22851 * cfghooks.c: Likewise.
22852 * cfghooks.h (struct cfg_hooks): Likewise.
22853 * cfgrtl.c: Likewise.
22854 * cfgrtl.h: Likewise.
22855 * cgraph.c (cgraph_node::get_body): Likewise.
22856 * coretypes.h: Likewise.
22857 * domwalk.c: Likewise.
22858 * domwalk.h: Likewise.
22859 * dumpfile.c (struct dump_option_value_info): Likewise.
22860 (dump_enable_all): Likewise.
22861 (dump_switch_p_1): Likewise.
22862 (opt_info_switch_p): Likewise.
22863 * dumpfile.h (enum tree_dump_index): Likewise.
22864 (struct dump_file_info): Likewise.
22865 * genemit.c: Likewise.
22866 * generic-match-head.c: Likewise.
22867 * gengtype.c (open_base_files): Likewise.
22868 * gimple-pretty-print.c: Likewise.
22869 * gimple-pretty-print.h: Likewise.
22870 * graph.c (print_graph_cfg): Likewise.
22871 * graphite-scop-detection.c (dot_all_sese): Likewise.
22872 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22873 * loop-unroll.c (report_unroll): Likewise.
22874 * passes.c (pass_manager::register_one_dump_file): Likewise.
22875 * print-tree.c: Likewise.
22876 * statistics.c: Likewise.
22877 * tree-cfg.c: Likewise.
22878 * tree-cfg.h: Likewise.
22879 * tree-dfa.c: Likewise.
22880 * tree-dfa.h: Likewise.
22881 * tree-dump.c (dump_function): Likewise.
22882 * tree-dump.h (struct dump_info): Likewise.
22883 * tree-pretty-print.c: Likewise.
22884 * tree-pretty-print.h: Likewise.
22885 * tree-ssa-live.c: Likewise.
22886 * tree-ssa-live.h: Likewise.
22887 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22888 * tree-vect-loop.c: Likewise.
22889 * tree-vect-slp.c: Likewise.
22891 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22892 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22894 PR tree-optimization/80457
22895 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22896 of all arguments to a statement as scalar_to_vec operations.
22897 (vectorizable_call): Adjust call to vect_model_simple_cost for
22899 (vectorizable_conversion): Likewise.
22900 (vectorizable_assignment): Likewise.
22901 (vectorizable_shift): Likewise.
22902 (vectorizable_operation): Likewise.
22903 (vectorizable_comparison): Likewise.
22904 (vect_is_simple_cond): Record the def types for operands.
22905 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22906 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22907 for statement argument count.
22909 2017-05-16 Carl Love <cel@us.ibm.com>
22911 * config/rs6000/rs6000-c: Add support for built-in functions
22912 vector unsigned long long vec_bperm (vector unsigned long long,
22913 vector unsigned char)
22914 vector signed long long vec_mule (vector signed int,
22916 vector unsigned long long vec_mule (vector unsigned int,
22917 vector unsigned int)
22918 vector signed long long vec_mulo (vector signed int,
22920 vector unsigned long long vec_mulo (vector unsigned int,
22921 vector unsigned int)
22922 vector signed char vec_sldw (vector signed char,
22923 vector signed char,
22925 vector unsigned char vec_sldw (vector unsigned char,
22926 vector unsigned char,
22928 vector signed short vec_sldw (vector signed short,
22929 vector signed short,
22931 vector unsigned short vec_sldw (vector unsigned short,
22932 vector unsigned short,
22934 vector signed int vec_sldw (vector signed int,
22937 vector unsigned int vec_sldw (vector unsigned int,
22938 vector unsigned int,
22940 vector signed long long vec_sldw (vector signed long long,
22941 vector signed long long,
22943 vector unsigned long long vec_sldw (vector unsigned long long,
22944 vector unsigned long long,
22946 * config/rs6000/rs6000-c: Add support for built-in functions
22947 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22948 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22949 * doc/extend.texi: Update the built-in documentation for the
22950 new built-in functions.
22952 2017-05-16 Marek Polacek <polacek@redhat.com>
22956 * tree.c (save_expr): Don't fold the expression.
22958 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22960 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22961 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22962 and (?*y,m). Update insn attributes.
22964 2017-05-16 Martin Liska <mliska@suse.cz>
22966 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
22967 flags argument of print_gimple_stmt, print_gimple_expr,
22968 print_generic_stmt and print_generic_expr.
22969 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
22970 * coretypes.h: Likewise.
22971 * except.c (dump_eh_tree): Likewise.
22972 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
22973 * gimple-pretty-print.h: Likewise.
22974 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
22975 (backprop::push_to_worklist): Likewise.
22976 (backprop::pop_from_worklist): Likewise.
22977 (backprop::process_use): Likewise.
22978 (backprop::intersect_uses): Likewise.
22979 (note_replacement): Likewise.
22980 * gimple-ssa-store-merging.c
22981 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
22982 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
22983 (pass_store_merging::execute): Likewise.
22984 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
22985 (ssa_base_cand_dump_callback): Likewise.
22986 (dump_incr_vec): Likewise.
22987 (replace_refs): Likewise.
22988 (replace_mult_candidate): Likewise.
22989 (create_add_on_incoming_edge): Likewise.
22990 (create_phi_basis): Likewise.
22991 (insert_initializers): Likewise.
22992 (all_phi_incrs_profitable): Likewise.
22993 (introduce_cast_before_cand): Likewise.
22994 (replace_one_candidate): Likewise.
22995 * gimplify.c (gimplify_expr): Likewise.
22996 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
22997 (set_rename): Likewise.
22998 (rename_uses): Likewise.
22999 (copy_loop_phi_nodes): Likewise.
23000 (add_close_phis_to_merge_points): Likewise.
23001 (copy_loop_close_phi_args): Likewise.
23002 (copy_cond_phi_args): Likewise.
23003 (graphite_copy_stmts_from_block): Likewise.
23004 (translate_pending_phi_nodes): Likewise.
23005 * graphite-poly.c (print_pdr): Likewise.
23006 (dump_gbb_cases): Likewise.
23007 (dump_gbb_conditions): Likewise.
23008 (print_scop_params): Likewise.
23009 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23010 (build_cross_bb_scalars_use): Likewise.
23011 (gather_bbs::before_dom_children): Likewise.
23012 * hsa-dump.c (dump_hsa_immed): Likewise.
23013 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23014 (get_replacement_map): Likewise.
23015 * ipa-inline-analysis.c (dump_condition): Likewise.
23016 (estimate_function_body_sizes): Likewise.
23017 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23018 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23019 * ipa-prop.c (ipa_dump_param): Likewise.
23020 (ipa_print_node_jump_functions_for_edge): Likewise.
23021 (ipa_modify_call_arguments): Likewise.
23022 (ipa_modify_expr): Likewise.
23023 (ipa_dump_param_adjustments): Likewise.
23024 (ipa_dump_agg_replacement_values): Likewise.
23025 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23026 * ipa-pure-const.c (check_stmt): Likewise.
23027 (pass_nothrow::execute): Likewise.
23028 * ipa-split.c (execute_split_functions): Likewise.
23029 * omp-offload.c (dump_oacc_loop_part): Likewise.
23030 (dump_oacc_loop): Likewise.
23031 * trans-mem.c (tm_log_emit): Likewise.
23032 (tm_memopt_accumulate_memops): Likewise.
23033 (dump_tm_memopt_set): Likewise.
23034 (dump_tm_memopt_transform): Likewise.
23035 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23036 (print_loop): Likewise.
23037 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23038 (chkp_gather_checks_info): Likewise.
23039 (chkp_get_check_result): Likewise.
23040 (chkp_remove_check_if_pass): Likewise.
23041 (chkp_use_outer_bounds_if_possible): Likewise.
23042 (chkp_reduce_bounds_lifetime): Likewise.
23043 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23044 (chkp_mark_completed_bounds): Likewise.
23045 (chkp_register_incomplete_bounds): Likewise.
23046 (chkp_mark_invalid_bounds): Likewise.
23047 (chkp_maybe_copy_and_register_bounds): Likewise.
23048 (chkp_build_returned_bound): Likewise.
23049 (chkp_get_bound_for_parm): Likewise.
23050 (chkp_build_bndldx): Likewise.
23051 (chkp_get_bounds_by_definition): Likewise.
23052 (chkp_generate_extern_var_bounds): Likewise.
23053 (chkp_get_bounds_for_decl_addr): Likewise.
23054 * tree-chrec.c (chrec_apply): Likewise.
23055 * tree-data-ref.c (dump_data_reference): Likewise.
23056 (dump_subscript): Likewise.
23057 (dump_data_dependence_relation): Likewise.
23058 (analyze_overlapping_iterations): Likewise.
23059 * tree-inline.c (expand_call_inline): Likewise.
23060 (tree_function_versioning): Likewise.
23061 * tree-into-ssa.c (dump_defs_stack): Likewise.
23062 (dump_currdefs): Likewise.
23063 (dump_names_replaced_by): Likewise.
23064 (dump_update_ssa): Likewise.
23065 (update_ssa): Likewise.
23066 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23067 * tree-parloops.c (build_new_reduction): Likewise.
23068 (try_create_reduction_list): Likewise.
23069 (ref_conflicts_with_region): Likewise.
23070 (oacc_entry_exit_ok_1): Likewise.
23071 (oacc_entry_exit_single_gang): Likewise.
23072 * tree-pretty-print.h: Likewise.
23073 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23074 (get_scalar_evolution): Likewise.
23075 (add_to_evolution): Likewise.
23076 (get_loop_exit_condition): Likewise.
23077 (analyze_evolution_in_loop): Likewise.
23078 (analyze_initial_condition): Likewise.
23079 (analyze_scalar_evolution): Likewise.
23080 (instantiate_scev): Likewise.
23081 (number_of_latch_executions): Likewise.
23082 (gather_chrec_stats): Likewise.
23083 (final_value_replacement_loop): Likewise.
23084 (scev_const_prop): Likewise.
23085 * tree-sra.c (dump_access): Likewise.
23086 (disqualify_candidate): Likewise.
23087 (create_access): Likewise.
23088 (reject): Likewise.
23089 (maybe_add_sra_candidate): Likewise.
23090 (create_access_replacement): Likewise.
23091 (analyze_access_subtree): Likewise.
23092 (analyze_all_variable_accesses): Likewise.
23093 (sra_modify_assign): Likewise.
23094 (initialize_constant_pool_replacements): Likewise.
23095 (find_param_candidates): Likewise.
23096 (decide_one_param_reduction): Likewise.
23097 (replace_removed_params_ssa_names): Likewise.
23098 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23099 * tree-ssa-copy.c (dump_copy_of): Likewise.
23100 (copy_prop_visit_cond_stmt): Likewise.
23101 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23102 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23103 (record_equivalences_from_stmt): Likewise.
23104 * tree-ssa-dse.c (compute_trims): Likewise.
23105 (delete_dead_call): Likewise.
23106 (delete_dead_assignment): Likewise.
23107 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23108 (forward_propagate_into_cond): Likewise.
23109 (pass_forwprop::execute): Likewise.
23110 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23111 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23113 (move_computations_worker): Likewise.
23114 (execute_sm): Likewise.
23115 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23116 (remove_exits_and_undefined_stmts): Likewise.
23117 (remove_redundant_iv_tests): Likewise.
23118 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23119 (adjust_iv_update_pos): Likewise.
23120 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23121 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23122 (value_replacement): Likewise.
23123 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23124 * tree-ssa-pre.c (print_pre_expr): Likewise.
23125 (get_representative_for): Likewise.
23126 (create_expression_by_pieces): Likewise.
23127 (insert_into_preds_of_block): Likewise.
23128 (eliminate_insert): Likewise.
23129 (eliminate_dom_walker::before_dom_children): Likewise.
23130 (eliminate): Likewise.
23131 (remove_dead_inserted_code): Likewise.
23132 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23133 * tree-ssa-reassoc.c (get_rank): Likewise.
23134 (eliminate_duplicate_pair): Likewise.
23135 (eliminate_plus_minus_pair): Likewise.
23136 (eliminate_not_pairs): Likewise.
23137 (undistribute_ops_list): Likewise.
23138 (eliminate_redundant_comparison): Likewise.
23139 (update_range_test): Likewise.
23140 (optimize_range_tests_var_bound): Likewise.
23141 (optimize_vec_cond_expr): Likewise.
23142 (rewrite_expr_tree): Likewise.
23143 (rewrite_expr_tree_parallel): Likewise.
23144 (linearize_expr): Likewise.
23145 (break_up_subtract): Likewise.
23146 (linearize_expr_tree): Likewise.
23147 (attempt_builtin_powi): Likewise.
23148 (attempt_builtin_copysign): Likewise.
23149 (transform_stmt_to_copy): Likewise.
23150 (transform_stmt_to_multiply): Likewise.
23151 (dump_ops_vector): Likewise.
23152 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23153 (print_scc): Likewise.
23154 (set_ssa_val_to): Likewise.
23155 (visit_reference_op_store): Likewise.
23156 (visit_use): Likewise.
23157 (sccvn_dom_walker::before_dom_children): Likewise.
23158 (run_scc_vn): Likewise.
23159 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
23161 (expr_hash_elt::print): Likewise.
23162 (const_and_copies::pop_to_marker): Likewise.
23163 (const_and_copies::record_const_or_copy_raw): Likewise.
23164 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
23165 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
23166 (dump_predicates): Likewise.
23167 (find_uninit_use): Likewise.
23168 (warn_uninitialized_phi): Likewise.
23169 (pass_late_warn_uninitialized::execute): Likewise.
23170 * tree-ssa.c (verify_vssa): Likewise.
23171 (verify_ssa): Likewise.
23172 (maybe_optimize_var): Likewise.
23173 * tree-vrp.c (dump_value_range): Likewise.
23174 (dump_all_value_ranges): Likewise.
23175 (dump_asserts_for): Likewise.
23176 (register_edge_assert_for_2): Likewise.
23177 (vrp_visit_cond_stmt): Likewise.
23178 (vrp_visit_switch_stmt): Likewise.
23179 (vrp_visit_stmt): Likewise.
23180 (vrp_visit_phi_node): Likewise.
23181 (simplify_cond_using_ranges_1): Likewise.
23182 (fold_predicate_in): Likewise.
23183 (evrp_dom_walker::before_dom_children): Likewise.
23184 (evrp_dom_walker::push_value_range): Likewise.
23185 (evrp_dom_walker::pop_value_range): Likewise.
23186 (execute_early_vrp): Likewise.
23188 2017-05-16 Richard Biener <rguenther@suse.de>
23190 * dwarf2out.c (loc_list_from_tree_1): Do not create
23191 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
23193 2017-05-16 Richard Biener <rguenther@suse.de>
23195 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
23197 (note_variable_value_in_expr): If we resolved the decl ref
23198 do not push to the stack.
23200 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
23202 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
23203 operations in fast-math mode.
23204 (vaddq_f16): Likewise.
23205 (vmul_f16): Likewise.
23206 (vmulq_f16): Likewise.
23207 (vsub_f16): Likewise.
23208 (vsubq_f16): Likewise.
23209 * config/arm/neon.md (add<mode>3): New.
23211 (fma:<VH:mode>3): New. Also remove outdated comment.
23214 2017-05-16 Martin Liska <mliska@suse.cz>
23218 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23219 (odr_types_equivalent_p): Likewise.
23221 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23223 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23225 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23228 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23229 non-interunit SSE move alternatives with '?'.
23230 (zero-extendsidi peephole2): New peephole to skip intermediate
23231 general register in SSE zero-extend sequence.
23233 2017-05-15 Jeff Law <law@redhat.com>
23235 * reorg.c (relax_delay_slots): Create a new variable to hold
23236 the temporary target rather than clobbering TARGET_LABEL.
23238 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23239 missing argument to extract_bit_field call.
23240 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23242 2017-05-15 Martin Liska <mliska@suse.cz>
23245 * gcc.c (process_command): Do not allow empty argument of -o option.
23247 2017-05-15 Renlin Li <renlin.li@arm.com>
23249 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23250 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23251 * config/aarch64/constraints.md (Usf): Add long call check.
23252 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23253 (call_value): Likewise.
23254 (sibcall): Likewise.
23255 (sibcall_value): Likewise.
23257 (call_value_insn): New.
23258 (sibcall_insn): Update rtx pattern.
23259 (sibcall_value_insn): Likewise.
23260 (call_internal): Remove.
23261 (call_value_internal): Likewise.
23262 (sibcall_internal): Likewise.
23263 (sibcall_value_internal): Likewise.
23264 (call_reg): Likewise.
23265 (call_symbol): Likewise.
23266 (call_value_reg): Likewise.
23267 (call_value_symbol): Likewise.
23269 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23272 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23274 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23276 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23277 compatible with CCGOCmode and with CCZmode.
23279 2017-05-14 Martin Sebor <msebor@redhat.com>
23281 PR middle-end/77671
23282 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23283 (gimple_fold_builtin_snprintf): Same.
23284 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23285 (gimple_fold_builtin_snprintf): Same.
23286 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23287 of character types.
23288 (is_call_safe): New function.
23289 (try_substitute_return_value): Call it.
23290 (try_simplify_call): New function.
23291 (pass_sprintf_length::handle_gimple_call): Call it.
23293 2017-05-14 Martin Sebor <msebor@redhat.com>
23295 PR middle-end/80669
23296 * builtins.c (expand_builtin_stpncpy): Simplify.
23298 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23300 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23301 * config/i386/i386.h
23302 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23303 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23304 (struct machine_function): Add new members call_ms2sysv,
23305 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23306 (struct machine_frame_state): New fields sp_realigned and
23307 sp_realigned_offset.
23308 * config/i386/i386.c
23309 (enum xlogue_stub): New enum.
23310 (enum xlogue_stub_sets): New enum.
23311 (class xlogue_layout): New class.
23312 (struct ix86_frame): New fields stack_realign_allocate_offset,
23313 stack_realign_offset and outlined_save_offset. Modify comments to
23314 detail stack layout when using out-of-line stubs.
23315 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23316 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23317 -mcall-ms2sysv-xlogues.
23318 (stub_managed_regs): New static variable.
23319 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23320 registers managed by out-of-line stub.
23321 (disable_call_ms2sysv_xlogues): New function.
23322 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23323 m->call_ms2sysv when appropriate and compute frame layout for
23325 (sp_valid_at, fp_valid_at): New inline functions.
23326 (choose_basereg): New function.
23327 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23329 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23330 Use align parameter of choose_baseaddr to generated aligned SSE movs
23332 (pro_epilogue_adjust_stack): Modify to track
23333 machine_frame_state::sp_realigned.
23334 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23335 (ix86_nsaved_sseregs): Likewise.
23336 (ix86_emit_save_regs): Likewise.
23337 (ix86_emit_save_regs_using_mov): Likewise.
23338 (ix86_emit_save_sse_regs_using_mov): Likewise.
23339 (get_scratch_register_on_entry): Likewise.
23340 (gen_frame_set): New function.
23341 (gen_frame_load): Likewise.
23342 (gen_frame_store): Likewise.
23343 (emit_outlined_ms2sysv_save): Likewise.
23344 (emit_outlined_ms2sysv_restore): Likewise.
23345 (ix86_expand_prologue): Modify stack re-alignment code and call
23346 emit_outlined_ms2sysv_save when appropriate.
23347 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23348 parameter rtx_insn *insn, which allows the function to be used to only
23349 generate the notes.
23350 (ix86_expand_epilogue): Modify validity checks of frame and stack
23351 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23352 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23353 * config/i386/predicates.md
23354 (save_multiple): New predicate.
23355 (restore_multiple): Likewise.
23356 * config/i386/sse.md
23357 (save_multiple<mode>): New pattern.
23358 (save_multiple_realign<mode>): Likewise.
23359 (restore_multiple<mode>): Likewise.
23360 (restore_multiple_and_return<mode>): Likewise.
23361 (restore_multiple_leave_return<mode>): Likewise.
23362 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23364 2017-05-14 Julia Koval <julia.koval@intel.com>
23366 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23367 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23368 (__builtin_ia32_xsetbv): New builtins.
23369 * config/i386/i386.c (ix86_expand_special_args_builtin):
23371 (ix86_expand_builtin): Special expand for new intrinsics.
23372 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23373 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23374 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23376 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23378 * cfganal.c (inverted_post_order_compute): Change argument type
23380 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23381 * df-core.c (rest_of_handle_df_initialize): Adjust.
23382 (rest_of_handle_df_finish): Likewise.
23383 (df_analyze_1): Likewise.
23384 (df_analyze): Likewise.
23385 (loop_inverted_post_order_compute): Change argument to be a vec *.
23386 (df_analyze_loop): Adjust.
23387 (df_get_n_blocks): Likewise.
23388 (df_get_postorder): Likewise.
23389 * df.h (struct df_d): Change field to be a vec.
23390 * lcm.c (compute_laterin): Adjust.
23391 (compute_available): Likewise.
23392 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23393 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23394 * tree-ssa-pre.c (compute_antic): Likewise.
23396 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23398 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23399 (depth_first_search::depth_first_search): Change structure init
23400 function to this constructor.
23401 (depth_first_search::add_bb): Rename function to this member.
23402 (depth_first_search::execute): Likewise.
23403 (flow_dfs_compute_reverse_finish): Adjust.
23405 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23407 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23408 (longest_simple_path): Likewise.
23409 * shrink-wrap.c (spread_components): Likewise.
23410 (disqualify_problematic_components): Likewise.
23411 (emit_common_heads_for_components): Likewise.
23412 (emit_common_tails_for_components): Likewise.
23413 (insert_prologue_epilogue_for_components): Likewise.
23415 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23417 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23420 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23422 * df-core.c (df_set_blocks): Start using auto_bitmap.
23423 (df_compact_blocks): Likewise.
23424 * df-problems.c (df_rd_confluence_n): Likewise.
23425 * df-scan.c (df_insn_rescan_all): Likewise.
23426 (df_process_deferred_rescans): Likewise.
23427 (df_update_entry_block_defs): Likewise.
23428 (df_update_exit_block_uses): Likewise.
23429 (df_entry_block_bitmap_verify): Likewise.
23430 (df_exit_block_bitmap_verify): Likewise.
23431 (df_scan_verify): Likewise.
23432 * lra-constraints.c (lra_constraints): Likewise.
23433 (undo_optional_reloads): Likewise.
23434 (lra_undo_inheritance): Likewise.
23435 * lra-remat.c (calculate_gen_cands): Likewise.
23436 (do_remat): Likewise.
23437 * lra-spills.c (assign_spill_hard_regs): Likewise.
23438 (spill_pseudos): Likewise.
23439 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23440 (bitmap_set_subtract_values): Likewise.
23442 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23444 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23445 management with auto_bitmap.
23446 (fix_inter_tick): Likewise.
23447 (fix_recovery_deps): Likewise.
23448 * ira.c (add_store_equivs): Likewise.
23449 (find_moveable_pseudos): Likewise.
23450 (split_live_ranges_for_shrink_wrap): Likewise.
23451 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23452 (rtx_reuse_manager::seen_def_p): Likewise.
23453 (rtx_reuse_manager::set_seen_def): Likewise.
23454 * print-rtl.h (class rtx_reuse_manager): Likewise.
23456 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23458 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23460 (migrate_btr_def): Likewise.
23461 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23462 * df-core.c (loop_post_order_compute): Likewise.
23463 (loop_inverted_post_order_compute): Likewise.
23464 * hsa-common.h: Likewise.
23465 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23466 * init-regs.c (initialize_uninitialized_regs): Likewise.
23467 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23468 (inline_small_functions): Likewise.
23469 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23470 * ira.c (combine_and_move_insns): Likewise.
23471 (build_insn_chain): Likewise.
23472 * loop-invariant.c (find_invariants): Likewise.
23473 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23474 * predict.c (tree_predict_by_opcode): Likewise.
23475 (predict_paths_leading_to): Likewise.
23476 (predict_paths_leading_to_edge): Likewise.
23477 (estimate_loops_at_level): Likewise.
23478 (estimate_loops): Likewise.
23479 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23480 (spread_components): Likewise.
23481 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23482 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23483 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23484 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23485 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23486 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23487 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23488 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23489 (mark_threaded_blocks): Likewise.
23490 (thread_through_all_blocks): Likewise.
23491 * tree-ssa.c (verify_ssa): Likewise.
23492 (execute_update_addresses_taken): Likewise.
23493 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23495 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23497 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23499 (post_order_compute): Likewise.
23500 (inverted_post_order_compute): Likewise.
23501 (pre_and_rev_post_order_compute_fn): Likewise.
23503 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23505 * genrecog.c (int_set::int_set): Explicitly construct our
23506 auto_vec base class.
23507 * vec.h (auto_vec::auto_vec): New constructor.
23509 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23511 * bitmap.h (class auto_bitmap): New constructor taking
23512 bitmap_obstack * argument.
23514 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23516 * bitmap.h (class auto_bitmap): Change type of m_bits to
23517 bitmap_head, and adjust ctor / dtor and member operators.
23519 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23521 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23522 when returned register mode doesn't match original mode.
23524 2017-05-12 Jeff Law <law@redhat.com>
23525 Jakub Jelinek <jakub@redhat.com>
23527 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23528 we look for cc setter after the compare-elim changes.
23529 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23530 within the vector to match what compare-elim now expects.
23531 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23532 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23534 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23535 after the compare-elim changes.
23536 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23537 the vector to match what compare-elim now expects.
23538 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23539 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23540 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23541 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23542 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23544 * config/visium/visium.c (single_set_and_flags): Fix where
23545 we look for cc setter after the compare-elim changes.
23546 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23547 with the vector to match what compare-elim now expects.
23548 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23549 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23550 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23551 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23552 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23553 (neg<mode>2_insn_set_overflow): Likewise.
23555 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23557 PR middle-end/79794
23558 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23559 maybe_expand_insn call, set ops[0].target. If still set after call,
23560 set alt_rtl. Add extra arg to recursive calls.
23561 (extract_bit_field): Add alt_rtl argument. Pass to
23563 * expmed.h (extract_bit_field): Fix prototype.
23564 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23565 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23566 to extract_bit_field_calls.
23567 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23568 Pass alt_rtl to extract_bit_field calls.
23569 * calls.c (store_unaligned_arguments_into_psuedos)
23570 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23571 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23573 * optabs.h (struct expand_operand): Add target bitfield.
23575 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23577 * compare-elim.c (try_eliminate_compare): Canonicalize
23578 operation with embedded compare to
23579 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23580 (set (reg) (operation)].
23582 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23584 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23587 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23588 cost of adding a carry flag for ADC instruction.
23589 [case MINUS]: Ignore the cost of subtracting a carry flag
23590 for SBB instruction.
23592 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23594 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23596 * config/rs6000/bmiintrin.h: New file.
23597 * config/rs6000/bmi2intrin.h: New file.
23598 * config/rs6000/x86intrin.h: New file.
23600 2017-05-12 Jeff Law <law@redhat.com>
23602 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23605 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23607 PR middle-end/80707
23608 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23610 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23612 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23613 early expansion of vector divide builtins.
23614 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23615 builtins identified as having unsigned arguments.
23617 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23619 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23620 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23621 expansion of vector logical operations (and, andc, or, xor,
23624 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23626 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23627 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23629 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23631 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23632 early GIMPLE expansion of vector multiplies.
23634 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23636 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23637 TARGET_HAVE_MOVT conditional.
23638 (movt splitter): Likewise.
23640 2017-05-12 Richard Biener <rguenther@suse.de>
23642 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23643 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23644 Fold all stmts not inplace.
23646 2017-05-12 Richard Biener <rguenther@suse.de>
23648 PR tree-optimization/80713
23649 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23650 inserted_exprs bit for not removed stmts.
23652 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23654 PR middle-end/69921
23655 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
23656 parallelized" attribute for parallelized OpenACC kernels.
23657 * omp-offload.c (execute_oacc_device_lower): Use it.
23659 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
23660 Set "oacc kernels" attribute.
23661 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
23662 parameter. Adjust all users.
23663 (oacc_fn_attrib_kernels_p): Remove function.
23664 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
23665 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
23666 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
23667 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
23668 assert "oacc kernels" attribute is set.
23670 2017-05-11 Carl Love <cel@us.ibm.com>
23672 * config/rs6000/rs6000-c: Add support for built-in functions
23673 vector unsigned char vec_popcnt (vector signed char)
23674 vector unsigned char vec_popcnt (vector unsigned char)
23675 vector unsigned short vec_popcnt (vector signed short)
23676 vector unsigned short vec_popcnt (vector unsigned short)
23677 vector unsigned int vec_popcnt (vector signed int)
23678 vector unsigned int vec_popcnt (vector unsigned int)
23679 vector unsigned long long vec_popcnt (vector signed long long)
23680 vector unsigned long long vec_popcnt (vector unsigned long long)
23681 vector signed long long vec_slo (vector signed long long,
23682 vector signed char)
23683 vector signed long long vec_slo (vector signed long long,
23684 vector unsigned char)
23685 vector unsigned long long vec_slo (vector unsigned long long,
23686 vector signed char)
23687 vector unsigned long long vec_slo (vector unsigned long long,
23688 vector unsigned char)
23689 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23690 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23691 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23692 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23693 * doc/extend.texi: Update the built-in documentation file for the
23694 new built-in functions.
23696 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23698 * attribs.h (sorted_attr_string): Move machine independent
23699 functions for target clone support from the i386 port to common
23700 code. Rename ix86_function_versions to common_function_versions.
23701 Rename make_name to make_unique_name.
23702 (common_function_versions): Likewise.
23703 (make_unique_name): Likewise.
23704 (make_dispatcher_decl): Likewise.
23705 (is_function_default_version): Likewise.
23706 * attribs.c (attr_strcmp): Likewise.
23707 (sorted_attr_string): Likewise.
23708 (common_function_versions): Likewise.
23709 (make_unique_name): Likewise.
23710 (make_dispatcher_decl): Likewise.
23711 (is_function_default_version): Likewise.
23712 * config/i386/i386.c (attr_strcmp): Likewise.
23713 (sorted_attr_string): Likewise.
23714 (ix86_function_versions): Likewise.
23715 (make_name): Likewise.
23716 (make_dispatcher_decl): Likewise.
23717 (is_function_default_version): Likewise.
23718 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23720 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23723 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23724 Account for direct move costs for vec_construct of integer
23727 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23730 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23731 (UNSPEC_STX_ATOMIC): Ditto.
23732 (loaddi_via_sse): New insn.
23733 (storedi_via_sse): Ditto.
23734 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23735 Update corresponding peephole2 patterns.
23736 (atomic_storedi_fpu): Ditto.
23738 2017-05-11 Julia Koval <julia.koval@intel.com>
23740 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23741 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23743 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23744 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23745 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23747 2017-05-11 Nathan Sidwell <nathan@acm.org>
23749 * graphite-poly.c: Include dumpfile.h.
23751 * dumpfle.h (dump_function): Declare here ...
23752 * tree-dump.h (dump_function): ... not here.
23753 * dumpfile.c: #include tree-cfg.h.
23754 (dump_function): Move here from ...
23755 * tree-dump.c (dump_function): ... here.
23756 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23757 * graphite-poly.c: Don't include tree-dump.h.
23758 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23759 * print-tree.c: Likewise.
23760 * stor-layout.c: Likewise.
23761 * tree-nested.c: Likewise.
23763 * dumpfile.c (dump_start): Use TDF_FLAGS.
23764 (dump_enable_all): Fix TDF_KIND check thinko.
23766 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23768 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23769 array entries to represent two legal parameterizations of the
23770 overloaded __builtin_cmpb function, as represented by the
23771 P6_OV_BUILTIN_CMPB constant.
23772 (altivec_resolve_overloaded_builtin): Add special case handling
23773 for the __builtin_cmpb function, as represented by the
23774 P6_OV_BUILTIN_CMPB constant.
23775 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23776 (BU_P6_64BIT_2): New macro.
23777 (BU_P6_OVERLOAD_2): New macro
23778 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23779 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23780 (CMPB): Add overload support to represent both 32-bit and 64-bit
23781 compare-bytes function.
23782 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23783 support for TARGET_CMPB.
23784 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23785 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23786 documentation of the __builtin_cmpb overloaded built-in function.
23788 2017-05-11 Richard Biener <rguenther@suse.de>
23790 PR tree-optimization/80705
23791 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23792 bases are not vectorizable.
23794 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23796 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23797 when counting register pressure.
23799 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23801 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23802 (iv_ca_more_deps): Renamed to ...
23803 (iv_ca_compare_deps): ... this.
23804 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23806 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23808 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23810 (determine_group_iv_costs): ... here.
23811 (find_inv_vars_cb): Record inv var if it's not recorded before.
23813 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23815 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23816 (get_shiftadd_cost): Ditto.
23818 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23820 * tree-ssa-address.c: Include header file.
23821 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23823 (add_to_parts): Refactor.
23824 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23825 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23828 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23830 PR tree-optimization/53090
23831 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23833 (extract_cond_operands): Detect condition with IV on both sides
23834 and return COMP_IV_EXPR_2.
23835 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23836 (rewrite_use_compare): Simplify by removing call to function
23837 extract_cond_operands.
23839 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23841 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23842 (extract_cond_operands): Detect condition comparing against non-
23843 invariant bound and return appropriate enum value.
23844 (find_interesting_uses_cond): Update use of extract_cond_operands.
23845 Handle its return value accordingly.
23846 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23848 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23850 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23851 nonlinear iv_use computation in loop invariant sensitive way.
23853 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23855 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23856 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23858 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23860 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23861 (dump_cand): Support iv_cand.inv_exprs.
23862 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23864 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23867 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23869 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23871 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23872 as local function. Include necessary header files.
23873 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23875 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23877 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23879 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23881 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23882 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23883 RSHIFT_EXPR and BIT_NOT_EXPR.
23885 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23887 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23888 (adjust_setup_cost): New parameter supporting round up adjustment.
23889 (struct address_cost_data): Delete.
23890 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23891 (split_address_cost, ptr_difference_cost): Delete.
23892 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23893 (struct ainc_cost_data): New struct.
23894 (get_address_cost_ainc): New function.
23895 (get_address_cost, get_computation_cost): Reimplement.
23896 (determine_group_iv_cost_address): Record inv_expr for all uses of
23898 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23899 (iv_ca_has_deps): Reimplemented to ...
23900 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23902 (iv_ca_extend): Call iv_ca_more_deps.
23904 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23906 * tree-ssa-address.c (struct mem_address): Move to header file.
23907 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23908 * tree-ssa-address.h (struct mem_address): Move from C file.
23909 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23911 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23913 * tree-affine.h (aff_combination_type): New interface.
23914 (aff_combination_zero_p): Remove static.
23915 (aff_combination_const_p): New interface.
23916 (aff_combination_singleton_var_p): New interfaces.
23918 2017-05-11 Richard Biener <rguenther@suse.de>
23920 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23921 Skip unreachable blocks and destinations.
23922 (eliminate): Move stmt removal and fixup ...
23923 (fini_eliminate): ... here. Skip inserted exprs.
23924 (pass_pre::execute): Move fini_pre after fini_eliminate.
23925 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23926 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23927 PRE to get rid of dead code that has invalid SSA form and
23928 split critical edges again.
23930 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23932 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23934 2017-05-11 Richard Biener <rguenther@suse.de>
23936 * passes.c (execute_function_todo): Verify loops if they are
23937 said to be up-to-date.
23938 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23939 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23941 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23944 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23945 handle calling assemble_external ourself.
23948 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23949 modes with zero size. Enhance comment.
23951 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23953 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23954 built-ins for vec_xl and vec_xst with short and char pointer
23957 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23959 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23960 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23961 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23962 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23963 (_mm_maskz_min_round_ss): New intrinsics.
23964 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23965 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23966 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
23967 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
23968 (__builtin_ia32_minss_mask_round): New builtins.
23969 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23970 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23971 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
23973 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
23974 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
23976 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
23979 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23981 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
23982 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
23983 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
23984 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
23985 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
23986 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
23987 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
23988 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23989 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23990 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
23991 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
23992 (__builtin_ia32_mulss_mask_round): New builtins.
23993 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23994 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23995 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
23997 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
23998 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24000 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24003 2017-05-10 Julia Koval <julia.koval@intel.com>
24005 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24006 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24007 (_mm256_setr_m128i): New intrinsics.
24009 2017-05-10 Julia Koval <julia.koval@intel.com>
24011 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24012 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24013 (_mm_maskz_rcp14_ss): New intrinsics.
24014 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24015 (__builtin_ia32_rcp14ss_mask): New builtins.
24016 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24018 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24020 PR tree-optimization/51513
24021 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24022 (assert_unreachable_fallthru_edge_p): Use it.
24023 (group_case_labels_stmt): Likewise.
24024 * tree-cfg.h: Prototype it.
24025 * stmt.c: Include cfghooks.h and tree-cfg.h.
24026 (emit_case_dispatch_table) <gap_label>: New local variable.
24027 Use it to fill dispatch table gaps.
24028 Test for default_label before updating probabilities.
24029 (expand_case) <default_label>: Remove unneeded initialization.
24030 Test for unreachable default case statement and remove its edge.
24031 Set default_label accordingly.
24032 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24034 2017-05-10 Carl Love <cel@us.ibm.com>
24036 * config/rs6000/rs6000-c: Add support for built-in functions
24037 vector signed char vec_neg (vector signed char)
24038 vector signed short int vec_neg (vector short int)
24039 vector signed int vec_neg (vector signed int)
24040 vector signed long long vec_neg (vector signed long long)
24041 vector float vec_neg (vector float)
24042 vector double vec_neg (vector double)
24043 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24045 * config/rs6000/altivec.h: Add define for vec_neg
24046 * doc/extend.texi: Update the built-in documentation for the
24047 new built-in functions.
24049 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24051 PR tree-optimization/77644
24052 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24054 2017-05-10 Nathan Sidwell <nathan@acm.org>
24056 * dumpfile.h (TDI_lang_all): New.
24057 (TDF_KIND): New. Renumber others
24058 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24060 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24062 (get_dump_file_name): Adjust suffix generation.
24063 (dump_enable_all): Use TDF_KIND.
24064 * doc/invoke.texi (-fdump-lang-all): Document.
24066 * dumpfile.h: Tabify.
24068 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24071 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24072 Move member access before delete.
24074 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24076 * tree-inline.c (expand_call_inline): Split block at stmt
24079 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24082 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24083 are now unused after splitting mov{sf,sd}_hardfloat.
24084 (f32_lr2): Likewise.
24085 (f32_lm): Likewise.
24086 (f32_lm2): Likewise.
24087 (f32_li): Likewise.
24088 (f32_li2): Likewise.
24089 (f32_lv): Likewise.
24090 (f32_sr): Likewise.
24091 (f32_sr2): Likewise.
24092 (f32_sm): Likewise.
24093 (f32_sm2): Likewise.
24094 (f32_si): Likewise.
24095 (f32_si2): Likewise.
24096 (f32_sv): Likewise.
24097 (f32_dm): Likewise.
24098 (f32_vsx): Likewise.
24099 (f32_av): Likewise.
24100 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24101 For movsf, order stores so the VSX stores occur before the GPR
24102 store which encourages the register allocator to use a traditional
24103 FPR instead of a GPR. For movsd, order the stores so that the GPR
24104 store comes before the VSX stores to allow the power6 to work.
24105 This is due to the power6 not having a 32-bit integer store
24106 instruction from a FPR.
24107 (movsf_hardfloat): Likewise.
24108 (movsd_hardfloat): Likewise.
24110 2017-05-09 Martin Sebor <msebor@redhat.com>
24112 PR translation/80280
24113 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24116 PR translation/80280
24117 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24118 data member added in r247778.
24119 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24121 2017-05-09 Nathan Sidwell <nathan@acm.org>
24123 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24125 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24128 2017-05-09 Marek Polacek <polacek@redhat.com>
24130 * doc/invoke.texi: Fix typo.
24132 2017-05-09 Richard Biener <rguenther@suse.de>
24134 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24135 (vrp_val_is_min): Likewise.
24136 (set_value_range_to_value): Likewise.
24137 (set_value_range_to_nonnegative): Likewise.
24138 (gimple_assign_nonzero_p): Likewise.
24139 (gimple_stmt_nonzero_p): Likewise.
24140 (vrp_int_const_binop): Likewise. Remove unreachable case.
24141 (adjust_range_with_scev): Adjust comments.
24142 (compare_range_with_value): Likewise.
24143 (extract_range_from_phi_node): Likewise.
24144 (test_for_singularity): Likewise.
24146 2017-05-09 Richard Biener <rguenther@suse.de>
24148 * tree-vrp.c (get_single_symbol): Add assert that we don't
24149 get overflowed constants as invariant part.
24150 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24151 checks. Use wi::cmp instead of recursing for integer constants.
24152 (compare_values): Just ignore whether we assumed undefined
24153 overflow instead of failing the compare.
24154 (extract_range_for_var_from_comparison_expr): Add comment before the
24155 TREE_NO_WARNING sets.
24156 (test_for_singularity): Likewise.
24157 (extract_range_from_comparison): Do not disable optimization
24158 when we assumed undefined overflow.
24159 (extract_range_basic): Remove init of unused var.
24161 2017-05-09 Richard Biener <rguenther@suse.de>
24163 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
24164 (extract_range_from_multiplicative_op_1): Adjust.
24165 (extract_range_from_binary_expr_1): Use int_const_binop.
24167 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
24170 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
24171 rs6000_store_data_bypass_p in seven define_bypass directives and
24172 in several comments.
24173 * config/rs6000/rs6000-protos.h: Add prototype for
24174 rs6000_store_data_bypass_p function.
24175 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
24176 function implements slightly different (rs6000-specific) semantics
24177 than store_data_bypass_p, returning false rather than aborting
24178 with assertion error when arguments do not satisfy the
24179 requirements of store data bypass.
24180 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
24181 rs6000_store_data_bypass_p.
24183 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
24185 * config/xtensa/xtensa-protos.h
24186 (xtensa_initial_elimination_offset): New declaration.
24187 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24188 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
24189 macro definition, add case for FRAME_POINTER_REGNUM when
24190 FRAME_GROWS_DOWNWARD.
24191 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
24192 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
24193 xtensa_initial_elimination_offset.
24195 2017-05-08 Nathan Sidwell <nathan@acm.org>
24197 * doc/invoke.texi: Alphabetize -fdump options.
24199 2017-05-08 Martin Sebor <msebor@redhat.com>
24201 PR translation/80280
24202 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
24204 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
24206 * target.def (compute_frame_layout): New optional target hook.
24207 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
24208 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
24209 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
24211 * reload1.c (verify_initial_elim_offsets): Likewise.
24212 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
24213 (use_simple_return_p): Call arm_compute_frame_layout if needed.
24214 (arm_get_frame_offsets): Split up into this ...
24215 (arm_compute_frame_layout): ... and this function.
24217 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24219 * config/aarch64/constraints.md (Usa): New constraint.
24220 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24222 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24224 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24225 with_multilib_list after it has been checked.
24227 2017-05-08 Richard Biener <rguenther@suse.de>
24229 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24230 (bitmap_set_subtract_values): Likewise.
24232 2017-05-08 Richard Biener <rguenther@suse.de>
24234 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24235 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24237 (gimple_stmt_nonzero_warnv_p): Rename to ...
24238 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24240 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24241 (extract_range_basic): Adjust, do not disable propagation on
24242 strict overflow sensitive simplification.
24243 (vrp_visit_cond_stmt): Likewise.
24245 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24247 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24248 body size unconditionally.
24250 2017-05-07 Jeff Law <law@redhat.com>
24253 2017-05-06 Jeff Law <law@redhat.com>
24254 PR tree-optimization/78496
24255 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24258 PR tree-optimization/78496
24259 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24260 (simplify_stmt_using_ranges): Call it.
24261 (vrp_dom_walker::before_dom_children): Extract equivalences
24262 from an ASSERT_EXPR with an equality comparison against a
24265 2017-05-06 Jeff Law <law@redhat.com>
24267 PR tree-optimization/78496
24268 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24271 PR tree-optimization/78496
24272 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24273 (simplify_stmt_using_ranges): Call it.
24274 (vrp_dom_walker::before_dom_children): Extract equivalences
24275 from an ASSERT_EXPR with an equality comparison against a
24278 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24280 * lra-constraints.c (lra_copy_reg_equiv): New function.
24281 (split_reg): Use it to copy equivalence information from the
24282 original register to the spill register.
24284 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24286 PR rtl-optimization/75964
24287 * simplify-rtx.c (simplify_const_relational_operation): Remove
24288 invalid handling of comparisons of integer ABS.
24290 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24292 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24293 initialize to zero.
24294 (init_regs): Remove declaration.
24295 (function_arg_advance_32): Initialize error_p as boolean variable.
24297 2017-05-05 Nathan Sidwell <nathan@acm.org>
24299 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24300 lines. Use for (;;).
24302 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24304 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24305 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24306 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24307 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24308 VF=2 that require versioning.
24310 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24312 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24315 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24317 * diagnostic.h (diagnostic_override_option_index): Convert from
24318 macro to inline function.
24320 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24322 * diagnostic.c (last_module_changed_p): New function.
24323 (set_last_module): New function.
24324 (diagnostic_report_current_module): Convert macro usage to
24325 the above functions.
24326 * diagnostic.h (diagnostic_context::last_module): Strengthen
24327 from const line_map * to const line_map_ordinary *.
24328 (diagnostic_last_module_changed): Delete macro.
24329 (diagnostic_set_last_module): Delete macro.
24331 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24333 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24334 with diagnostic_report_diagnostic.
24335 (diagnostic_n_impl_richloc): Likewise.
24336 * diagnostic.h (report_diagnostic): Delete macro.
24337 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24338 with diagnostic_report_diagnostic.
24339 * substring-locations.c (format_warning_va): Likewise.
24341 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24343 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24344 save/restor of format_spec. Move option-printing code to...
24345 (print_option_information): ...this new function, and
24346 reimplement by simply printing to the pretty_printer,
24347 rather than appending to the format string.
24349 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24351 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24352 handling logic into...
24353 (update_effective_level_from_pragmas): ...this new function.
24355 2017-05-04 Andrew Waterman <andrew@sifive.com>
24357 * config/riscv/riscv.opt (mstrict-align): New option.
24358 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24359 (SLOW_UNALIGNED_ACCESS): Define.
24360 (riscv_slow_unaligned_access): Declare.
24361 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24363 (riscv_slow_unaligned_access): New variable.
24364 (rocket_tune_info): Set slow_unaligned_access to true.
24365 (optimize_size_tune_info): Set slow_unaligned_access to false.
24366 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24367 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24368 (riscv_option_override): Set riscv_slow_unaligned_access.
24369 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24371 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24373 * config/riscv/riscv.md: Unify indentation.
24375 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24380 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24382 (extendsi<mode>2): Add support for doing sign extension via
24383 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24384 don't have ISA 3.0 instructions.
24385 (extendsi<mode>2 splitter): Likewise.
24386 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24387 generate the normal insns since SImode can now go in vector
24388 registers. Disallow the special UNSPECs needed for previous
24389 machines to hide SImode being used. Add new insns
24390 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24391 (fix_trunc<mode>si2_stfiwx): Likewise.
24392 (fix_trunc<mode>si2_internal): Likewise.
24393 (fixuns_trunc<mode>si2): Likewise.
24394 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24395 (fctiw<u>z_<mode>_smallint): Likewise.
24396 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24397 of floating point to 32-bit integer from doing a direct move to
24398 the GPR registers to do a store.
24399 (fctiwz_<mode>): Break long line.
24401 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24403 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24404 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24405 (addr_list, addr_offset_valid_p): New.
24406 (split_address_groups): Check offset validity with above function.
24407 (gt-tree-ssa-loop-ivopts.h): Include header file.
24409 2017-05-05 Nathan Sidwell <nathan@acm.org>
24411 * config.gcc (arm*-*-*): Add missing 'fi'.
24413 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24415 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24416 included in -fopt-info does not matter.
24417 * doc/optinfo.texi (-fopt-info): Fix description of default
24418 behavour. Explicitly say order of options included in -fopt-info
24421 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24423 * config.gcc: Allow combinations of aprofile and rmprofile values for
24424 --with-multilib-list.
24425 * config/arm/t-multilib: New file.
24426 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24427 variables. Remove setting of ISA and floating-point ABI in
24428 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24429 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24430 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24431 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24433 * config/arm/t-rmprofile: Likewise except for the matches changes.
24434 * doc/install.texi (--with-multilib-list): Document the combination of
24435 aprofile and rmprofile values and warn about pitfalls in doing that.
24437 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24439 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24440 (movdi_aarch64): Likewise.
24442 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24444 PR tree-optimization/80632
24445 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24447 (build_arrays): Initialize it for virtual phis.
24448 (fix_phi_nodes): Use it for virtual phis.
24450 PR tree-optimization/80558
24451 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24452 [x, y] op z into [x op, y op z] for op & or | if conditions
24455 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24456 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24459 * config/arm/arm.md (use_literal_pool): Remove.
24460 (64-bit immediate split): No longer takes cost into consideration
24461 if arm_disable_literal_pool is enabled.
24462 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24463 used when arm_disable_literal_pool is enabled.
24464 (arm_max_const_double_inline_cost): Remove use of
24465 arm_disable_literal_pool.
24466 (push_minipool_fix): Add assert.
24467 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24468 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24469 (no_literal_pool_sf_immediate): New.
24471 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24473 PR tree-optimization/80613
24474 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24475 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24477 2017-05-05 Richard Biener <rguenther@suse.de>
24479 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24481 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24483 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24484 of this flag from insn conditions due to removal from r247495.
24486 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24488 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24490 (arm_early_store_addr_dep_ptr): Likewise.
24491 * config/arm/aarch-common-protos.h
24492 (arm_early_load_addr_dep_ptr): Add prototype.
24493 (arm_early_store_addr_dep_ptr): Likewise.
24494 * config/arm/cortex-a53.md: Add new bypasses.
24496 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24498 * tree.c (next_type_uid): Change type to unsigned.
24499 (type_hash_canon): Decrement back next_type_uid if
24500 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24501 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24504 2017-05-04 Martin Sebor <msebor@redhat.com>
24506 * builtins.c: Fix a trivial typo in a comment.
24508 PR middle-end/79234
24509 * builtins.c (check_sizes): Adjust to handle reading past the end.
24510 Avoid printing excessive upper bound of ranges. Use %E to print
24511 tree nodes instead of converting them to %wu.
24512 (expand_builtin_memchr): New function.
24513 (compute_dest_size): Rename...
24514 (compute_objsize): ...to this.
24515 (expand_builtin_memcpy): Adjust.
24516 (expand_builtin_mempcpy): Adjust.
24517 (expand_builtin_strcat): Adjust.
24518 (expand_builtin_strcpy): Adjust.
24519 (check_strncat_sizes): Adjust.
24520 (expand_builtin_strncat): Adjust.
24521 (expand_builtin_strncpy): Adjust and simplify.
24522 (expand_builtin_memset): Adjust.
24523 (expand_builtin_bzero): Adjust.
24524 (expand_builtin_memcmp): Adjust.
24525 (expand_builtin): Handle memcmp.
24526 (maybe_emit_chk_warning): Check strncat just once.
24528 2017-05-04 Martin Sebor <msebor@redhat.com>
24530 PR preprocessor/79214
24531 PR middle-end/79222
24532 PR middle-end/79223
24533 * builtins.c (check_sizes): Add inlining context and issue
24534 warnings even when -Wno-system-headers is set.
24535 (check_strncat_sizes): Same.
24536 (expand_builtin_strncat): Same.
24537 (expand_builtin_memmove): New function.
24538 (expand_builtin_stpncpy): Same.
24539 (expand_builtin): Handle memmove and stpncpy.
24541 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24543 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24544 which is not used any more.
24546 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24548 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24550 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24552 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24553 (cortexa53_tunings): Likewise.
24554 (cortexa57_tunings): Likewise.
24555 (cortexa72_tunings): Likewise.
24556 (cortexa73_tunings): Likewise.
24558 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24560 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24561 Set loop alignment to 8.
24563 2017-05-04 Martin Sebor <msebor@redhat.com>
24565 PR translation/80280
24566 * builtins.c (expand_builtin_object_size): Add missing quoting to
24567 %D and like directives.
24568 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24569 (hsa_type_for_tree_type): Same.
24570 (verify_function_arguments): Same.
24571 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24572 * varasm.c (get_section): Same.
24575 2017-05-04 Martin Sebor <msebor@redhat.com>
24577 PR translation/80280
24578 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24580 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24582 * config/aarch64/aarch64.c (generic_addrcost_table):
24583 Change HI/TI mode setting.
24585 2017-05-04 Martin Jambor <mjambor@suse.cz>
24587 PR tree-optimization/80622
24588 * tree-sra.c (comes_initialized_p): New function.
24589 (build_accesses_from_assign): Only set write lazily when
24590 comes_initialized_p is false.
24591 (analyze_access_subtree): Use comes_initialized_p.
24592 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24593 instead of testing for PARM_DECL.
24595 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24597 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24598 constraint on operand 0 to allow more general addressing modes.
24599 Adjust output template.
24600 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24602 * config/aarch64/aarch64-protos.h
24603 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24604 * config/aarch64/constraints.md (Dp): New address constraint.
24605 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24608 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24610 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24611 update use of estimate_ipcp_clone_size_and_time.
24612 (estimate_local_effects): Update use of
24613 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24614 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24615 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24616 Return nonspecialized time.
24618 2017-05-04 Richard Biener <rguenther@suse.de>
24620 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24621 for the last VUSE which def dominates the PHI. Directly call
24623 (get_continuation_for_phi_1): Remove.
24625 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24627 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24628 to explain the use of truncating division. Cap the number of
24629 iterations to the maximum given by nb_iterations_upper_bound,
24632 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24634 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24635 * configure: Regenerate.
24636 * config.in: Regenerate.
24637 * config/i386/driver-mingw32.c: new file.
24638 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24639 * config.host: Link driver-mingw32.o on MinGW host.
24640 * doc/install.texi: Document new --enable-mingw-wildcard configure
24643 2017-05-04 Marek Polacek <polacek@redhat.com>
24645 PR tree-optimization/80612
24646 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24648 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24649 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24651 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24652 (movt splitter): Likewise.
24653 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24654 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
24655 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
24656 block for Thumb-1 with MOVT.
24657 (thumb2_legitimate_address_p): Move code block ...
24658 (can_avoid_literal_pool_for_label_p): ... into this new function.
24659 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
24661 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
24662 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
24663 "M-profile targets with the MOVT instruction".
24665 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24667 * config/arm/arm-builtins.c (arm_init_builtins): Rename
24668 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
24669 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24671 2017-05-04 Martin Liska <mliska@suse.cz>
24673 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24674 variable cond_code.
24676 2017-05-04 Richard Biener <rguenther@suse.de>
24678 * tree.c (array_at_struct_end_p): Handle arrays at struct
24679 end with flexarrays more conservatively. Refactor and treat
24680 arrays of arrays or aggregates more strict. Fix
24681 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24682 * tree.c (array_at_struct_end_p): Adjust prototype.
24683 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24684 * gimple-fold.c (get_range_strlen): Likewise.
24685 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24687 2017-05-04 Richard Biener <rguenther@suse.de>
24689 PR tree-optimization/31130
24690 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24692 (supports_overflow_infinity): Likewise.
24693 (is_negative_overflow_infinity): Likewise.
24694 (is_positive_overflow_infinity): Likewise.
24695 (is_overflow_infinity): Likewise.
24696 (stmt_overflow_infinity): Likewise.
24697 (overflow_infinity_range_p): Likewise.
24698 (usable_range_p): Remove as always returning true.
24699 (make_overflow_infinity): Remove.
24700 (negative_overflow_infinity): Likewise.
24701 (positive_overflow_infinity): Likewise.
24702 (avoid_overflow_infinity): Likewise.
24703 (set_value_range): Adjust accordingly.
24704 (set_value_range_to_nonnegative): Likewise, remove now unused
24705 overflow_infinity arg.
24706 (vrp_operand_equal_p): Adjust.
24707 (update_value_range): Likewise.
24708 (range_int_cst_singleton_p): Likewise.
24709 (operand_less_p): Likewise.
24710 (compare_values_warnv): Likewise.
24711 (extract_range_for_var_from_comparison_expr): Likewise.
24712 (vrp_int_const_binop): Likewise.
24713 (zero_nonzero_bits_from_vr): Likewise.
24714 (extract_range_from_multiplicative_op_1): Likewise.
24715 (extract_range_from_binary_expr_1): Likewise.
24716 (extract_range_from_unary_expr): Likewise.
24717 (extract_range_from_comparison): Likewise.
24718 (extract_range_basic): Likewise.
24719 (adjust_range_with_scev): Likewise.
24720 (compare_ranges): Likewise.
24721 (compare_range_with_value): Likewise.
24722 (dump_value_range): Likewise.
24723 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24725 (simplify_cond_using_ranges): Adjust.
24727 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24729 * brig-builtins.def: Added a builtin for class_f64.
24730 * builtin-types.def: Added a builtin type needed by class_f64.
24732 2017-05-03 Jason Merrill <jason@redhat.com>
24734 * timevar.def: Add TV_CONSTEXPR.
24736 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24738 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24740 2017-05-03 Martin Jambor <mjambor@suse.cz>
24742 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24743 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24744 * ipa-cp.c (ipcp_propagate_stage): Do not call
24745 ipa_update_after_lto_read.
24746 * ipa-inline.c (ipa_inline): Likewise.
24748 2017-05-03 Martin Jambor <mjambor@suse.cz>
24750 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24751 tag. Added a default constructor and a destructor.
24752 (ipa_edge_args_sum_t): New class;
24753 (ipa_edge_args_sum): Declare.
24754 (ipa_edge_args_vector): Remove declaration.
24755 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24756 (ipa_free_edge_args_substructures): Remove declaration.
24757 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24758 (ipa_edge_args_info_available_for_edge_p): Likewise.
24759 * ipa-prop.c (ipa_edge_args_vector): Removed.
24760 (edge_removal_hook_holder): Likewise.
24761 (edge_duplication_hook_holder): Likewise.
24762 (ipa_edge_args_sum): New variable.
24763 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24764 ipa_edge_args_vector.
24765 (ipa_free_edge_args_substructures): Likewise.
24766 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24767 ipa_edge_args_vector.
24768 (ipa_edge_removal_hook): Turned into method
24769 ipa_edge_args_sum_t::remove.
24770 (ipa_edge_duplication_hook): Turned into method
24771 ipa_edge_args_sum_t::duplicate.
24772 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24773 registering edge hooks.
24774 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24775 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24776 ipa_edge_args_sum instead of ipa_edge_args_vector.
24777 * ipa-profile.c (ipa_profile): Likewise.
24779 2017-05-03 Martin Jambor <mjambor@suse.cz>
24781 * symbol-summary.h (function_summary): New method exists.
24782 (function_summary::symtab_removal): Deallocate through release.
24783 (call_summary): New class.
24784 (gt_ggc_mx): New overload.
24785 (gt_pch_nx): Likewise.
24786 (gt_pch_nx): Likewise.
24788 2017-05-03 Jeff Law <law@redhat.com>
24790 PR tree-optimization/78496
24791 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24792 from simplify_cond_using_ranges. Split off code to walk
24793 backwards through casts into ...
24794 (simplify_cond_using_ranges_2): New function.
24795 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24796 (execute_vrp): After identifying jump threads, call
24797 simplify_cond_using_ranges_2.
24799 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24802 * ipa-inline.h (inline_summary): Add ctor.
24803 (create_ggc): Do not use ggc_cleared_alloc.
24805 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24806 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24808 * gcc.c (handle_braces): Support escaping in switch matching
24810 * doc/invoke.texi (Spec Files): Document it.
24811 Remove superfluous @code markup in items.
24813 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24815 * diagnostic-show-locus.c (struct column_range): New struct.
24816 (get_affected_columns): New function.
24817 (get_printed_columns): New function.
24818 (struct correction): New struct.
24819 (correction::ensure_capacity): New function.
24820 (correction::ensure_terminated): New function.
24821 (struct line_corrections): New struct.
24822 (line_corrections::~line_corrections): New dtor.
24823 (line_corrections::add_hint): New function.
24824 (layout::print_trailing_fixits): Reimplement in terms of the new
24826 (selftest::test_overlapped_fixit_printing): New function.
24827 (selftest::diagnostic_show_locus_c_tests): Call it.
24829 2017-05-03 Nathan Sidwell <nathan@acm.org>
24831 Canonicalize canonical type hashing
24832 * tree.h (type_hash_canon_hash): Declare.
24833 * tree.c (type_hash_list, attribute_hash_list): Move into
24834 type_hash_canon_hash.
24835 (build_type_attribute_qual_variant): Break out hash code calc into
24836 type_hash_canon_hash.
24837 (type_hash_canon_hash): New. Generic type hash computation.
24838 (build_range_type_1, build_array_type_1, build_function_type,
24839 build_method_type_directly, build_offset_type, build_complex_type,
24840 make_vector_type): Call it.
24842 2017-05-03 Richard Biener <rguenther@suse.de>
24844 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24845 When all DRs have unknown misaligned do not always peel
24846 when there is a store but apply the same costing model as if
24847 there were only loads.
24849 2017-05-03 Richard Biener <rguenther@suse.de>
24852 PR tree-optimization/80492
24853 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24854 compare_base_decls returning dont-know properly.
24856 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24858 * config/arm/iterators.md (CCSI): New mode iterator.
24859 (arch): New mode attribute.
24860 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24861 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24862 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24863 code iterator for success result mode.
24864 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24865 the corresponding new insn generators.
24867 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24870 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24871 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24873 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24875 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24876 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24877 (DDR_A): Wrap DDR argument in brackets.
24878 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24879 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24880 (DDR_REVERSED_P): Likewise.
24882 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24884 PR tree-optimization/79472
24885 * tree-switch-conversion.c (struct switch_conv_info): Add
24886 contiguous_range and default_case_nonstandard fields.
24887 (collect_switch_conv_info): Compute contiguous_range and
24888 default_case_nonstandard fields, don't clear final_bb if
24889 contiguous_range and only the default case doesn't have the required
24891 (check_all_empty_except_final): Set default_case_nonstandard instead
24892 of failing if contiguous_range and the default case doesn't have empty
24894 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24895 and only the default case doesn't have the required constants. Skip
24897 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24898 if default_case_nonstandard.
24899 (build_constructors): Build constant 1 just once. Assert that default
24900 values aren't inserted in between cases if contiguous_range. Skip
24902 (build_arrays): Skip virtual phis.
24903 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24904 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24905 Handle virtual phis.
24906 (gen_inbound_check): Handle default_case_nonstandard case.
24907 (process_switch): Adjust check_final_bb caller. Call
24908 gather_default_values with the first non-default case instead of
24909 default case if default_case_nonstandard.
24911 2017-05-02 Nathan Sidwell <nathan@acm.org>
24913 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24914 check. Fix formatting.
24916 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24918 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24919 errors when comparing specialized and unspecialized times.
24921 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24923 * diagnostic-show-locus.c
24924 (layout::should_print_annotation_line_p): Make private.
24925 (layout::print_annotation_line): Make private.
24926 (layout::annotation_line_showed_range_p): Make private.
24927 (layout::show_ruler): Make private.
24928 (layout::print_source_line): Make private. Pass in line and
24929 line_width, rather than calling location_get_source_line. Drop
24931 (layout::print_leading_fixits): New method.
24932 (layout::print_any_fixits): Rename to...
24933 (layout::print_trailing_fixits): ...this, and make private.
24934 Don't print newline fixits.
24935 (diagnostic_show_locus): Move logic for printing one row into...
24936 (layout::print_line): ...this new function. Move the
24937 location_get_source_line call and error-handling from
24938 print_source_line to here. Call print_leading_fixits, and rename
24939 print_any_fixits to print_trailing_fixits.
24940 (selftest::test_fixit_insert_containing_newline): Update now that
24941 newlines are partially supported.
24942 (selftest::test_fixit_insert_containing_newline_2): New test.
24943 (selftest::test_fixit_replace_containing_newline): Update comments.
24944 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24945 * edit-context.c (class added_line): New class.
24946 (class edited_line): Describe newline handling in comment.
24947 (edited_line::actually_edited_p): New method.
24948 (edited_line::print_content): Delete redundant decl.
24949 (edited_line::m_predecessors): New field.
24950 (edited_file::print_content): Call edited_line::print_content.
24951 (edited_file::print_diff): Update to support newlines.
24952 (edited_file::print_diff_hunk): Likewise.
24953 (edited_file::print_run_of_changed_lines): New function.
24954 (edited_file::print_diff_line): Convert to...
24955 (print_diff_line): ...this.
24956 (edited_file::get_effective_line_count): New function.
24957 (edited_line::edited_line): Initialize new field m_predecessors.
24958 (edited_line::~edited_line): Clean up m_predecessors.
24959 (edited_line::apply_fixit): Handle newlines.
24960 (edited_line::get_effective_line_count): New function.
24961 (edited_line::print_content): New function.
24962 (edited_line::print_diff_lines): New function.
24963 (selftest::test_applying_fixits_insert_containing_newline): New
24965 (selftest::test_applying_fixits_replace_containing_newline): New
24967 (selftest::insert_line): New function.
24968 (selftest::test_applying_fixits_multiple_lines): Add example of
24970 (selftest::edit_context_c_tests): Call the new tests.
24972 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24974 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
24975 parameter cand. Update dump information.
24976 (get_computation_cost): Update uses.
24978 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24980 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
24981 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
24982 (get_computation_at, rewrite_use_address): Update use of
24983 get_computation_aff.
24985 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24987 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
24988 (get_computation): Delete.
24989 (get_computation_cost): Implement like get_computation_cost_at.
24990 Use get_computation_at.
24991 (get_computation_cost_at): Delete.
24992 (rewrite_use_nonlinear_expr): Use get_computation_at.
24993 (rewrite_use_compare, remove_unused_ivs): Ditto.
24995 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24997 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
24999 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25001 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25002 (ivopts_global_cost_for_size): Rename parameter and update uses.
25003 (iv_ca_recount_cost): Update uses.
25004 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25005 candidates seperately in n_invs and n_cands.
25006 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25008 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25010 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25011 (find_inv_vars_cb): New.
25012 (find_depends): Renamed to ...
25013 (find_inv_vars): ... this.
25014 (add_candidate_1, force_var_cost): Call find_inv_vars.
25015 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25017 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25019 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25020 inv_vars. Add inv_exprs.
25021 (struct iv_cand): Rename depends_on to inv_vars.
25022 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25023 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25024 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25025 (dump_cand): Dump inv_vars.
25026 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25027 (record_invariant, find_depends, add_candidate_1): Ditto.
25028 (set_group_iv_cost, force_var_cost): Ditto.
25029 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25030 (get_computation_cost_at, get_computation_cost): Ditto.
25031 (determine_group_iv_cost_generic): Ditto.
25032 (determine_group_iv_cost_address): Ditto.
25033 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25034 (determine_group_iv_costs): Ditto.
25035 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25036 (iv_ca_set_remove_invariants): Renamed to ...
25037 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25038 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25039 (iv_ca_set_add_invariants): Renamed to ...
25040 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25041 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25042 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25043 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25044 (create_new_ivs): Remove useless dump.
25046 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25048 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25050 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25051 (iv_ca_set_no_cp, create_new_iv): Ditto.
25053 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25055 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25057 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25059 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25060 function tree_check2.
25062 2017-05-02 Martin Liska <mliska@suse.cz>
25064 * doc/gcov.texi: Add missing preposition.
25065 * gcov.c (function_info::function_info): Properly fill up
25066 all member variables.
25068 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25070 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25072 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25074 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25076 2017-05-02 Martin Liska <mliska@suse.cz>
25079 * lto-streamer-in.c (lto_read_tree_1): Remove
25080 LTO_STREAMER_DEBUG.
25081 * lto-streamer.c (struct tree_hash_entry): Likewise.
25082 (struct tree_entry_hasher): Likewise.
25083 (tree_entry_hasher::hash): Likewise.
25084 (tree_entry_hasher::equal): Likewise.
25085 (lto_streamer_init): Likewise.
25086 (lto_orig_address_map): Likewise.
25087 (lto_orig_address_get): Likewise.
25088 (lto_orig_address_remove): Likewise.
25089 * lto-streamer.h: Likewise.
25090 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25091 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25093 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25095 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25096 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25097 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25098 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25099 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25100 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25101 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25102 (mm_maskz_sub_ss): New intrinsics.
25103 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25104 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25105 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25106 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25107 (__builtin_ia32_subss_mask_round): New builtins.
25108 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25109 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25110 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25112 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25113 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25115 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25118 2017-05-02 Martin Jambor <mjambor@suse.cz>
25120 PR tree-optimization/78687
25121 * tree-sra.c (access): New field parent.
25122 (process_subtree_disqualification): New function.
25123 (disqualify_candidate): Call it.
25124 (build_accesses_from_assign): Reset write flag if creating an
25126 (build_access_subtree): Fill in parent field and also prpagate
25127 down grp_write flag.
25128 (create_artificial_child_access): New parameter set_grp_write, set
25129 grp_write to its value.
25130 (propagate_subaccesses_across_link): Also propagate grp_write flag
25132 (propagate_all_subaccesses): Push the closest parent back to work
25133 queue if add_access_to_work_queue returned true.
25135 2017-05-02 Richard Biener <rguenther@suse.de>
25137 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25138 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25139 -fstrict-overflow documentation.
25140 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25141 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25142 flag_strict_overflow.
25143 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25144 * lto-opts.c (lto_write_options): Do not stream it.
25145 * lto-wrapper.c (merge_and_complain): Do not handle it.
25146 * opts.c (default_options_table): Do not set -fstrict-overflow.
25147 (finish_options): Likewise do not clear it when sanitizing.
25148 * simplify-rtx.c (simplify_const_relational_operation): Do not
25149 test flag_strict_overflow.
25151 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25153 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25154 using enabled attribute.
25155 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
25156 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
25157 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
25158 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
25159 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
25160 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
25161 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
25162 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
25163 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
25164 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
25166 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25168 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
25170 2017-05-02 Richard Biener <rguenther@suse.de>
25172 PR tree-optimization/80591
25174 2017-04-10 Richard Biener <rguenther@suse.de>
25176 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25179 2017-05-02 Richard Biener <rguenther@suse.de>
25181 PR tree-optimization/80549
25182 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
25183 (cleanup_tree_cfg_noloop): Create forwarders to known loop
25184 headers if they do not have a preheader.
25186 2017-05-02 Martin Liska <mliska@suse.cz>
25189 * common.opt: Fix typo.
25190 * doc/invoke.texi: Likewise.
25192 2017-05-01 Jan Beulich <jbeulich@suse.com>
25194 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
25195 swapping, add (x,x,m,x,n) alternative.
25197 2017-05-01 Nathan Sidwell <nathan@acm.org>
25199 * calls.c (combine_pending_stack_adjustment_and_call): Remove
25200 unnecessary unadjusted_alignment check.
25202 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25205 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
25207 * gimplify.c (gimplify_cilk_detach): New function.
25208 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
25209 * tree-core.h: Document EXPR_CILK_SPAWN.
25210 * tree.h (EXPR_CILK_SPAWN): Define.
25212 2017-05-01 David Malcolm <dmalcolm@redhat.com>
25214 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
25215 to use new fixit_hint representation, using the "replace" logic.
25216 (get_line_span_for_fixit_hint): Likewise.
25217 (layout::print_any_fixits): Likewise.
25218 (selftest::test_one_liner_many_fixits): Rename to...
25219 (selftest::test_one_liner_many_fixits_1): ...this, and update
25220 comment and expected output to reflect that the multiple fix-it
25221 hints are now consolidated into one insertion.
25222 (selftest::test_one_liner_many_fixits_2): New test.
25223 (selftest::test_diagnostic_show_locus_one_liner): Update for
25225 (selftest::test_fixit_consolidation): Update for fix-it API
25227 * diagnostic.c (print_parseable_fixits): Likewise.
25228 * edit-context.c (edited_line::m_line_events): Convert from
25229 auto_vec <line_event *> to auto_vec <line_event>.
25230 (class line_event): Convert from abstract base class to a concrete
25231 class, taking over the role of replace_event.
25232 (class insert_event): Delete.
25233 (class replace_event): Rename to class line_event. Convert to
25235 (edit_context::add_fixits): Reimplement.
25236 (edit_context::apply_insert): Delete.
25237 (edit_context::apply_replace): Rename to...
25238 (edit_context::apply_fixit): ...this. Convert to half-open range.
25239 (edited_file::apply_insert): Delete.
25240 (edited_file::apply_replace): Rename to...
25241 (edited_file::apply_fixit): ...this.
25242 (edited_line::~edited_line): Drop deletion of events.
25243 (edited_line::apply_insert): Delete.
25244 (edited_line::apply_replace): Rename to...
25245 (edited_line::apply_fixit): ...this. Convert to half-open range.
25246 Update for change to type of m_line_events.
25247 * edit-context.h (edit_context::apply_insert): Delete.
25248 (edit_context::apply_replace): Rename to...
25249 (edit_context::apply_fixit): ...this.
25251 2017-05-01 Martin Sebor <msebor@redhat.com>
25253 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25256 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25259 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25260 __get_cpuid_max returns 0.
25261 (__get_cpuid_count): Ditto.
25263 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25265 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25266 replacement expression is another instance of one of its arguments.
25268 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25271 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25272 check for stack push/pop autoinc.
25273 * config/i386/i386.c (ix86_agi_dependent): Return false
25274 if the only reason why modified_in_p returned true is that
25275 addr is SP based and set_insn is a push or pop.
25277 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25279 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25282 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25285 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25286 (account_size_time): Use two predicates - exec_pred and
25288 (evaluate_conditions_for_known_args): Compute both clause and
25290 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25291 (inline_summary_t::duplicate): Update.
25292 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25294 (compute_inline_parameters): Likewise.
25295 (estimate_edge_size_and_time): Update caluclation of time.
25296 (estimate_node_size_and_time): Compute both time and nonspecialized
25298 (estimate_ipcp_clone_size_and_time): Update.
25299 (inline_merge_summary): Update.
25300 (do_estimate_edge_time): Update.
25301 (do_estimate_edge_size): Update.
25302 (do_estimate_edge_hints): Update.
25303 (inline_read_section, inline_write_summary): Stream both new predicates.
25304 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25306 (compute_inlined_call_time): Cleanup.
25307 (big_speedup_p): Update.
25308 (edge_badness): Update.
25309 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25310 (size_time_entry): Replace predicate by exec_predicate and
25311 nonconst_predicate.
25312 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25313 (estimate_edge_time): Return also nonspec_time.
25314 (reset_edge_growth_cache): Update.
25316 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25318 PR rtl-optimization/80491
25319 * ifcvt.c (noce_process_if_block): When looking for x setter
25320 with missing else_bb, don't check only the insn right before
25321 cond_earliest, but look for the last insn that x is modified in
25322 within the same bb.
25324 PR rtl-optimization/80491
25325 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25327 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25329 PR tree-optimization/80487
25330 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25332 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25334 PR tree-optimization/79697
25335 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25336 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25337 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25339 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25340 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25342 2017-04-28 Martin Sebor <msebor@redhat.com>
25344 PR tree-optimization/80523
25345 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25346 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25348 (maybe_warn, format_directive, parse_directive): Use new functions.
25349 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25351 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25353 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25355 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25357 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25358 target_header_dir): Set correctly.
25359 * configure: Regenerated.
25360 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25361 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25362 instead of SYSTEM_HEADER_DIR.
25364 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25366 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25367 (estimate_local_effects): Likewise.
25368 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25369 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25370 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25371 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25372 do_estimate_edge_time, estimate_edge_time): Likewise.
25373 * ipa-inline-analysis.c (estimate_node_size_and_time,
25374 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25375 (estimate_time_after_inlining): Remove.
25377 2017-04-28 Martin Liska <mliska@suse.cz>
25379 * doc/gcov.texi: Enhance documentation of gcov.
25381 2017-04-28 Martin Liska <mliska@suse.cz>
25383 * doc/gcov.texi: Sort options in alphabetic order.
25384 * doc/gcov-dump.texi: Likewise.
25385 * doc/gcov-tool.texi: Likewise.
25386 * gcov.c (print_usage): Likewise.
25387 * gcov-dump.c (print_usage): Likewise.
25388 * gcov-tool.c (print_merge_usage_message): Likewise.
25389 (print_rewrite_usage_message): Likewise.
25390 (print_overlap_usage_message): Likewise.
25392 2017-04-28 Martin Liska <mliska@suse.cz>
25394 PR gcov-profile/53915
25395 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25397 2017-04-28 Martin Liska <mliska@suse.cz>
25399 PR gcov-profile/79891
25400 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25401 is marked by compiler as living on a line.
25402 (get_cycles_count): Remove usage of the union.
25403 (output_intermediate_file): Likewise.
25404 (find_source): Fix GNU coding style.
25405 (accumulate_line_counts): Remove old non-all block mode.
25406 (output_lines): Remove usage of the union.
25407 * profile.c (output_location): Include all BBs, even if
25408 belonging to a same line (and file) as a previous BB.
25410 2017-04-28 Martin Liska <mliska@suse.cz>
25412 * gcov.c (process_args): Handle new argument 'w'.
25413 (read_graph_file): Assign ID to BBs.
25414 (output_branch_count): Display BB # if verbose flag is set.
25415 (output_lines): Likewise for arcs.
25416 (print_usage): Add '--verbose' option help.
25417 * doc/gcov.texi: Document --verbose (-w) option.
25419 2017-04-28 Martin Liska <mliska@suse.cz>
25421 * gcov.c (struct block_location_info): New struct.
25422 (process_file): Fill up the new structure.
25423 (read_graph_file): Replace usage of encoding by the newly added
25425 (add_line_counts): Likewise.
25426 (accumulate_line_counts): Remove usage of the union.
25427 (function_info::function_info): New function.
25428 (function_info::~function_info): Likewise.
25429 (process_file): Call delete instead of release_function.
25430 (release_function): Release the function.
25431 (release_structures): Call delete instead of release_function.
25432 (solve_flow_graph): Replace usage of num_blocks.
25433 (find_exception_blocks): Likewise.
25434 (output_lines): Fix GNU coding style.
25436 2017-04-28 Martin Liska <mliska@suse.cz>
25439 * coverage.c (coverage_remove_note_file): New function.
25440 * coverage.h: Declare the function.
25441 * toplev.c (finalize): Clean if an error has been seen.
25443 2017-04-28 Martin Liska <mliska@suse.cz>
25445 PR gcov-profile/80031
25446 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25447 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25448 * gcov.c (read_graph_file): Read just number of blocks.
25449 * profile.c (branch_prob): Do not stream 0 flags per a basic
25452 2017-04-28 Martin Liska <mliska@suse.cz>
25454 * gcov-dump.c (tag_*): Add new argument to declarations.
25455 (dump_gcov_file): Likewise.
25456 (tag_blocks): Add and use new argument depth.
25457 (tag_arcs): Likewise.
25458 (tag_lines): Likewise.
25459 (tag_counters): Likewise.
25460 (tag_summary): Likewise.
25461 (dump_working_sets): Use depth to do a proper indentation.
25463 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25466 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25467 * symtab.c (symtab_node::debug_symtab): Move definition here.
25469 2017-04-28 Richard Biener <rguenther@suse.de>
25471 * lto-streamer.h (LTO_major_version): Bump to 7.
25473 2017-04-28 Richard Biener <rguenther@suse.de>
25475 * tree-vrp.c (assert_info): New struct.
25476 (add_assert_info): New helper.
25477 (register_edge_assert_for_2): Refactor to add asserts to a vector
25479 (register_edge_assert_for_1): Likewise.
25480 (register_edge_assert_for): Likewise.
25481 (finish_register_edge_assert_for): New helper actually registering
25482 asserts where live on edge.
25483 (find_conditional_asserts): Adjust.
25484 (find_switch_asserts): Likewise.
25485 (evrp_dom_walker::try_find_new_range): Generalize.
25486 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25488 2017-04-27 Marek Polacek <polacek@redhat.com>
25491 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25492 arg10 and arg11 to itype.
25494 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25496 * doc/extend.texi (Object Size Checking): Improve grammar.
25498 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25501 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25502 that the logic for permitting reciprocal estimates matches that
25505 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25508 * tree.c (type_cache_hasher::equal): Only compare
25509 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25510 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25511 non-aggregate element types.
25512 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25513 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25515 2017-04-27 Richard Biener <rguenther@suse.de>
25517 PR middle-end/80533
25518 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25519 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25520 keeping a reference to a trailing array.
25522 2017-04-27 Richard Biener <rguenther@suse.de>
25524 PR middle-end/80539
25525 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25526 being in loop-closed SSA form conservatively.
25527 (chrec_fold_multiply_poly_poly): Likewise.
25529 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25531 PR middle-end/79665
25532 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25533 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25535 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25538 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25539 (aarch64_function_arg_alignment): Return unsigned int again, but still
25540 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25541 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25542 Don't emit -Wpsabi note.
25543 (aarch64_function_arg_boundary): Likewise.
25544 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25547 2017-04-26 Nathan Sidwell <nathan@acm.org>
25549 * tree.h (crc32_unsigned_n): Declare.
25550 (crc32_unsigned, crc32_unsigned): Make inline.
25551 * tree.c (crc32_unsigned_bits): Replace with ...
25552 (crc32_unsigned_n): ... this.
25553 (crc32_unsigned, crc32_byte): Remove.
25554 (crc32_string): Remove unnecessary braces.
25556 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25558 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25559 * ipa-inline-analysis.c (MAX_TIME): Remove.
25560 (account_size_time): Use sreal for time.
25561 (dump_inline_summary): Update.
25562 (estimate_function_body_sizes): Update.
25563 (estimate_edge_size_and_time): Update.
25564 (estimate_calls_size_and_time): Update.
25565 (estimate_node_size_and_time): Update.
25566 (inline_merge_summary): Update.
25567 (inline_update_overall_summary): Update.
25568 (estimate_time_after_inlining): Update.
25569 (inline_read_section): Update.
25570 (inline_write_summary): Update.
25571 * ipa-inline.c (compute_uninlined_call_time): Update.
25572 (compute_inlined_call_time): Update.
25573 (recursive_inlining): Update.
25574 (inline_small_functions): Update.
25575 (dump_overall_stats): Update.
25576 * ipa-inline.h: Include sreal.h.
25577 (size_time_entry): Turn time to sreal.
25578 (inline_summary): Turn self_time nad time to sreal.
25580 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25582 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25584 (sreal::stream_out, sreal::stream_in): New.
25585 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25587 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25589 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25592 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25595 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25597 (dimode_scalar_chain::compute_convert_gain): Ditto.
25598 (dimode_scalar_chain::make_vector_copies): Ditto.
25599 (dimode_scalar_chain::convert_reg): Ditto.
25600 (dimode_scalar_chain::convert_insn): Ditto.
25601 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25602 (VI248_AVX512BW_1): New mode iterator.
25603 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25604 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25607 2017-04-25 Martin Sebor <msebor@redhat.com>
25609 PR tree-optimization/80497
25610 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25611 constants are representable in HOST_WIDE_INT.
25612 (parse_directive): Ditto.
25614 2017-04-25 Martin Sebor <msebor@redhat.com>
25617 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25618 (new_zero_array): Adjust signature.
25619 (dom_info::dom_init): Used unsigned rather that size_t.
25620 (dom_info::dom_info): Same.
25622 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25623 Jakub Jelinek <jakub@redhat.com>
25626 * config/arm/arm.c: Include gimple.h.
25627 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25628 returns negative, increment ncrn only if it returned positive.
25629 (arm_needs_doubleword_align): Return int instead of bool,
25630 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25631 members, but if there is any such non-FIELD_DECL
25632 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25633 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25634 returns negative, increment nregs only if it returned positive.
25635 (arm_setup_incoming_varargs): Likewise.
25636 (arm_function_arg_boundary): Emit -Wpsabi note if
25637 arm_needs_doubleword_align returns negative, return
25638 DOUBLEWORD_ALIGNMENT only if it returned positive.
25640 2017-04-25 Marek Polacek <polacek@redhat.com>
25643 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25644 first argument to type.
25646 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25649 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25650 type checks to test for compatibility instead of equality.
25652 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25653 Jakub Jelinek <jakub@redhat.com>
25656 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
25658 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
25659 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
25660 the alignment computation, but return their maximum in warn_alignment.
25661 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25662 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
25664 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
25665 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25668 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25670 * config/arc/simdext.md (dmpyh): Fix typo.
25672 2017-04-25 Richard Biener <rguenther@suse.de>
25674 PR tree-optimization/80492
25675 * alias.c (compare_base_decls): Handle registers with asm
25676 specification conservatively.
25677 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25678 compare_base_decls returning dont-know properly.
25680 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25682 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25683 (legitimate_offset_address_p): New function.
25684 (arc_legitimate_address_p): Use above function.
25686 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25688 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25690 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25692 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25693 ACCH registers whenever they are available.
25695 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25697 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25698 double regs fix when not used.
25700 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25702 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25704 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25705 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25707 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25709 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25710 when emitting short ADD instructions.
25712 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25714 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25716 (cmpsi_cc_c_insn): Likewise.
25717 (cbranchsi4_scratch): Compute proper instruction length using
25718 compact_hreg_operand.
25719 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25721 2017-04-25 Richard Biener <rguenther@suse.de>
25723 PR middle-end/80509
25724 * passes.c (pass_manager::pass_manager): Initialize
25725 m_name_to_pass_map.
25727 2017-04-25 Richard Biener <rguenther@suse.de>
25729 PR tree-optimization/79201
25730 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25732 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25735 * config/s390/vector.md: Split MEM->GPR vector moves for
25736 non-s_operand addresses.
25738 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25741 * config/s390/predicates.md (reload_const_wide_int_operand): New
25743 * config/s390/s390.md ("movti"): Remove d/P alternative.
25744 ("movti_bigconst"): New pattern definition.
25746 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25749 * s390-protos.h (s390_expand_cs_hqi): Removed.
25750 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25751 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25752 modes as well as CCZ1mode and CCZmode.
25753 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25754 signature of s390_emit_compare_and_swap.
25755 (s390_expand_cs_hqi): Likewise, make static.
25756 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25757 compare-and-swap, in some cases.
25758 (s390_expand_cs): Wrapper function.
25759 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25761 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25762 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25763 patterns for small and large integers. Forbid symref memory operands.
25764 Move expander to s390.c. Require cc register.
25765 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25766 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25767 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25768 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25769 symref memory operands. Remove CC mode and call s390_match_ccmode
25771 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25773 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25775 * config/s390/s390.md (define_peephole2): New peephole to help
25776 combining the load-and-test pattern with volatile memory.
25778 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25780 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25781 with CCZmode for TARGET_Z196.
25783 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25785 PR rtl-optimization/80501
25786 * combine.c (make_compound_operation_int): Set subreg_code to SET
25787 even for AND with mask of the sign bit of mode.
25789 PR rtl-optimization/80500
25790 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25791 sum's initial value.
25793 2017-04-25 Julian Brown <julian@codesourcery.com>
25794 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25796 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25798 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25800 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25802 2017-04-25 Julian Brown <julian@codesourcery.com>
25803 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25805 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25806 (thunderx2t99_sha): New Reservation.
25808 2017-04-25 Julian Brown <julian@codesourcery.com>
25809 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25811 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25812 type for 1-element load.
25814 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25816 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25818 2017-04-24 Martin Jambor <mjambor@suse.cz>
25820 PR tree-optimization/80293
25821 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25822 char arrays not totally scalarizable if it is false.
25823 (analyze_all_variable_accesses): Pass correct value in the new
25824 parameter. Add a statistics counter.
25826 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25828 PR middle-end/79931
25829 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25831 2017-04-24 Richard Biener <rguenther@suse.de>
25833 PR tree-optimization/80494
25834 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25835 out for complex types.
25837 2017-04-24 Richard Biener <rguenther@suse.de>
25839 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25840 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25841 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25842 (DFS): Adjust and never fail.
25843 (sccvn_dom_walker::fail): Remove.
25844 (sccvn_dom_walker::before_dom_children): Adjust.
25845 (run_scc_vn): Likewise and never fail.
25846 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25847 (pass_fre::execute): Likewise.
25849 2017-04-24 Richard Biener <rguenther@suse.de>
25851 PR tree-optimization/79725
25852 * tree-ssa-sink.c (statement_sink_location): Return whether
25853 failure reason was zero uses. Move that check later.
25854 (sink_code_in_bb): Deal with zero uses by removing the stmt
25857 2017-04-24 Richard Biener <rguenther@suse.de>
25860 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25861 pointer-based references.
25863 2017-04-24 Richard Biener <rguenther@suse.de>
25866 * pass_manager.h (pass_manager::operator new): Remove.
25867 (pass_manager::operator delete): Likewise.
25868 * passes.c (pass_manager::operator new): Remove.
25869 (pass_manager::operator delete): Likewise.
25870 (pass_manager::pass_manager): Zero individual pass members.
25872 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25875 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25876 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25877 Check "XEXP (src, 1)" operand here.
25878 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25879 Check "XEXP (src, 1)" operand here.
25880 (dimode_scalar_chain::make_vector_copies): Detect count register
25881 of a shift instruction. Zero extend count register from QImode
25882 to DImode to satisfy vector shift pattern count operand predicate.
25883 Substitute vector shift count operand with a DImode copy.
25884 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25887 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25889 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25890 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25891 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25892 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25893 (UNSPEC_NOREX_MEM): Remove definition.
25895 2017-04-21 Richard Biener <rguenther@suse.de>
25897 PR tree-optimization/79547
25898 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25899 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25900 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25901 without any constraints.
25903 2017-04-21 Richard Biener <rguenther@suse.de>
25905 PR tree-optimization/78847
25906 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25908 2017-04-21 Richard Biener <rguenther@suse.de>
25910 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25911 (build_distinct_type_copy): Likewise.
25912 (build_variant_type_copy): Likewise.
25913 * tree.c (build_qualified_type): Pass down mem-stat info.
25914 (build_distinct_type_copy): Likewise.
25915 (build_variant_type_copy): Likewise.
25917 2017-04-21 Richard Biener <rguenther@suse.de>
25919 PR tree-optimization/80237
25920 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25922 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25923 for a simplified result.
25925 2016-04-21 Richard Biener <rguenther@suse.de>
25927 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25928 sth as strict as a simple_iv but a chrec without symbols and an
25929 operand defined in the loop we are peeling (and not some subloop).
25930 (propagate_constants_for_unrolling): Propagate all constants.
25932 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25935 * config/i386/i386.c (print_reg): Remove assert for disalowed
25936 regno values, call output_operand_lossage instead.
25938 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25941 * config/i386/constraints.md (Yc): New register constraint.
25942 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25943 Use Yc constraint for alternative 2 of operand 0. Remove
25944 preferred_for_speed attribute.
25946 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25948 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25949 lastprivate clauses in SIMT case.
25951 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25953 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25955 2017-04-20 Richard Biener <rguenther@suse.de>
25957 PR tree-optimization/57796
25958 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25959 as N scalar stores.
25960 (vect_model_load_cost): Cost gathers as N scalar loads.
25962 2017-04-20 Richard Biener <rguenther@suse.de>
25964 * ggc-page.c (ggc_allocated_p): Rename to ...
25965 (safe_lookup_page_table_entry): ... this and return the lookup
25967 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
25969 2017-04-20 Richard Biener <rguenther@suse.de>
25971 PR tree-optimization/80453
25972 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
25973 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
25974 from the conditions.
25975 (vn_phi_eq): Pass them down.
25976 (vn_phi_lookup): Record them.
25977 (vn_phi_insert): Likewise.
25979 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
25981 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
25982 uninitialized variable warning to avoid buffer overrun.
25984 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25987 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
25988 is suppressed for '{ 0 }' in C.
25990 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25992 * BASE-VER: Set to 8.0.0.
25994 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
25996 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
25997 priority .init_array and .fini_array section with SECTION_NOTYPE
26000 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26002 PR middle-end/80423
26003 * tree.h (build_array_type): Add typeless_storage default argument.
26004 * tree.c (type_cache_hasher::equal): Also compare
26005 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26006 (build_array_type): Add typeless_storage argument, set
26007 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26009 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26010 (build_array_type): Likewise. Add typeless_storage argument.
26012 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26013 Jakub Jelinek <jakub@redhat.com>
26015 PR tree-optimization/80426
26016 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26017 operation on symbolic operands, also compute the overflow for the
26018 invariant part when the operation degenerates into a negation.
26020 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26023 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26024 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26027 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26029 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26032 * config/avr/avr.c (tree.h): Include it.
26033 (cgraph.h): Include it.
26034 (avr_encode_section_info): Don't warn for uninitialized progmem
26035 variable if it's just an alias.
26037 2017-04-19 Richard Biener <rguenther@suse.de>
26040 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26041 when needed by AutoPGO.
26043 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26046 * doc/lto.texi: Remove an extra 'that'.
26048 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26050 PR rtl-optimization/80429
26051 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26052 are only used in debug insns.
26054 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26055 Vladimir Makarov <vmakarov@redhat.com>
26057 * config/sparc/predicates.md (input_operand): Add comment. Return
26058 true for any memory operand when LRA is in progress.
26059 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26061 2017-04-18 Jeff Law <law@redhat.com>
26064 * mips.md ({return,simple_return}_internal): Do not overwrite
26067 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26069 PR tree-optimization/80443
26070 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26071 instead of adding 1, subtract -1 and similarly instead of subtracting
26074 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26076 PR rtl-optimization/80357
26077 * haifa-sched.c (tmp_bitmap): New variable.
26078 (model_recompute): Handle duplicate use records.
26079 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26080 (free_global_sched_pressure_data): Free it.
26082 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26085 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26086 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26087 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26088 instead of SYSTEM_HEADER_DIR.
26090 2017-04-18 Jeff Law <law@redhat.com>
26092 PR middle-end/80422
26093 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26094 predecessors after walking up the insn chain.
26096 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26099 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26100 sizetype type into debug info.
26102 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26105 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26106 unneeded test for TARGET_UPPER_REGS_SF.
26107 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26109 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26112 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26113 instead of gsi_after_labels.
26115 2017-04-18 Jeff Law <law@redhat.com>
26117 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26121 2017-04-13 Jeff Law <law@redhat.com>
26122 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26123 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26125 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26128 * config/avr/avr.c (intl.h): Include it.
26129 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26131 2017-04-18 Martin Liska <mliska@suse.cz>
26133 PR gcov-profile/78783
26134 * gcov-tool.c (gcov_output_files): Validate that destination
26135 file is either removed by the tool or by a user.
26137 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26138 Guy Benyei <guybe@mellanox.com>
26140 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26141 block, and do not negate it, the stored id is already negative.
26143 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26145 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26147 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26150 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26151 masks of options that should be turned off if the VSX vector
26152 options are turned off.
26153 (OTHER_P8_VECTOR_MASKS): Likewise.
26154 (OTHER_VSX_VECTOR_MASKS): Likewise.
26155 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
26156 rs6000_disable_incompatible_switches to validate no type switches
26158 (rs6000_incompatible_switch): New function to disallow turning on
26159 other vector options if -mno-vsx, -mno-power8-vector, or
26160 -mno-power9-vector are specified.
26162 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26164 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
26166 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26168 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
26169 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
26170 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
26171 (ARG_POINTER_CFA_OFFSET): Likewise.
26173 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26175 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
26176 conditions to take advantage of various optimizations.
26178 2017-04-13 Jeff Law <law@redhat.com>
26180 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26181 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26182 (zero_extendsidi2_dext): Likewise.
26184 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26187 * fold-const.c (fold_ternary_loc): Revert
26188 use op0 instead of fold_convert_loc (loc, type, arg0) part of
26191 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
26193 PR rtl-optimization/80343
26194 * lra-remat.c (update_scratch_ops): Assign original hard reg to
26195 new scratch pseudo.
26197 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
26200 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
26201 to ubsan_encode_value.
26203 2017-04-13 Jeff Law <law@redhat.com>
26205 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
26206 appearing in DEBUG_INSNs.
26208 2017-04-13 Martin Liska <mliska@suse.cz>
26210 PR gcov-profile/80413
26211 * gcov-io.c (gcov_write_string): Copy to buffer just when
26212 allocated size is greater than zero.
26214 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26217 * dwarf2out.c (decls_for_scope): Ignore declarations of
26218 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26220 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26223 * ipa-visibility.c (non_local_p): Fix typos.
26224 (localize_node): When localizing symbol in same comdat group,
26225 dissolve the group only when we know external symbols are going
26227 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26229 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26231 PR tree-optimization/79390
26232 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26233 order does not result in usable sequence, retry with reversed operand
26239 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26240 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26241 op0 instead of fold_convert_loc (loc, type, arg0).
26243 2017-04-12 Jeff Law <law@redhat.com>
26245 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26246 has a delay slot in the generated code.
26248 * config/cris/cris.md (cris_preferred_reload_class): Return
26249 GENNONACR_REGS rather than GENERAL_REGS.
26251 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26254 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26255 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26256 signedness of the result type.
26258 2017-04-12 Richard Biener <rguenther@suse.de>
26259 Jeff Law <law@redhat.com>
26261 PR tree-optimization/80359
26262 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26263 trim stores to TARGET_MEM_REFs.
26265 2017-04-12 Richard Biener <rguenther@suse.de>
26267 PR tree-optimization/79390
26268 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26269 threading case even more.
26271 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26274 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26275 for quad_address_p for TImode, instead of just not indexed_address.
26277 2017-04-12 Richard Biener <rguenther@suse.de>
26278 Bernd Edlinger <bernd.edlinger@hotmail.de>
26280 PR middle-end/79671
26281 * alias.c (component_uses_parent_alias_set_from): Handle
26282 TYPE_TYPELESS_STORAGE.
26283 (get_alias_set): Likewise.
26284 * tree-core.h (tree_type_common): Add typeless_storage flag.
26285 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26286 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26287 for types containing members with TYPE_TYPELESS_STORAGE.
26288 (place_field): Likewise.
26289 (layout_type): Likewise for ARRAY_TYPE.
26290 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26291 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26292 TYPE_TYPELESS_STORAGE.
26293 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26295 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26298 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26299 first argument to type.
26301 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26305 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26306 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26307 (rs6000_expand_binop_builtin): Likewise.
26308 (rs6000_expand_ternop_builtin): Likewise; also add missing
26309 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26310 vshasigma built-ins.
26311 * doc/extend.texi: Document that vec_xxpermdi's third argument
26312 must be a constant.
26314 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26316 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26317 Use shift_const cost parameter when calculating gain of STV shifts.
26319 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26321 PR rtl-optimization/70478
26322 * lra-constraints.c (process_alt_operands): Check memory for
26323 disfavoring memory insn operand.
26325 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26327 PR middle-end/80100
26328 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26329 left shift in unsigned HOST_WIDE_INT type.
26331 PR rtl-optimization/80385
26332 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26333 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26336 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26337 if they have any depend clauses.
26339 2017-04-11 Martin Liska <mliska@suse.cz>
26342 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26343 * ipa-split.c (split_function): Create a local comdat symbol
26344 if caller is in a comdat group.
26346 2017-04-11 Martin Liska <mliska@suse.cz>
26349 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26352 2017-04-11 Martin Sebor <msebor@redhat.com>
26354 PR middle-end/80364
26355 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26356 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26358 (directive::set_width, directive::set_precision, format_character):
26360 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26363 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26366 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26367 conflict, set target->arch_name instead of target->cpu_name.
26369 2017-04-11 Richard Biener <rguenther@suse.de>
26371 PR tree-optimization/80374
26372 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26373 build_zero_cst, remove fold_convertible_p check again.
26375 2017-04-11 Martin Liska <mliska@suse.cz>
26378 * ubsan.c (instrument_object_size): Do not instrument register
26381 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26384 * config/i386/i386-builtin-types.def
26385 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26386 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26387 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26388 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26389 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26390 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26391 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26392 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26393 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26394 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26395 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26396 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26397 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26398 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26399 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26400 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26401 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26402 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26403 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26404 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26405 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26406 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26407 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26408 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26409 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26410 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26411 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26412 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26413 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26414 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26415 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26416 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26417 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26418 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26419 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26420 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26421 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26422 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26423 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26424 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26425 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26426 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26427 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26428 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26429 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26430 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26432 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26433 flag to second_arg_count, handle 4 argument function type _COUNT
26434 aliases, handle second_arg_count on second argument rather than last.
26436 2017-04-10 Jeff Law <law@redhat.com>
26438 PR tree-optimization/80374
26439 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26440 record anything if we can not convert integer_zero_node to the
26443 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26446 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26447 Enhance special handling given to the TARGET_P9_MINMAX option in
26448 relation to certain other options.
26450 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26452 PR tree-optimization/80153
26453 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26454 remove POINTER_PLUS_EXPR's base part directly, rather than through
26457 2017-04-10 Richard Biener <rguenther@suse.de>
26458 Bin Cheng <bin.cheng@arm.com>
26460 PR tree-optimization/80153
26461 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26462 the first element of pointer type aff_tree. Build result expr in
26464 (add_elt_to_tree): Convert to type unconditionally. Remove other
26465 fold_convert calls.
26466 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26467 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26469 2017-04-10 Richard Biener <rguenther@suse.de>
26471 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26474 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26476 PR rtl-optimization/70478
26477 * lra-constraints.c (curr_small_class_check): New.
26478 (update_and_check_small_class_inputs): New.
26479 (process_alt_operands): Update curr_small_class_check. Disfavor
26480 alternative insn memory operands. Check available regs for small
26483 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26486 * config/mips/mips.opt (-mvirt): Update description.
26487 * doc/invoke.texi (-mvirt): Likewise.
26489 2017-04-10 Richard Biener <rguenther@suse.de>
26491 PR middle-end/80362
26492 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26493 looking for NEGATE_EXPR in -A / -B to A / B folding.
26495 2017-04-10 Martin Liska <mliska@suse.cz>
26497 PR gcov-profile/80224
26498 * gcov.c (print_usage): Fix usage string.
26499 (get_gcov_intermediate_filename): Remove.
26500 (output_gcov_file): Use both for normal and intermediate format.
26501 (generate_results): Do not initialize special file for
26502 intermediate format.
26504 2017-04-10 Richard Biener <rguenther@suse.de>
26506 PR tree-optimization/80304
26507 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26510 2017-04-10 Nathan Sidwell <nathan@acm.org>
26513 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26514 (rs6000_init_builtins): Use it.
26516 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26518 * config/arm/arm.md (<mrc>): Add mode to SET source.
26519 (<mrrc>): Likewise.
26521 2017-04-10 Richard Biener <rguenther@suse.de>
26523 PR middle-end/80344
26524 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26526 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26529 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26530 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26531 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26532 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26533 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26534 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26535 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26536 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26537 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26538 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26539 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26540 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26541 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26542 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26543 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26544 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26545 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26546 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26547 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26548 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26549 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26550 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26551 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26553 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26555 PR rtl-optimization/70478
26556 * lra-constraints.c: Reverse the last patch.
26558 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26560 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26561 Add comment for WCHAR_T.
26563 2017-04-08 Martin Liska <mliska@suse.cz>
26566 2017-04-07 Martin Liska <mliska@suse.cz>
26569 * ipa-split.c (split_function): Add function part to a same comdat
26572 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26575 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26577 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26579 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26580 (vec_pairing): ...to this.
26581 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26582 (rs6000_sched_init): Adjust for name change.
26583 (struct rs6000_sched_context): Likewise.
26584 (rs6000_init_sched_context): Likewise.
26585 (rs6000_set_sched_context): Likewise.
26587 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26593 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26595 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26596 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26597 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26599 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26601 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26603 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26605 PR rtl-optimization/70703
26606 * ira-color.c (update_conflict_hard_regno_costs): Use
26607 int64_t instead of HOST_WIDE_INT.
26609 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26611 PR rtl-optimization/70478
26612 * lra-constraints.c (process_alt_operands): Disfavor alternative
26613 insn memory operands.
26615 2017-04-07 Jeff Law <law@redhat.com>
26617 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26618 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26620 2017-04-07 Martin Liska <mliska@suse.cz>
26623 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26624 Show error message instead of an ICE.
26626 2017-04-07 Martin Liska <mliska@suse.cz>
26629 * ipa-split.c (split_function): Add function part to a same comdat
26632 2017-04-07 Richard Biener <rguenther@suse.de>
26634 PR middle-end/80341
26635 * tree.c (get_unwidened): Also handle ! for_type case for
26637 * convert.c (do_narrow): Split out from ...
26638 (convert_to_integer_1): ... here. Do not pass final truncation
26639 type to get_unwidened for TRUNC_DIV_EXPR.
26641 2017-04-07 Richard Biener <rguenther@suse.de>
26643 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26645 (aff_combination_const): Adjust.
26646 (aff_combination_scale): Likewise.
26647 (aff_combination_add_elt): Likewise.
26648 (aff_combination_add_cst): Likewise.
26649 (aff_combination_convert): Likewise.
26650 (add_elt_to_tree): Likewise. Remove unused argument.
26651 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26653 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
26655 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
26657 * config/arm/arm.c (arm_default_short_enums): Use
26658 ARM_DEFAULT_SHORT_ENUMS.
26659 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
26661 2017-04-06 Jakub Jelinek <jakub@redhat.com>
26664 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
26665 members with redundant out-of-class redeclaration.
26667 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26670 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26671 * config/i386/i386.md (*zero_extendsidi2):
26672 Add (?*x,*x) and (?*v,*v) alternatives.
26674 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26677 * config/i386/i386.c (ix86_expand_builtin)
26678 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26679 mode from insn data. Convert operands to insn operand mode.
26680 Copy operands that don't satisfy insn predicate to a register.
26682 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26684 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26687 2017-04-06 Richard Biener <rguenther@suse.de>
26689 PR tree-optimization/80334
26690 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26691 preserve alignment of accesses.
26693 2017-04-06 Richard Biener <rguenther@suse.de>
26695 PR tree-optimization/80262
26696 * tree-sra.c (build_ref_for_offset): Preserve address-space
26698 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26699 Drop useless address-space information on MEM_REF offsets.
26701 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26703 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26705 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26707 PR rtl-optimization/70703
26708 * ira-color.c (update_conflict_hard_regno_costs): Use
26709 HOST_WIDE_INT instead of long.
26711 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26714 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26715 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26717 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26718 for x86_64 target. Handle -m3dnowa option.
26720 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26722 PR rtl-optimization/70703
26723 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26724 (update_conflict_hard_regno_costs): Use long instead of unsigned
26725 arithmetic for cost calculation.
26727 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26728 Bernd Edlinger <bernd.edlinger@hotmail.de>
26731 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26734 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26737 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26738 ptr_mode with Pmode throughout.
26739 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26740 into probe_stack_range and use DImode.
26742 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26745 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26746 call_eh_return is true.
26748 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26750 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26751 Initialize last_match_fntype_index.
26753 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26756 * tree-nvr.c: Include internal-fn.h.
26757 (pass_return_slot::execute): Ignore internal calls without
26760 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26761 Richard Biener <rguenther@suse.de>
26764 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26765 captures used multiple times, except for the last use.
26766 * generic-match-head.c: Include gimplify.h.
26768 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26770 PR tree-optimization/79390
26771 * target.h (struct noce_if_info): Declare.
26772 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26773 * target.def (noce_conversion_profitable_p): New target hook.
26774 * ifcvt.h (struct noce_if_info): New type, moved from ...
26775 * ifcvt.c (struct noce_if_info): ... here.
26776 (noce_conversion_profitable_p): Renamed to ...
26777 (default_noce_conversion_profitable_p): ... this. No longer
26779 (noce_try_store_flag_constants, noce_try_addcc,
26780 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26781 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26782 instead of noce_conversion_profitable_p.
26783 * config/i386/i386.c: Include ifcvt.h.
26784 (ix86_option_override_internal): Don't override
26785 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26786 (ix86_noce_conversion_profitable_p): New function.
26787 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26788 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26789 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26790 * doc/tm.texi: Regenerated.
26792 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26794 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26797 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26800 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26801 instructions for small multiply cores.
26803 2017-04-04 Jeff Law <law@redhat.com>
26805 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26807 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26808 that are not set by the loop over the elements.
26810 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26813 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26814 int mode, convert_modes it to mode as unsigned, otherwise use
26815 lowpart_subreg to mode rather than SImode.
26816 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26817 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26818 Use DImode instead of SImode for the shift count operand.
26819 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26822 2017-04-04 Richard Biener <rguenther@suse.de>
26824 PR middle-end/80281
26825 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26826 arithmetic done for the negate or the plus. Simplify.
26827 (A - (-B) -> A + B): Likewise.
26828 * fold-const.c (split_tree): Make sure to not negate pointers.
26830 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26832 PR rtl-optimization/60818
26833 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26834 a compare of comparisons with the thing compared if this results
26835 in a different machine mode.
26837 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26839 * alias.c (base_alias_check): Fix typo in comment.
26840 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26841 * cgraphunit.c (symbol_table::compile): Likewise.
26842 * collect2.c (maybe_run_lto_and_relink): Likewise.
26843 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26844 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26845 * config/avr/avr.c (avr_map_op_t): Likewise.
26846 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26847 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26848 * config/epiphany/epiphany.md (movcc): Likewise.
26849 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26850 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26852 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26853 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26854 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26855 * config/sh/sh.c (sh_rtx_costs): Likewise.
26856 * fold-const.c (fold_truth_andor): Likewise.
26857 * genautomata.c (collapse_flag): Likewise.
26858 * gengtype.h (struct type::u::s): Likewise.
26859 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26860 * input.c (FORMAT_AMOUNT): Likewise.
26861 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26862 (known_aggs_to_agg_replacement_list): Likewise.
26863 * ipa-inline-analysis.c: Likewise.
26864 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26865 * ipa-polymorphic-call.c
26866 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26867 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26868 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26870 * modulo-sched.c (apply_reg_moves): Likewise.
26871 * omp-expand.c (build_omp_regions_1): Likewise.
26872 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26873 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26874 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26875 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26876 * value-prof.c: Likewise.
26877 * var-tracking.c (val_reset): Likewise.
26879 2017-04-03 Richard Biener <rguenther@suse.de>
26881 PR tree-optimization/80275
26882 * fold-const.c (split_address_to_core_and_offset): Handle
26885 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26887 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26888 descriptors is at least equal to that of functions.
26890 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26892 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26894 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26897 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26898 (mov<IMOD4:mode>): New expander.
26899 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26901 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26903 PR rtl-optimization/79405
26904 * fwprop.c (propagations_left): New variable.
26905 (forward_propagate_into): Decrement it.
26906 (fwprop_init): Initialize it.
26907 (fw_prop): If the variable has reached zero, stop propagating.
26908 (fwprop_addr): Ditto.
26910 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26913 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26914 a FUNCTION_DECL, pass it as decl instead of origin to
26917 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26919 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26922 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26925 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26926 TARGET_VSX_SMALL_INTEGER.
26928 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26930 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26931 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26933 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26935 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26936 extraction from odd-numbered MSA register.
26938 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26940 PR middle-end/80173
26941 * expmed.c (store_bit_field_1): Don't attempt to create
26942 a word subreg out of hard registers wider than word if they
26943 have HARD_REGNO_NREGS of 1 for their mode.
26945 PR middle-end/80163
26946 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26947 conversions to integer types wider than word and pointer.
26950 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26951 (rtx_equal_for_cselib_p): Pass 0 to it.
26952 * cselib.c (cselib_hasher::equal): Likewise.
26953 (rtx_equal_for_cselib_1): Add depth argument. If depth
26954 is 128, don't look up VALUE locs and punt. Increment
26955 depth in recursive calls when walking VALUE locs.
26957 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26959 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26960 (make_gcov_file_name): Use the canonical path name for generating
26962 (read_line): Fix handling of files with ascii null bytes.
26964 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26966 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
26967 to initialise a vector register instead
26968 of using a const_int.
26970 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26972 PR translation/80189
26973 * gimplify.c (omp_default_clause): Use %qs instead of %s in
26974 diagnostic messages.
26976 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
26979 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
26980 (dfp_diex_<mode>): Update mode of operand 1.
26981 * doc/extend.texi (dxex, dxexq): Document change to return type.
26982 (diex, diexq): Document change to argument type.
26984 2017-03-30 Martin Jambor <mjambor@suse.cz>
26987 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
26988 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
26989 it reflects the signature changes performed at the callee side.
26990 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
26991 to cgraph_build_function_type_skip_args.
26992 (build_function_decl_skip_args): Adjust call to the above function.
26994 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26997 * config/i386/sse.md
26998 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
26999 register as dest whenever it is a MEM not rtx_equal_p to the
27000 corresponding dup operand, and when forcing into reg move the
27001 reg into the memory afterwards.
27002 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27003 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27004 for the force_reg mode.
27005 (avx512vl_vextractf128<mode>): Use register as dest either
27006 always when a MEM, or when it is a MEM not rtx_equal_p to the
27007 corresponding dup operand, or even not when it is a CONST_VECTOR
27008 depending on the mode and lo vs. hi.
27009 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27011 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27012 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27013 Likewise. Require that operands[2] is even.
27014 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27015 Remove extraneous parens. Require that operands[2] is a multiple
27017 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27018 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27019 disallow memory then.
27021 2017-03-30 Richard Biener <rguenther@suse.de>
27023 PR tree-optimization/77498
27024 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27025 to non-constants over backedges.
27027 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27029 PR rtl-optimization/80233
27030 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27031 as last_combined_insn. Do not test for BARRIER_P separately.
27033 2017-03-29 Andreas Schwab <schwab@suse.de>
27036 * calls.c (prepare_call_address): Convert funexp to Pmode before
27037 copying to temp reg.
27039 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27041 PR tree-optimization/80158
27042 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27043 Handle possible future case of more than one alternate
27045 (replace_rhs_if_not_dup): Likewise.
27046 (replace_one_candidate): Likewise.
27048 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27050 PR rtl-optimization/80193
27051 * ira.c (ira): Do not check allocation for LRA.
27053 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27055 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27056 (nvptx_output_simt_exit): Declare.
27057 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27058 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27059 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27060 (nvptx_declare_function_name): ...here. Emit declaration of local
27061 memory space buffer for omp_simt_enter insn.
27062 (nvptx_output_unisimt_switch): New.
27063 (nvptx_output_softstack_switch): New.
27064 (nvptx_output_simt_enter): New.
27065 (nvptx_output_simt_exit): New.
27066 * config/nvptx/nvptx.h (struct machine_function): New fields
27067 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27068 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27069 (UNSPECV_SIMT_EXIT): Ditto.
27070 (omp_simt_enter_insn): New insn.
27071 (omp_simt_enter): New expansion.
27072 (omp_simt_exit): New insn.
27073 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27075 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27076 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27077 (expand_GOMP_SIMT_EXIT): New.
27078 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27079 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27080 (GOMP_SIMT_EXIT): Ditto.
27081 * target-insns.def (omp_simt_enter): New insn.
27082 (omp_simt_exit): Ditto.
27083 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27085 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27086 (lower_rec_input_clauses): Likewise.
27087 (lower_lastprivate_clauses): Handle SIMT privatization.
27089 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27090 (ompdevlow_adjust_simt_enter): New.
27091 (find_simtpriv_var_op): New.
27092 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27093 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27095 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27096 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27097 (copy_decl_for_dup_finish): Ditto.
27099 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27101 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27104 * config/i386/i386.c (ix86_option_override_internal): Always
27105 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27107 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27109 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27111 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27113 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27114 mark new edge's irreducible flag accordign to it.
27115 (vect_do_peeling): Check loop preheader edge's irreducible flag
27116 and pass it to function slpeel_add_loop_guard.
27118 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27120 PR tree-optimization/80218
27121 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27122 Update block frequencies and counts.
27124 2017-03-28 Richard Biener <rguenther@suse.de>
27126 PR tree-optimization/78644
27127 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27128 of a simplification result we may not use it at all.
27130 2017-03-28 Richard Biener <rguenther@suse.de>
27133 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27134 without arguments, generate default definition of a SSA name.
27136 2017-03-28 Richard Biener <rguenther@suse.de>
27138 PR middle-end/80222
27139 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27140 TYPE_REF_CAN_ALIAS_ALL references.
27141 * fold-const.c (fold_indirect_ref_1): Likewise.
27143 2017-03-28 Martin Liska <mliska@suse.cz>
27146 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27147 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27149 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27150 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27152 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27153 (EXTRA_SPECS): Define.
27154 (SUBTARGET_EXTRA_SPECS): Likewise.
27155 (SUBTARGET_CPP_SPEC): Likewise.
27156 * config/arc/elf.h (EXTRA_SPECS): Renamed to
27157 SUBTARGET_EXTRA_SPECS.
27158 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
27160 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27162 * config/arc/simdext.md (vst64_insn): Update pattern.
27163 (vld32wh_insn): Likewise.
27164 (vld32wl_insn): Likewise.
27165 (vld64_insn): Likewise.
27166 (vld32_insn): Likewise.
27168 2017-03-28 Marek Polacek <polacek@redhat.com>
27171 * fold-const.c (fold_comparison): Use protected_set_expr_location
27172 instead of SET_EXPR_LOCATION.
27174 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
27176 * tree.c (add_expr): Avoid name lookup warning.
27178 2017-03-27 Jeff Law <law@redhat.com>
27180 PR tree-optimization/80216
27181 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
27182 function name. Limit recursion depth.
27183 (record_temporary_equivalences): Corresponding changes.
27185 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
27187 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
27190 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27193 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
27195 * cfgcleanup.c (reg_note_cfa_p): New array.
27196 (insns_have_identical_cfa_notes): New function.
27197 (old_insns_match_p): Don't cross-jump in between /f
27198 and non-/f instructions. If both i1 and i2 are frame related,
27199 verify all CFA notes, their order and content.
27201 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
27204 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
27205 HImode and SImode with zero extend to DImode to one insn.
27206 (bswap<mode>2_extenddi): Likewise.
27207 (bswapsi2_extenddi): Likewise.
27208 (bswaphi2_extendsi): Likewise.
27209 (bswaphi2): Combine bswap HImode and SImode into one insn.
27210 Separate memory insns from swapping register.
27211 (bswapsi2): Likewise.
27212 (bswap<mode>2): Likewise.
27213 (bswaphi2_internal): Delete, no longer used.
27214 (bswapsi2_internal): Likewise.
27215 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
27216 store, and gpr<-gpr swap insns.
27217 (bswap<mode>2_store): Likewise.
27218 (bswaphi2_reg): Register only splitter, combine with the splitter.
27219 (bswaphi2 splitter): Likewise.
27220 (bswapsi2_reg): Likewise.
27221 (bswapsi2 splitter): Likewise.
27222 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27223 the insns into load, store, and register/register insns.
27224 (bswapdi2_ldbrx): Likewise.
27225 (bswapdi2_load): Likewise.
27226 (bswapdi2_store): Likewise.
27227 (bswapdi2_reg): Likewise.
27229 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27231 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27232 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27234 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27237 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27239 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27240 special handling for target option conflicts between dform
27241 options (-mpower9-dform, -mpower9-dform-vector,
27242 -mpower9-dform-scalar) and -mno-direct-move.
27244 2017-03-27 Richard Biener <rguenther@suse.de>
27246 PR tree-optimization/80181
27247 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27249 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27251 * config/arc/predicates.md (move_double_src_operand): Replace the
27252 call to move_double_src_operand with a call to address_operand.
27254 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27256 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27257 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27258 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27260 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27262 * config/arc/predicates.md (long_immediate_loadstore_operand):
27263 Consider scaled addresses cases.
27265 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27267 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27268 restored when in interrupt.
27269 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27270 doesn't have delay slot.
27272 2017-03-27 Richard Biener <rguenther@suse.de>
27275 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27276 inlined thunk clones.
27278 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27281 * asan.c (instrument_derefs): Copy over last operand from
27282 original COMPONENT_REF to the new COMPONENT_REF with
27283 DECL_BIT_FIELD_REPRESENTATIVE.
27284 * ubsan.c (instrument_object_size): Likewise.
27286 2017-03-27 Richard Biener <rguenther@suse.de>
27288 PR tree-optimization/80170
27289 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27290 sure DR/SCEV didnt fold in constants we do not see when looking
27291 at the reference base alignment.
27293 2017-03-27 Richard Biener <rguenther@suse.de>
27295 PR middle-end/80171
27296 * gimple-fold.c (fold_ctor_reference): Properly guard against
27297 NULL return value from canonicalize_constructor_val.
27299 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27302 * config/i386/i386.c (ix86_expand_builtin)
27303 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27304 flags reg setting and flags reg using instructions.
27305 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27306 clobbering instructions to zero extend op2.
27308 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27310 * doc/install.texi (Configuration) <--with-aix-soname>:
27311 Update link to AIX ld.
27313 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27315 PR rtl-optimization/80160
27316 PR rtl-optimization/80159
27317 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27318 reg_alternate_class into account.
27320 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27323 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27324 to consider in curr_insn_transform.
27326 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27328 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27329 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27330 and emit_mode_inner.
27332 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27334 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27335 argument to the overloaded builtin variants. Use the new flag to
27336 deprecate certain builtin variants.
27337 * config/s390/s390-builtin-types.def: Add new builtin types.
27338 * config/s390/s390-builtins.h: Support new flags field for
27339 overloaded builtins.
27340 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27341 (s390_macro_to_expand): Enable vector float data type.
27342 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27343 builtins by incrementing the __VEC__ version number.
27344 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27346 (s390_resolve_overloaded_builtin): Emit error messages depending
27347 on the builtin flags.
27348 * config/s390/s390.c (s390_expand_builtin): Support additional
27349 flags argument. Change error message to match the messages
27350 emitted in s390-c.c.
27351 * config/s390/s390.md: New UNSPEC_* constants.
27352 (op_type): Add new instruction types.
27353 * config/s390/vecintrin.h: Add new builtins and test data class
27355 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27356 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27357 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27358 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27359 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27360 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27362 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27363 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27364 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27365 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27367 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27368 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27369 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27370 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27372 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27373 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27374 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27375 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27376 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27377 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27378 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27380 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27381 ("vec_scatter_element<V_HW_4:mode>_DI")
27382 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27383 ("vec_fpint<mode>", "vflls")
27384 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27385 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27386 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27387 ("*vec_cmphe<mode>_cc"): ... these.
27389 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27390 mode constant instead of magic value.
27392 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27394 * config/s390/s390.c (s390_expand_vec_compare): Support other
27395 vector floating point modes than just V2DF.
27396 (s390_expand_vcond): Likewise.
27397 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27398 (s390_cannot_change_mode_class): Prevent mode changes between TF
27399 and V1TF in vector registers.
27400 * config/s390/s390.md (DF, SF): New mode attributes.
27401 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27402 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27403 SFmode support for VRs.
27404 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27406 (VFT, VF_HW): New mode iterators.
27407 (vw, sdx): New mode attributes.
27408 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27409 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27410 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27411 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27412 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27413 also the new vector floating point modes. Renaming to ...
27415 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27416 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27417 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27418 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27419 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27420 ("vec_unordered<mode>"): ... these.
27422 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27423 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27424 ("*vec_extendv2df"): New insn definitions.
27426 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27428 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27429 ("mulditi3_2", "*muldi3_sign"): New patterns.
27430 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27431 rename the pattern definition.
27433 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27435 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27437 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27439 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27441 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27442 instruction if possible.
27443 * config/s390/vector.md (vec_halfnumelts): New mode
27445 ("*vec_vllezlf<mode>"): New pattern.
27447 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27449 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27450 ("popcountv4si2", "popcountv2di2"): Rename to ...
27451 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27452 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27454 ("popcount<mode>2_vxe"): New pattern.
27456 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27458 * common/config/s390/s390-common.c (processor_flags_table): Add
27460 * config.gcc: Add arch12.
27461 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27462 Default to arch12 for unknown CPU model numbers.
27463 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27464 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27465 PROCESSOR_max sanity check.
27466 * config/s390/s390-opts.h (enum processor_type): Add
27468 * config/s390/s390.c (processor_table): Add arch12.
27469 (s390_expand_builtin): Add check for B_VXE flag.
27470 (s390_issue_rate): Add PROCESSOR_ARCH12.
27471 (s390_get_sched_attrmask): Likewise.
27472 (s390_get_unit_mask): Likewise.
27473 (s390_sched_score): Enable z13 scheduling for arch12.
27474 (s390_sched_reorder): Likewise.
27475 (s390_sched_variable_issue): Likewise.
27476 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27478 (s390_tune_attr): Use z13 scheduling also for arch12.
27479 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27480 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27481 (TARGET_VXE_P): New macros.
27482 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27483 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27484 * config/s390/s390.opt: Add arch12 as processor_type.
27486 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27488 * config/s390/s390.md
27489 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27490 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27491 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27493 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27494 Rename expanders to ...
27496 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27497 ("fixuns_truncdddi2_emu"): ... these.
27499 ("fixuns_trunc<mode>si2_emu"): New expander.
27501 ("*fixuns_truncdfdi2_z13"): Rename to ...
27502 ("*fixuns_truncdfdi2_vx"): ... this.
27504 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27506 * config/s390/2964.md: Remove the single element vector compare
27507 instructions which are no longer used.
27508 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27510 (s390_canonicalize_comparison): Remove handling of DFmode
27512 (s390_expand_vec_compare_scalar): Remove function.
27513 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27514 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27516 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27518 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27520 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27522 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27523 will anyway by matched by mov<mode>_64dfp.
27525 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27527 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27528 vlef/vstef. Add missing operand to vleif.
27530 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27532 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27533 pair for all vector types with 64 bit elements.
27534 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27535 * config/s390/vector.md (V_HW_64): ... here.
27536 (V_128_NOSINGLE): New mode iterator.
27537 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27538 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27539 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27540 ("*vec_load_pairv2di"): Change to ...
27541 ("*vec_load_pair<mode>"): ... this one.
27543 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27545 * config/s390/constraints.md: Add comments.
27546 (jKK): Reject element sizes > 8 bytes.
27547 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27549 * config/s390/s390.md: Add the s_operand checks formerly in
27550 s390_split_ok_p to various splitters where they are still
27552 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27553 for 128 bit vectors. Plus two splitters.
27555 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27557 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27560 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27563 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27564 error if the boundary argument is not constant.
27566 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27568 PR rtl-optimization/80112
27569 * loop-doloop.c (doloop_condition_get): Don't check condition
27570 if cmp isn't SET with IF_THEN_ELSE src.
27572 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27574 PR tree-optimization/80158
27575 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27576 replacing a candidate statement, also replace it for the
27577 candidate's alternate interpretation.
27578 (replace_rhs_if_not_dup): Likewise.
27579 (replace_one_candidate): Likewise.
27581 2017-03-24 Richard Biener <rguenther@suse.de>
27583 PR tree-optimization/80167
27584 * graphite-isl-ast-to-gimple.c
27585 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27587 (translate_isl_ast_to_gimple::get_rename): Likewise.
27589 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27591 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27592 handling of certain combinations of target options, including the
27593 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27594 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27596 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27599 * config/arm/arm.md (*load_multiple): Add reload_completed to
27600 matching condition.
27602 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27603 Richard Biener <rguenth@suse.de>
27605 PR tree-optimization/79908
27606 PR tree-optimization/80136
27607 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27608 been cast away, gimplify_and_add suffices.
27610 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27612 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27614 2017-03-23 Richard Biener <rguenther@suse.de>
27616 PR tree-optimization/80032
27617 * gimplify.c (gimple_push_cleanup): Forced unconditional
27618 cleanups still have to go to the conditional_cleanups
27621 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27623 PR tree-optimization/80072
27624 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27626 (next_operand_entry_id): Change type to unsigned int.
27627 (sort_by_operand_rank): Make sure to return the right return value
27628 even if unsigned fields are bigger than INT_MAX.
27629 (struct oecount): Change cnt and id type to unsigned int.
27630 (oecount_hasher::equal): Formatting fix.
27631 (oecount_cmp): Make sure to return the right return value
27632 even if unsigned fields are bigger than INT_MAX.
27633 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27636 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27637 TREE_READONLY on result if writing it more than once.
27640 * doc/invoke.texi (-fsanitize=thread): Document that with
27641 -fnon-call-exceptions atomics are not able to throw
27645 * tsan.c: Include tree-eh.h.
27646 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27647 maybe_clean_or_replace_eh_stmt where needed.
27648 (instrument_memory_accesses): Add cfg_changed argument.
27649 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27650 if it returned true.
27651 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27653 PR rtl-optimization/63191
27654 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
27655 wrapper function, moved the whole old content into ...
27656 (ix86_delegitimize_address_1): ... this. New inline function.
27657 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
27658 true as last argument instead of ix86_delegitimize_address.
27660 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27662 * config/aarch64/aarch64.c (generic_branch_cost): Copy
27663 cortexa57_branch_cost.
27665 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27667 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
27669 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27672 * doc/md.texi (Constraints): Document wA constraint.
27673 * config/rs6000/constraints.md (wA): New.
27674 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27675 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27676 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27677 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27679 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27682 * gimplify.c (is_oacc_declared): New function.
27683 (oacc_default_clause): Use it to set default flags for acc declared
27684 variables inside parallel regions.
27685 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27686 declared variables.
27687 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27688 declare attribute to any decl as necessary.
27690 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27693 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27694 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27695 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27696 (arm_arch_lpae): This.
27697 * config/arm/arm.c (arm_arch7ve): Rename into ...
27698 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27699 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27702 2017-03-22 Martin Liska <mliska@suse.cz>
27705 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27706 error message instead of an ICE.
27708 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27710 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27712 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27714 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27716 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27719 2017-03-21 Martin Sebor <msebor@redhat.com>
27721 * doc/extend.texi: Use "cannot" instead of "can't."
27722 * doc/hostconfig.texi: Same.
27723 * doc/install.texi: Same.
27724 * doc/invoke.texi: Same.
27725 * doc/loop.texi: Same.
27726 * doc/md.texi: Same.
27727 * doc/objc.texi: Same.
27728 * doc/rtl.texi: Same.
27729 * doc/tm.texi: Same.
27730 * doc/tm.texi.in: Same.
27731 * doc/trouble.texi: Same.
27733 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27736 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27737 (collect_checksum_attributes): Set it.
27738 (die_checksum_ordered): Use it.
27740 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27742 PR tree-optimization/79908
27743 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27744 change: For a VA_ARG whose LHS has been cast away, use
27745 force_gimple_operand to construct the side effects.
27747 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27749 PR translation/80001
27750 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27751 more amenable to translation.
27752 (oacc_loop_auto_partitions): Likewise.
27754 2017-03-21 Marek Polacek <polacek@redhat.com>
27755 Martin Sebor <msebor@redhat.com>
27757 PR tree-optimization/80109
27758 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27759 on INTEGRAL_TYPE_P.
27761 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27762 Segher Boessenkool <segher@kernel.crashing.org>
27765 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27766 check reg_used_between_p between insn and one of succ or succ2
27767 depending on if succ is artificial insn not inserted into insn
27770 2017-03-21 Martin Liska <mliska@suse.cz>
27772 PR gcov-profile/80081
27773 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27774 * doc/gcc.texi: Include gcov-dump stuff.
27775 * doc/gcov-dump.texi: New file.
27777 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27779 PR rtl-optimization/79150
27780 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27781 conditional jump, if the jump is the last insn of the loop.
27783 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27784 Richard Biener <rguenth@suse.de>
27786 PR tree-optimization/79908
27787 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27788 been cast away, use force_gimple_operand to construct the side
27791 2017-03-21 Martin Liska <mliska@suse.cz>
27793 PR libfortran/79956
27794 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27797 2017-03-21 Brad Spengler <spender@grsecurity.net>
27800 * plugin.c (htab_hash_plugin): New function.
27801 (add_new_plugin): Use it and adjust.
27802 (parse_plugin_arg_opt): Adjust.
27803 (init_one_plugin): Likewise.
27805 2017-03-21 Richard Biener <rguenther@suse.de>
27807 PR tree-optimization/80032
27808 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27809 if set force the cleanup to happen unconditionally.
27810 (gimplify_target_expr): Push inserted clobbers with force_uncond
27811 to avoid them being removed by control-dependent DCE.
27813 2017-03-21 Richard Biener <rguenther@suse.de>
27815 PR tree-optimization/80122
27816 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27817 va_arg_pack_len when the inlined call stmt requires pack
27819 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27821 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27824 * tsan.c (instrument_builtin_call): If the memory model argument
27825 is not a constant, assume it is valid.
27828 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27831 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27833 PR rtl-optimization/79910
27834 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27835 if its dest is used by an insn before I2 (other than the combined
27836 insns themselves, which are properly handled already).
27838 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27841 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27843 * combine.c (record_used_regs): New static function.
27844 (try_combine): Handle situations where there is an additional
27845 instruction between I2 and I3 which needs to have a LOG_LINK
27849 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27851 * combine.c (try_combine): Delete redundant i1 test. Call
27852 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27854 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27857 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27858 alternatives 13/14.
27860 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27862 PR tree-optimization/80054
27863 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27864 the optimization if a PHI or any of its arguments is not dominated
27865 by the candidate's basis. Use gphi* rather than gimple* as
27867 (replace_profitable_candidates): Clean up a gimple* variable that
27868 should be a gphi* variable.
27870 2017-03-20 Martin Sebor <msebor@redhat.com>
27873 * doc/extend.texi (attribute constructor): Document present limitation.
27875 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27878 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27879 __POWER9_VECTOR__ #ifdef control, change template definition to
27880 use Power9-specific built-in function.
27881 (vec_any_eq): Likewise.
27882 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27883 to control outcomes from this test.
27884 (vector_ae_<mode>p): For VEC_F modes, likewise.
27886 2017-03-20 Ian Lance Taylor <iant@google.com>
27888 * config/i386/i386.c (ix86_function_regparm): Save an extra
27889 register for -fsplit-stack with DECL_STATIC_CHAIN.
27891 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27894 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27895 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27897 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27899 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27901 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27904 2017-03-20 Marek Polacek <polacek@redhat.com>
27907 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27909 2017-03-20 Richard Biener <rguenther@suse.de>
27911 PR tree-optimization/80113
27912 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27913 allocate extra SSA name for PHI def.
27914 (add_close_phis_to_outer_loops): Likewise.
27915 (add_close_phis_to_merge_points): Likewise.
27916 (copy_loop_close_phi_args): Likewise.
27917 (copy_cond_phi_nodes): Likewise.
27919 2017-03-20 Martin Liska <mliska@suse.cz>
27921 PR middle-end/79753
27922 * tree-chkp.c (chkp_build_returned_bound): Do not build
27923 returned bounds for a LHS that's not a BOUNDED_P type.
27925 2017-03-20 Martin Liska <mliska@suse.cz>
27929 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27930 COMPLEX_CST and VECTOR_CST.
27932 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27935 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27936 target operand. A new splitter adds the clobber statement in case
27937 the target operand is dead anyway.
27939 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27941 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27942 to age-old versions of binutils and glibc.
27944 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27946 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27948 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27950 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27952 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27954 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27955 requirement for binutils 2.13.
27957 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27959 * combine.c (try_combine): Delete redundant i1 test. Call
27960 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27962 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27964 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27965 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
27967 <riscv64-*-elf>: Re-arrange section
27968 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
27969 <riscv32-*-linux>: Likewise.
27970 <riscv64-*-elf>: Likewise
27971 <riscv64-*-linux>: Likewise.
27973 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
27976 * aarch64.opt(verbose-cost-dump): Fix typo.
27978 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
27981 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
27982 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
27984 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27986 * reload.c (find_reloads): When reloading a nonoffsettable address,
27987 use RELOAD_OTHER for it and its address reloads.
27989 PR rtl-optimization/79910
27990 * combine.c (record_used_regs): New static function.
27991 (try_combine): Handle situations where there is an additional
27992 instruction between I2 and I3 which needs to have a LOG_LINK
27995 2017-03-17 Jeff Law <law@redhat.com>
27997 PR tree-optimization/71437
27998 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
27999 conditional in the hash table first.
28000 (vrp_dom_walker::before_dom_children): Extract condition from
28001 ASSERT_EXPR. Record condition, its inverion and any implied
28002 conditions as well.
28004 2017-03-17 Marek Polacek <polacek@redhat.com>
28005 Markus Trippelsdorf <markus@trippelsdorf.de>
28007 PR tree-optimization/80079
28008 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28011 2017-03-17 Richard Biener <rguenther@suse.de>
28013 PR middle-end/80075
28014 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28015 Properly verify the LHS before the RHS possibly claims to be
28017 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28020 2017-03-17 Martin Jambor <mjambor@suse.cz>
28022 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28023 (List of -O2 options): Likewise.
28024 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28027 2017-03-17 Tom de Vries <tom@codesourcery.com>
28029 * gcov-dump.c (print_usage): Print bug_report_url.
28031 2017-03-17 Richard Biener <rguenther@suse.de>
28033 PR middle-end/80050
28034 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28035 (parser::peek): Likewise.
28037 2017-03-17 Richard Biener <rguenther@suse.de>
28039 PR tree-optimization/80048
28040 * sese.c (free_sese_info): Properly release rename_map and
28041 copied_bb_map elements.
28043 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28045 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28046 Add linked-list forward and backlinks. Insert on
28047 construction, remove on destruction.
28048 (class pass_store_merging): Add m_stores_head field.
28049 (pass_store_merging::terminate_and_process_all_chains):
28050 Iterate over m_stores_head list.
28051 (pass_store_merging::terminate_all_aliasing_chains):
28053 (pass_store_merging::execute): Check for debug stmts first.
28054 Push new chains onto the m_stores_head stack.
28056 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28059 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28060 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28061 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28063 2017-03-16 Jeff Law <law@redhat.com>
28065 PR tree-optimization/71437
28066 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28067 member function. Implementation moved into after_dom_children
28068 member function and into the threader's thread_outgoing_edges
28070 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28071 some code into new thread_outgoing_edges.
28072 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28073 definition. Simplify marker handling (do it here). Assume we always
28074 have the available expression and the const/copies tables.
28075 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28077 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28078 * tree-vrp.c (equiv_stack): No longer file scoped.
28079 (vrp_dom_walker): New class.
28080 (vrp_dom_walker::before_dom_children): New member function.
28081 (vrp_dom_walker::after_dom_children): Likewise.
28082 (identify_jump_threads): Setup domwalker. Use it rather than
28083 walking edges in a random order by hand. Simplify setup/finalization.
28084 (finalize_jump_threads): Remove.
28085 (vrp_finalize): Do not call identify_jump_threads here.
28086 (execute_vrp): Do it here instead and call thread_through_all_blocks
28089 PR tree-optimization/71437
28090 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28092 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28094 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28095 (dom_opt_dom_walker::thread_across_edge): Remove
28096 handle_dominating_asserts argument. All callers changed.
28097 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28098 changes. Remove calls to lhs_of_dominating_assert. Other
28099 uses of handle_dominating_asserts turn into unconditional code
28100 (simplify_control_stmt_condition_1): Likewise.
28101 (simplify_control_stmt_condition): Likewise.
28102 (thread_through_normal_block, thread_across_edge): Likewise.
28103 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28104 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28105 object if it is not an SSA_NAME.
28106 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28107 before calling into the VRP specific simplifiers.
28108 (identify_jump_threads): Remove handle_dominating_asserts
28111 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28114 * tree-diagnostic.c (default_tree_printer): No longer static.
28115 * tree-diagnostic.h (default_tree_printer): New prototype.
28117 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28119 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28120 Change ins into fmov.
28122 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28124 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28125 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28126 Use h_con constraint for operand 1.
28127 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28128 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28130 2017-03-15 Jeff Law <law@redhat.com>
28132 PR tree-optimization/71437
28133 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28134 (record_temporary_equivalences): Use it.
28136 PR tree-optimization/71437
28137 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28138 tree-ssa-scopedtables.
28139 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28140 (record_conditions, record_cond, vuse_eq): Likewise.
28141 (record_edge_info): Adjust to API tweak of record_conditions.
28142 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28143 (record_temporary_equivalences, optimize_stmt): Likewise.
28144 (eliminate_redundant_computations): Likewise.
28145 (record_equivalences_from_stmt): Likewise.
28146 * tree-ssa-scopedtables.c: Include options.h and params.h.
28147 (vuse_eq): New function, moved from tree-ssa-dom.c
28148 (build_and_record_new_cond): Likewise.
28149 (record_conditions): Likewise. Accept vector of conditions rather
28150 than edge_equivalence structure for first argument.
28151 for the first argument.
28152 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28153 from tree-ssa-dom.c.
28154 (avail_exprs_stack::record_cond): Likewise.
28155 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
28156 from tree-ssa-dom.c.
28157 (avail_exprs_stack): Add new member functions lookup_avail_expr
28159 (record_conditions): Declare.
28161 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
28164 * lra-constraints.c (process_alt_operands): Increase reject for
28165 reloading an input/output operand.
28167 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28170 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
28171 insns to convert from signed/unsigned char/short to IEEE 128-bit
28173 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
28175 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
28178 * config/i386/i386.c (ix86_vector_duplicate_value): Create
28179 subreg of inner mode for values already in registers.
28181 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
28183 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
28184 iteration reg is used after the loop.
28186 2017-03-14 Martin Sebor <msebor@redhat.com>
28188 PR tree-optimization/79800
28189 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
28190 precision in negative-positive range.
28191 (format_floating): Call non-const overload with adjusted precision.
28193 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28196 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
28199 2017-03-14 Martin Sebor <msebor@redhat.com>
28201 PR middle-end/80020
28202 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
28203 * builtins.def (aligned_alloc): Use it.
28206 * Makefile.in (GTFILES): Add calls.c.
28207 * calls.c: Include "gt-calls.h".
28209 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
28211 PR rtl-optimization/79728
28212 * regs.h (struct target_regs): New field
28213 x_contains_allocatable_regs_of_mode.
28214 (contains_allocatable_regs_of_mode): New macro.
28215 * reginfo.c (init_reg_sets_1): Initialize it, and change
28216 contains_reg_of_mode so it includes global regs as well.
28217 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28218 rather than contains_regs_of_mode.
28220 2017-03-14 Martin Liska <mliska@suse.cz>
28222 * doc/invoke.texi: Document options that can't be combined with
28223 -fcheck-pointer-bounds.
28225 2017-03-14 Martin Liska <mliska@suse.cz>
28227 PR middle-end/79831
28228 * doc/invoke.texi (-Wchkp): Document the option.
28230 2017-03-14 Martin Liska <mliska@suse.cz>
28232 * Makefile.in: Install gcov-dump.
28234 2017-03-14 Martin Liska <mliska@suse.cz>
28236 * multiple_target.c (expand_target_clones): Bail out for
28237 an invalid attribute.
28239 2017-03-14 Richard Biener <rguenther@suse.de>
28241 * alias.c (struct alias_set_entry): Pack properly.
28242 * cfgloop.h (struct loop): Likewise.
28243 * cse.c (struct set): Likewise.
28244 * ipa-utils.c (struct searchc_env): Likewise.
28245 * loop-invariant.c (struct invariant): Likewise.
28246 * lra-remat.c (struct cand): Likewise.
28247 * recog.c (struct change_t): Likewise.
28248 * rtl.h (struct address_info): Likewise.
28249 * symbol-summary.h (function_summary): Likewise.
28250 * tree-loop-distribution.c (struct partition): Likewise.
28251 * tree-object-size.c (struct object_size_info): Likewise.
28252 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28253 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28254 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28255 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28256 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28257 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28258 (struct _stmt_vec_info): Likewise.
28260 2017-03-14 Martin Liska <mliska@suse.cz>
28263 * multiple_target.c (create_dispatcher_calls): Check that
28264 a target can create a function dispatcher.
28266 2017-03-14 Martin Liska <mliska@suse.cz>
28269 * multiple_target.c (expand_target_clones): Drop local.local
28270 flag for default implementation.
28272 2017-03-14 Richard Biener <rguenther@suse.de>
28274 PR tree-optimization/80030
28275 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28277 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28279 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28280 gcc_fallthrough() instead of __attribute__((fallthrough));
28282 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28284 * doc/gcc.texi: Remove "up" link to (DIR).
28285 * doc/gccint.texi: Ditto.
28287 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28289 * doc/install.texi (Specific) <avr>: Remove reference to
28292 2017-03-13 Jeff Law <law@redhat.com>
28294 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28295 attribute rather than comments.
28297 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28298 match_scratch operand is highest.
28300 2017-03-13 Martin Liska <mliska@suse.cz>
28302 PR middle-end/78339
28303 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28304 is a CHKP clone, use original declaration.
28306 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28308 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28309 (arc_conditional_register_usage): Use a different allocation order
28310 when optimizing for size.
28311 * common/config/arc/arc-common.c (arc_option_optimization_table):
28312 Section anchors default on when optimizing for size.
28314 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28316 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28318 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28320 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28321 * config/arc/arc.md (cpu_facility): Add cd variant.
28322 (*movqi_insn): Add code density variant.
28323 (*movhi_insn): Likewise.
28324 (*movqi_insn): Likewise.
28325 (*addsi3_mixed): Likewise.
28326 (subsi3_insn): Likewise.
28328 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28330 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28332 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28334 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28335 expressions with MINUS and UNARY ops.
28337 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28340 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28342 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28343 between vec_select and vector argument.
28344 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28345 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28346 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28347 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28348 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28349 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28351 2017-03-13 Richard Biener <rguenther@suse.de>
28354 * params.def (vect-max-peeling-for-alignment): Fix typo.
28356 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28358 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28359 issue that only occurred with binutils below 2.18.
28361 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28363 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28364 refer to binutils 2.11/2.12 minimum.
28366 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28368 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28369 ftp.kernel.org and simplify binutils requirement.
28371 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28373 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28375 (Optimize Options): Ditto. Also remove redundancy.
28377 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28379 PR translation/79848
28380 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28382 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28383 to G_ to avoid double translation.
28385 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28387 PR translation/79923
28388 * auto-profile.c (get_combined_location): Convert leading
28389 character of diagnostics to lower case and remove trailing period.
28390 (read_profile): Likewise for various diagnostics.
28391 * config/arm/arm.c (arm_option_override): Remove trailing period
28392 from various diagnostics.
28393 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28394 (msp430_expand_delay_cycles): Likewise.
28396 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28399 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28400 full command-line argument, rather than just "str".
28401 (aarch64_validate_march): Likewise.
28402 (aarch64_validate_mtune): Likewise.
28404 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28406 PR rtl-optimization/78911
28407 * lra-assigns.c (must_not_spill_p): New function.
28408 (spill_for): Use it.
28410 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28412 PR tree-optimization/79981
28413 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28414 ATOMIC_COMPARE_EXCHANGE ifn result.
28415 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28416 IFN_ATOMIC_COMPARE_EXCHANGE.
28418 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28421 * opts.c (parse_sanitizer_options): Add missing question mark to
28422 "did you mean" message.
28424 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28426 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28428 (VMULEUH_UNS): Likewise.
28429 (VMULOUB_UNS): Likewise.
28430 (VMULOUH_UNS): Likewise.
28431 * config/rs6000/rs6000.c (builtin_function_type): Remove
28432 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28434 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28437 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28438 x with result of extra_parsing_for_operand_code_0.
28439 (function_reader::extra_parsing_for_operand_code_0): Convert
28440 return type from void to rtx, returning x. When reading
28441 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28442 larger size containing struct block_symbol.
28444 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28446 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28447 -mfloat128-hardware without -m64.
28449 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28452 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28453 entries to the case statement that marks unsigned arguments to
28454 overloaded functions.
28456 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28458 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28459 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28461 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28464 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28465 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28467 2017-03-10 Martin Liska <mliska@suse.cz>
28471 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28472 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28475 2017-03-10 Olivier Hainque <hainque@adacore.com>
28477 * tree-switch-conversion (array_value_type): Start by resetting
28478 candidate type to it's main variant.
28480 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28482 PR rtl-optimization/79909
28483 * combine.c (try_combine): Use simplify_replace_rtx on individual
28484 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28485 of the whole CALL_INSN_FUNCTION_USAGE.
28487 PR tree-optimization/79972
28488 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28489 get_range_info on SSA_NAMEs. Formatting fixes.
28491 2017-03-10 Richard Biener <rguenther@suse.de>
28492 Jakub Jelinek <jakub@redhat.com>
28494 PR tree-optimization/77975
28495 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28496 edge to be constant.
28497 (get_val_for): For constant x return it. Formatting fix.
28498 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28499 would use the same bases as the current one.
28501 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28503 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28504 instead of vec_select for V1TImode.
28505 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28507 (VSX_LE_128): Add V1TI to this mode iterator.
28508 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28509 (*vsx_le_perm_store_<mode>): Likewise.
28510 (pre-reload splitter for VSX stores): Likewise.
28511 (post-reload splitter for VSX stores): Likewise.
28512 (*vsx_xxpermdi2_le_<mode>): Likewise.
28513 (*vsx_lxvd2x2_le_<mode>): Likewise.
28514 (*vsx_stxvd2x2_le_<mode>): Likewise.
28516 2017-03-09 Michael Eager <eager@eagercon.com>
28518 Correct failures with --enable-checking=yes,rtl.
28520 * config/microblaze/microblaze.c (microblaze_expand_shift):
28521 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28522 test for const0_rtx.
28523 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28524 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28526 2017-03-09 Richard Biener <rguenther@suse.de>
28528 PR tree-optimization/79977
28529 * graphite-scop-detection.c (scop_detection::merge_sese):
28530 Handle the case of extra exits to blocks dominating the entry.
28532 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28534 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28537 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28539 PR rtl-optimization/79949
28540 * lra-constraints.c (process_alt_operands): Check memory when
28541 trying to predict a cycle. Print about the overall increase.
28543 2017-03-09 Richard Biener <rguenther@suse.de>
28545 PR middle-end/79971
28546 * gimple-expr.c (useless_type_conversion_p): Preserve
28547 TYPE_SATURATING for fixed-point types.
28549 2017-03-09 Richard Biener <rguenther@suse.de>
28552 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28553 alignment of BLKmode params.
28555 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28558 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28559 (VALL_NO_V2Q): Likewise.
28561 * config/aarch64/aarch64-simd.md
28562 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28564 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28565 VALL_NO_V2Q mode iterator.
28566 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28568 2017-03-09 Martin Liska <mliska@suse.cz>
28570 PR tree-optimization/79631
28571 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28572 tree_int_cst_sign_bit just for INTEGER constants.
28574 2017-03-09 Martin Liska <mliska@suse.cz>
28578 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28581 2017-03-09 Marek Polacek <polacek@redhat.com>
28584 * tree.c (inchash::add_expr): Handle TREE_VEC.
28586 2017-03-09 Martin Liska <mliska@suse.cz>
28589 (chkp_narrow_size_and_offset): New function.
28590 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28591 (void chkp_parse_bit_field_ref): New function.
28592 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28593 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28595 2017-03-09 Martin Liska <mliska@suse.cz>
28598 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28599 (chkp_find_bounds_1): Remove gcc_unreachable.
28601 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28604 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28605 BUILT_IN_SYNC*, determine the access type from the size suffix and
28606 always build a MEM_REF with that type. Handle forgotten
28607 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28610 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28611 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28612 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28613 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28614 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28615 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28616 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28617 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28618 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28619 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28620 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28621 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28622 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28623 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28624 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28625 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28626 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28627 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28628 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28629 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28630 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28631 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28632 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28633 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28634 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28635 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28636 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28637 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28638 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28639 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28640 definitions outside of __OPTIMIZE__ guarded section.
28643 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28644 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28645 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28646 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28649 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28651 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28652 ("vfenez<mode>"): Add missing constraints.
28654 2017-03-08 Martin Sebor <msebor@redhat.com>
28657 * config/nds32/nds32.c (nds32_option_override):
28658 Fix misspelled diagnostic.
28660 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28663 * gimplify.c (gimplify_omp_for): Replace index var in outer
28664 taskloop statement with an artificial variable and add
28665 OMP_CLAUSE_PRIVATE clause for it.
28667 2017-03-08 Richard Biener <rguenther@suse.de>
28669 PR tree-optimization/79955
28670 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28671 for accesses that are completely outside of the variable.
28673 2017-03-08 Andrew Haley <aph@redhat.com>
28675 PR tree-optimization/79943
28676 * tree-ssa-loop-split.c (compute_new_first_bound): When
28677 calculating the new upper bound, (END-BEG) should be added, not
28680 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28682 * config/avr/avr.md (setmemhi): Make sure match_dup
28683 operand number comes before match_scratch.
28685 2017-03-08 Richard Biener <rguenther@suse.de>
28687 PR tree-optimization/79920
28688 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28689 with ncopies == 1 to ...
28690 (vect_transform_slp_perm_load): ... here. Properly compute
28691 all element loads by iterating VF times over the group. Do
28692 not handle ncopies (computed in a broken way) in
28693 vect_create_mask_and_perm.
28695 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28698 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28699 is a uniform vector, use uniform_vector_p return value instead of
28700 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28702 2017-03-07 Marek Polacek <polacek@redhat.com>
28704 PR middle-end/79809
28705 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28706 (alloca_call_type): Likewise.
28708 2017-03-07 Martin Liska <mliska@suse.cz>
28710 * gcov.c (process_args): Put comment to correct location.
28712 2017-03-07 Martin Liska <mliska@suse.cz>
28714 PR middle-end/68270
28715 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28716 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28717 (chkp_narrow_bounds_for_field): Likewise.
28718 (chkp_parse_array_and_component_ref): Pass one more argument to
28721 2017-03-07 Richard Biener <rguenther@suse.de>
28723 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28726 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28728 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28729 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28731 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28734 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28735 to end of description.
28736 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28738 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28740 PR rtl-optimization/79901
28741 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28743 (*avx512f_<code><mode>3<mask_name>): ... this.
28744 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28745 iterator instead of VI8_AVX2_AVX512BW.
28747 PR rtl-optimization/79901
28748 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28749 min/max expander, expand it using expand_vec_cond_expr.
28752 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28755 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28758 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28759 to void * for PCH reasons.
28760 * dwarf2out.c (output_loc_operands, output_die): Cast
28761 v.val_vec.array to unsigned char *.
28763 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28766 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28769 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28771 PR rtl-optimization/79571
28772 * lra-constraints.c (process_alt_operands): Calculate static
28773 reject and subtract it from overall when only addresses will be
28776 2017-03-06 Julia Koval <julia.koval@intel.com>
28779 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28780 incoming stack boundary to 128 for 64-bit targets.
28782 2017-03-06 Richard Biener <rguenther@suse.de>
28784 PR tree-optimization/79894
28785 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28786 to NULL after folding it.
28788 2017-03-06 Richard Biener <rguenther@suse.de>
28790 PR tree-optimization/79824
28791 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28792 check disabling peeling for gaps.
28794 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28796 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28797 attributes): Document gettimeofday.
28799 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28801 * config/s390/s390.c (s390_option_override_internal): Set
28802 PARAM_MIN_VECT_LOOP_BOUND
28804 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28806 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28807 * config/s390/s390.md: Likewise.
28809 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28812 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28813 (<avx2_avx512>_perm<mode>): Rename to ...
28814 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28816 (<avx512>_perm<mode>_mask): Rename to ...
28817 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28819 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28820 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28821 instead of VI8F_256_512.
28822 (avx512f_perm<mode>): New define_expand.
28823 (avx512f_perm<mode>_mask): Likewise.
28824 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28825 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28827 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28829 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28830 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28832 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28834 2017-03-06 Martin Liska <mliska@suse.cz>
28837 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28838 when having a SSA NAME w/o VAR_DECL assigned to it.
28840 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28842 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28843 msa_dpsub_<su>_d): Fix MODE for vec_select.
28845 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28847 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28849 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28851 2017-03-06 Richard Biener <rguenther@suse.de>
28853 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28854 * plugin.c (register_plugin_info): Likewise.
28855 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28857 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28859 * config/i386/sse.md (sse_storehps, sse_storelps,
28860 avx_<castmode><avxsizesuffix>_<castmode>,
28861 avx512f_<castmode><avxsizesuffix>_<castmode>,
28862 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28863 in condition that at least one operand is not a MEM.
28865 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28867 PR middle-end/79805
28868 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28869 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28871 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28872 gimple_call_nothrow_p flag based on whether original builtin can throw.
28873 If it can, emit following stmts on the fallthrough edge.
28874 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28875 don't create new bb if inserting just debug stmts on the edge, try to
28876 insert them on the fallthru bb or just reset debug stmts.
28878 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28881 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28882 restore recog_data (including the operand rtxes inside it) around
28883 the call to get_insn_template.
28885 2017-03-03 Martin Sebor <msebor@redhat.com>
28887 PR tree-optimization/79699
28888 * context.c (context::~context): Free MPFR caches to avoid
28889 a memory leak on program exit.
28891 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28893 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28894 Use wide_int::ulow () instead of .elt (0).
28896 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28898 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28899 (*pushxf): Limit oF constraint to 32bit targets and add oC
28900 constraint for 64bit targets.
28901 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28902 (*pushdf): Change rmF constraint to rmC.
28904 2017-03-03 Martin Liska <mliska@suse.cz>
28906 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28907 Remove unused variable.
28909 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28912 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28913 is a memory operand, increase num_memory.
28914 (ix86_expand_args_builtin): Likewise.
28916 2017-03-03 Jan Hubicka <jh@suse.cz>
28919 * ipa-devirt.c (maybe_record_node): Properly handle
28920 __cxa_pure_virtual visibility.
28922 2017-03-03 Martin Liska <mliska@suse.cz>
28924 PR tree-optimization/79803
28925 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28927 (pass_loop_prefetch::execute): Disabled optimization if an
28928 assumption about L1 cache size is not met.
28930 2017-03-03 Martin Liska <mliska@suse.cz>
28932 PR rtl-optimization/79574
28933 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28934 (hash_scan_set): Likewise.
28935 (dump_hash_table): Likewise.
28936 (hoist_code): Likewise.
28938 2017-03-03 Richard Biener <rguenther@suse.de>
28940 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28941 in place of uhigh ().
28942 (fixed_convert_from_real): Likewise.
28944 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28947 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28949 2017-03-03 Richard Biener <rguenther@suse.de>
28951 PR middle-end/79818
28952 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28953 TYPE_OVERFLOW_UNDEFINED check.
28955 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28957 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28959 (vector_ae_<mode>_p): Likewise.
28960 (vector_nez_<mode>_p): Likewise.
28961 (vector_ne_v2di_p): Likewise.
28962 (vector_ae_v2di_p): Likewise.
28963 (vector_ne_<mode>_p): Likewise.
28964 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28966 (vsx_tsqrt<mode>2_fe): Likewise.
28968 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
28971 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
28973 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28975 PR rtl-optimization/79780
28976 * cprop.c (one_cprop_pass): When second and further conditional trap
28977 in a single basic block is turned into an unconditional trap, turn it
28978 into a deleted note to avoid RTL verification failures.
28980 2017-03-02 Richard Biener <rguenther@suse.de>
28982 * fold-const.c (const_binop): Use ulow () instead of elt (0).
28984 2017-03-02 Richard Biener <rguenther@suse.de>
28986 PR tree-optimization/79345
28988 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
28989 param and abort the walk, returning -1 if it is hit.
28990 (walk_aliased_vdefs): Take a limit param and pass it on.
28991 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
28992 defaulting to 0 and return a signed int.
28993 * tree-ssa-uninit.c (struct check_defs_data): New struct.
28994 (check_defs): New helper.
28995 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
28996 about uninitialized memory.
28997 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
28998 bogus uninitialized warning.
28999 (fixed_convert_from_real): Likewise.
29001 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29003 PR tree-optimization/66768
29004 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29005 iv_use if base object can't be determined.
29007 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29009 PR tree-optimization/79345
29010 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29011 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29012 (get_pattern_stats): Initialize it.
29013 * genemit.c (gen_expand): Verify match_scratch numbers come after
29014 match_operand/match_dup numbers.
29015 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29016 match_scratch numbers.
29017 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29019 * config/s390/s390.md (trunctdsd2): Likewise.
29021 2017-03-02 Richard Biener <rguenther@suse.de>
29023 * wide-int.h (wide_int_storage::operator=): Implement in terms
29026 2017-03-02 Richard Biener <rguenther@suse.de>
29028 PR tree-optimization/79777
29029 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29030 the to insert expression to sth existing.
29032 2017-03-01 Martin Sebor <msebor@redhat.com>
29034 PR middle-end/79692
29035 * gimple-ssa-sprintf.c
29036 (directive::known_width_and_precision): New function.
29037 (format_integer): Use it.
29038 (get_mpfr_format_length): Consider the full range of precision
29039 when computing %g output with the # flag. Set the likely byte
29040 count to 3 rather than 1 when precision is indeterminate.
29041 (format_floating): Correct the lower bound of precision.
29043 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29045 * doc/invoke.texi: Document default code model for 64-bit Linux.
29047 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29050 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29051 udiv rather than div since input pattern is unsigned.
29053 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29055 * config/i386/i386.c (print_reg): Warn for values of
29056 unsupported size in integer register.
29058 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29061 * config/rs6000/predicates.md (current_file_function_operand): Do
29062 not allow self calls to be local if the function is replaceable.
29064 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29067 * config/rs6000/altivec.h (vec_ctz and others): Change the
29068 preprocessor macro that controls conditional compilation from
29069 _ARCH_PWR9 to __POWER9_VECTOR__.
29070 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29071 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29072 control (instead of _ARCH_PWR9 control) so that template
29073 definition uses power9-specific function.
29074 (vec_any_eq): Likewise.
29075 (vec_all_ne): Change macro definition to use a power9-specific
29076 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29077 _ARCH_PWR9 control).
29078 (vec_any_eq) Likewise.
29079 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29080 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29081 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29082 support for xvcmpnedp instruction.
29083 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29084 macro expansion so that Power9 implementation of vec_all_ne does
29085 not use the AltiVec predicate framework.
29086 (VCMPNEH_P): Likewise.
29087 (VCMPNEW_P): Likewise.
29088 (VCMPNED_P): Likewise.
29089 (VCMPNEFP_P): Likewise.
29090 (VCMPNEDP_P): Likewise.
29091 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29092 implementation of vec_any_eq to not use AltiVec predicate
29094 (VCMPAEH_P): Likewise.
29095 (VCMPAEW_P): Likewise.
29096 (VCMPAED_P): Likewise.
29097 (VCMPAEFP_P): Likewise.
29098 (VCMPAEDP_P): Likewise.
29099 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29100 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29101 not use the AltiVec predicate framework.
29102 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29103 of vec_any_eq to not use AltiVec predicate framework.
29104 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29105 support for predefined __POWER9_VECTOR__ macro to indicate that
29106 Power9 instruction selection is enabled.
29107 (altivec_overloaded_builtins): Remove extraneous
29108 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29109 function argument types RS6000_BTI_bool_V16QI and
29110 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29111 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29112 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29113 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29114 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29115 Power9 for implementations of vec_cmpne. Change the signature for
29116 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29117 (representing vec_all_ne) to remove the previously described first
29118 argument of type RS6000_BTI_INTSI, as this was an artifact of
29119 reliance on the AltiVec predicate framework, which is no longer
29120 used in the implementation of these functions. Add
29121 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29122 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29123 since, unlike the AltiVec predicate framework implementation, we
29124 do not share function descriptors between vec_alle and vec_anyeq.
29125 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29126 set of modes that receive special treatment even when
29127 TARGET_P9_VECTOR is true. The special treatment emits code that
29128 does not depend on Power9 instructions.
29129 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29130 define_expand to not rely on AltiVec predicate framework.
29131 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29133 (vector_ne_v2di_p): Change this define_expand to not rely on
29134 AltiVec predicate framework.
29135 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29137 (vector_ne_<mode>_p): Change this define_expand to not rely on
29138 AltiVec predicate framework.
29139 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29141 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29142 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29143 define_insn pattern.
29144 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29145 define_insn pattern because the xvcmpne<VSs>. instruction is not
29147 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29148 instruction is not supported.
29150 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29152 * config/nvptx/nvptx.c: Include intl.h.
29154 2017-03-01 Martin Jambor <mjambor@suse.cz>
29157 * ipa-prop.h (ipa_bits): Removed field known.
29158 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
29159 to pointers. Adjusted their comments to warn about their sharing.
29160 (ipcp_transformation_summary): Change bits to a vector of pointers.
29161 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
29162 (ipa_get_ipa_bits_for_value): Declare.
29163 * tree-vrp.h (value_range): Mark as GTY((for_user)).
29164 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
29165 (ipa_bits_hash_table): Likewise.
29166 (ipa_vr_ggc_hash_traits): Likewise.
29167 (ipa_vr_hash_table): Likewise.
29168 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
29169 being pointers and vr_known being removed.
29170 (ipa_set_jf_unknown): Likewise.
29171 (ipa_get_ipa_bits_for_value): New function.
29172 (ipa_set_jfunc_bits): Likewise.
29173 (ipa_get_value_range): New overloaded functions.
29174 (ipa_set_jfunc_vr): Likewise.
29175 (ipa_compute_jump_functions_for_edge): Use the above functions to
29176 construct bits and vr parts of jump functions.
29177 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
29178 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29180 (ipcp_grow_transformations_if_necessary): Also allocate
29181 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29183 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
29184 them. Fix too long lines.
29185 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
29186 vr_known being removed.
29187 (ipa_read_jump_function): Use new setter functions to construct bits
29188 and vr parts of jump functions or set them to NULL.
29189 (write_ipcp_transformation_info): Adjust for bits being pointers.
29190 (read_ipcp_transformation_info): Likewise.
29191 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
29193 Include gt-ipa-prop.h.
29194 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
29196 (ipcp_store_bits_results): Likewise.
29197 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
29198 Do not write to existing jump functions but use a temporary instead.
29200 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29203 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
29204 attempt to use its first operand as BIT_FIELD_REF base.
29206 2017-03-01 Richard Biener <rguenther@suse.de>
29208 PR middle-end/79721
29209 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
29210 interpolating formula in wrapping arithmetic.
29211 (chrec_apply): Convert chrec_evaluate return value to wanted type.
29213 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29215 PR tree-optimization/79734
29216 * tree-vect-generic.c (expand_vector_condition): Optimize
29217 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29218 Handle VEC_COND_EXPR where comparison has different inner width from
29219 type's inner width.
29221 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29223 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29224 markup, and similar issues. Remove @opindex entries for things
29225 that aren't options. Add missing -mmpy-option entries.
29227 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29229 PR tree-optimization/79737
29230 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29231 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29232 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29233 instead of byte_size. Formatting fix.
29234 (shift_bytes_in_array_right): Formatting fix.
29236 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29239 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29240 condition on optimize for the leaf function test.
29242 2017-02-28 Martin Liska <mliska@suse.cz>
29245 * read-rtl-function.c (function_reader::handle_unknown_directive):
29246 Bail out when one uses -flto.
29248 2017-02-28 Martin Liska <mliska@suse.cz>
29250 * common.opt: Replace space with tabular for options of <number>
29252 * config/i386/i386.opt: Show <number> value for
29253 -mlarge-data-threshold.
29254 * opts.c (print_filtered_help): Do not display number in hexadecimal
29257 2017-02-28 Martin Liska <mliska@suse.cz>
29259 * common.opt: Fix --help=option -Q for options which are of
29262 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29264 * config/i386/i386.c (print_reg): Error out for values
29265 of 8-bit size in invalid integer register.
29267 2017-02-28 Martin Sebor <msebor@redhat.com>
29269 PR tree-optimization/79691
29270 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29272 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29275 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29276 gcc_unreachable with output_operand_lossage.
29278 2017-02-28 Richard Biener <rguenther@suse.de>
29280 PR tree-optimization/79740
29281 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29283 (visit_nary_op): Insert the nary into the hashtable if we
29284 pattern-matched sth.
29285 * tree-ssa-pre.c (eliminate_insert): Robustify.
29287 2017-02-28 Richard Biener <rguenther@suse.de>
29289 PR middle-end/79731
29290 * fold-const.c (decode_field_reference): Reject out-of-bound
29293 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29295 * config/i386/i386.c: Include intl.h.
29296 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29297 instead of just cond ? "..." : "...".
29298 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29299 * coverage.c (read_counts_file): Likewise.
29300 * omp-offload.c: Include intl.h.
29301 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29302 of just cond ? "..." : "...".
29303 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29304 of just cond ? "..." : "...".
29306 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29309 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29311 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29312 'tune for' CPU name.
29313 * config/arm/arm-cpu-data.h: Regenerated.
29315 2017-02-28 Richard Biener <rguenther@suse.de>
29317 PR tree-optimization/79732
29318 * tree-inline.c (expand_call_inline): Do not shadow var.
29320 2017-02-28 Richard Biener <rguenther@suse.de>
29322 PR tree-optimization/79723
29323 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29324 address-space properly.
29326 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29328 * doc/optinfo.texi (Optimization groups): Fix option used for
29330 * doc/invoke.texi (-fopt-info): Document "omp".
29331 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29332 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29333 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29334 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29335 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29337 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29339 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29340 OPTGROUP_OMP with "omp".
29342 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29345 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29346 for arithmetic shift of unsigned V2DI.
29348 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29350 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29351 arc/linux.h headers.
29352 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29353 (LINK_SPEC): Likewise.
29354 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29355 (EXTRA_SPECS): Likewise.
29356 (STARTFILE_SPEC): Likewise.
29357 (ENDFILE_SPEC): Likewise.
29358 (LIB_SPEC): Likewise.
29359 (TARGET_SDATA_DEFAULT): Likewise.
29360 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29361 (MULTILIB_DEFAULTS): Likewise.
29362 (DWARF2_UNWIND_INFO): Likewise.
29363 * config/arc/big.h: New file.
29364 * config/arc/elf.h: Likewise.
29365 * config/arc/linux.h: Likewise.
29366 * config/arc/t-uClibc: Remove.
29368 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29370 PR tree-optimization/77536
29371 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29372 (tree_transform_and_unroll_loop): Use above function to compute the
29373 estimated niter of unrolled loop and use it when scaling profile.
29374 Also use count info rather than frequency if it's non-zero.
29375 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29376 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29377 (vect_transform_loop): Call above function.
29379 2017-02-27 Richard Biener <rguenther@suse.de>
29381 PR tree-optimization/45397
29382 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29383 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29384 (visit_nary_op): Add pattern matching for CSEing sign-changed
29385 or truncated operations with wider ones.
29387 2017-02-27 Richard Biener <rguenther@suse.de>
29389 PR tree-optimization/79690
29390 * tree-vect-stmts.c (vectorizable_store): Use vector type
29391 built from the DR with address-space.
29393 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29395 * doc/invoke.texi (Optimize Options): Refine the description
29396 of asan-use-after-return.
29398 2017-02-25 Alan Modra <amodra@gmail.com>
29400 PR rtl-optimization/79584
29401 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29402 base, not ad->base_term, the reg within base. Remove assertion
29403 that ad->base == ad->base_term. Replace gen_int_mode using
29404 bogus mode with const0_rtx.
29406 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29408 PR middle-end/79396
29409 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29410 FMA_EXPR like tcc_binary or tcc_unary.
29412 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29415 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29417 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29418 (output_loc_operands): Handle DW_OP_call_ref and
29419 DW_OP_GNU_variable_value.
29420 (struct variable_value_struct): New type.
29421 (struct variable_value_hasher): Likewise.
29422 (variable_value_hash): New variable.
29423 (string_types): Remove.
29424 (copy_loc_descr): New function.
29425 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29426 (prepend_loc_descr_to_each): New function.
29427 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29428 instead of add_loc_descr_to_each if the first argument is single
29429 location list and the second has multiple.
29430 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29431 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29432 when looking for variable value which doesn't have other location info.
29433 (loc_list_from_tree): Formatting fix.
29434 (gen_array_type_die): Simplify DW_AT_string_length handling.
29435 (adjust_string_types): Remove.
29436 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29437 string_types. Call resolve_variable_values.
29438 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29439 (resolve_addr_in_expr): Likewise. Add A argument.
29440 (copy_deref_exprloc): Remove deref argument. Adjust for the
29441 original expression being DW_OP_GNU_variable_value with optionally
29442 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29443 optionally after it.
29444 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29445 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29446 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29447 (variable_value_hasher::hash, variable_value_hasher::equal): New
29449 (resolve_variable_value_in_expr, resolve_variable_value,
29450 resolve_variable_values, note_variable_value_in_expr,
29451 note_variable_value): New functions.
29452 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29455 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29458 * opts.h (handle_generated_option): Add GENERATED_P argument.
29459 * opts-common.c (handle_option): Adjust function comment.
29460 (handle_generated_option): Add GENERATED_P argument, pass it to
29462 (control_warning_option): Pass false to handle_generated_option
29464 * opts.c (maybe_default_option): Pass true to handle_generated_option
29466 * optc-gen.awk: Likewise.
29468 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29470 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29471 a REG, look at the REG it is a SUBREG of.
29472 (splitter for cmpeqsi_t): Ditto.
29474 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29476 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29477 the special USEs with the pattern of the insn, not the insn itself.
29479 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29482 * doc/invoke.texi: Document -mload-store-pairs.
29484 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29485 Sandra Loosemore <sandra@codesourcery.com>
29487 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29488 argument isn't a CONST_INT.
29489 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29490 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29491 (nios2_validate_compare): Bypass alternate compare logic if *op2
29492 is not a CONST_INT.
29493 (ldstwm_operation_p): Return false if first_base is not a REG or
29494 if first_offset is not a CONST_INT.
29496 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29498 * config/cris/cris.md: Use correct operand in a define_peephole2.
29500 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29502 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29504 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29506 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29507 this_insn if it is an INSN or JUMP_INSN.
29508 (force_offsettable): Look at base, not at addr.
29509 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29510 on things that aren't necessarily CONST_INTs.
29512 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29514 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29515 -mfpmath=sse is the default also for x86-32 targets with SSE2
29516 instruction set when @option{-ffast-math} is enabled
29518 2017-02-24 Jeff Law <law@redhat.com>
29520 PR rtl-optimizatoin/79286
29521 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29524 2017-02-24 Richard Biener <rguenther@suse.de>
29526 PR tree-optimization/79389
29527 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29530 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29532 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29533 function comment to reflect reality.
29534 (loop_exits_before_overflow): Fix typo in function description.
29536 2017-02-24 Richard Biener <rguenther@suse.de>
29538 PR tree-optimization/79389
29539 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29540 properly that a threading opportunity exists. Detect conditional
29541 copy/constant propagation opportunities.
29543 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29545 * config/visium/visium.md (type): Add trap.
29546 (b): New mode attribute.
29547 (*btst): Rename into...
29548 (*btst<mode>): ...this and adjust.
29549 (*cbranchsi4_btst_insn): Rename into...
29550 (*cbranch<mode>4_btst_insn): ...this and adjust.
29551 (trap): New define_insn.
29553 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29555 PR tree-optimization/79389
29556 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29557 (noce_reversed_cond_code): New function.
29558 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29559 reversed_comparison_code. Formatting fix.
29560 (noce_try_store_flag): Test rev_cond != NULL in addition to
29561 reversed_comparison_code.
29562 (noce_try_store_flag_constants): Likewise.
29563 (noce_try_store_flag_mask): Likewise.
29564 (noce_try_addcc): Use rev_cond if non-NULL instead of
29565 reversed_comparison_code.
29566 (noce_try_cmove_arith): Likewise. Formatting fixes.
29567 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29568 (noce_find_if_block): Initialize rev_cond.
29569 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29570 instead of false as last argument never attempt to reverse it
29573 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29575 PR tree-optimization/79663
29576 * tree-predcom.c (combine_chains): Process refs in reverse order
29577 only for ZERO length chains, and add explaining comment.
29579 2017-02-23 Jeff Law <law@redhat.com>
29581 PR tree-optimization/79578
29582 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29583 in call to operand_equal_p.
29585 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29588 * config/i386/cpuid.h: Fix another undefined behavior.
29590 2017-02-23 Richard Biener <rguenther@suse.de>
29592 PR tree-optimization/79683
29593 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29594 vector types for data-refs.
29596 2017-02-23 Martin Liska <mliska@suse.cz>
29598 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29600 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29602 PR middle-end/79665
29603 * internal-fn.c (get_range_pos_neg): Moved to ...
29604 * tree.c (get_range_pos_neg): ... here. No longer static.
29605 * tree.h (get_range_pos_neg): New prototype.
29606 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29607 are known to be in between 0 and signed maximum inclusive, try to
29608 expand both unsigned and signed divmod and use the cheaper one from
29611 2017-02-22 Jeff Law <law@redhat.com>
29613 PR tree-optimization/79578
29614 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29615 to compare base operands.
29617 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29620 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29621 gpc_reg_operand instead of fpr_reg_operand.
29623 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29625 * config/mips/mips.c (mips_return_in_memory): Force FP
29626 vector types to be returned in memory for o32 ABI.
29628 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29630 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29631 instead of DW_TAG_member for static data member declarations and don't
29632 set no_linkage_name for static inline data members.
29633 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29636 2017-02-22 Martin Liska <mliska@suse.cz>
29638 * doc/invoke.texi: Replace inequality signs with square brackets
29641 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29644 * lra-constraints.c (simplify_operand_subreg): Handle
29645 WORD_REGISTER_OPERATIONS targets.
29647 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29650 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29651 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29652 elimination by swapping fld*.
29654 2017-02-22 Richard Biener <rguenther@suse.de>
29656 PR tree-optimization/79673
29657 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
29658 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
29659 irrelevant address-space qualifiers and avoiding a
29660 ADDR_SPACE_CONVERT_EXPR from fold_convert.
29662 2017-02-22 Richard Biener <rguenther@suse.de>
29664 PR tree-optimization/79666
29665 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
29666 to not symbolically negate if that may introduce undefined
29669 2017-02-22 Martin Liska <mliska@suse.cz>
29672 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29673 * data-streamer-out.c (streamer_write_gcov_count_stream):
29675 * value-prof.c (stream_out_histogram_value): Make assert more
29676 precise based on type of counter.
29678 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29681 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29682 Use nonimmediate_operand instead of memory_operand for operand 1.
29683 (float-extend standard_x87sse_constant_load splitter): Ditto.
29685 2017-02-21 Jeff Law <law@redhat.com>
29687 PR tree-optimization/79621
29688 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29689 blocks with edges to themselves.
29691 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29694 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29695 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29696 Use gimple_call_builtin_p.
29699 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29700 on temporarily removed DEBUG_INSNs.
29702 PR tree-optimization/79649
29703 * tree-loop-distribution.c (classify_partition): Give up on
29704 non-generic address space loads/stores.
29706 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29708 * doc/loop.texi (Loop manipulation): Remove nonexistent
29709 tree_ssa_loop_version from the documentation.
29710 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29712 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29715 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29716 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29717 * config/rs6000/rs6000.c: Include except.h.
29718 (rs6000_expand_split_stack_prologue): Call
29719 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29721 2017-02-21 Martin Jambor <mjambor@suse.cz>
29724 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29725 have been analyzed.
29727 2017-02-21 Martin Jambor <mjambor@suse.cz>
29729 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29730 for backward compatibility only.
29731 * doc/invoke.texi (Option Summary): Remove all references to
29732 -fipa-cp-alignment.
29734 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29738 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29740 * lra-constraints.c (curr_insn_transform): Handle
29741 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29743 2017-02-21 Martin Liska <mliska@suse.cz>
29745 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29747 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29749 PR translation/79638
29750 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29752 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29755 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29756 (arm_function_ok_for_sibcall): Return false for an indirect call by
29757 descriptor if all the argument registers are used.
29758 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29759 alignment of the function.
29761 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29763 PR tree-optimization/61441
29764 * simplify-rtx.c (simplify_const_unary_operation): For
29765 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29766 the sNaN unmodified.
29768 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29770 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29771 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29772 instead of SYSTEM_HEADER_DIR.
29774 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29775 Martin Liška <mliska@suse.cz>
29777 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29778 Fix typos and grammar, use active voice, and clarify.
29780 2017-02-20 Marek Polacek <polacek@redhat.com>
29782 PR middle-end/79537
29783 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29786 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29788 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29791 * config/i386/i386.c (ix86_expand_builtin): Handle
29792 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29793 ix86_builtins_isa[fcode].isa as a requirement of those
29794 flags and any other flag in the bitmask.
29795 (ix86_init_mmx_sse_builtins): Use 0 instead of
29796 ~OPTION_MASK_ISA_64BIT as mask.
29797 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29798 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29799 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29800 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29802 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29805 * lra-constraints.c (split_reg): Check requested split mode
29806 is supported by the register.
29808 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29810 * lra-constraints.c (simplify_operand_subreg): Remove early
29813 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29816 * lra-constraints.c (curr_insn_transform): Tighten condition
29817 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29819 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29822 * lra-constraints.c (curr_insn_transform): Handle
29823 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29825 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29828 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29830 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29832 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29835 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29838 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29840 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29843 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29845 * final.c (last_columnnum, override_columnnum): New variables.
29846 (final_start_function): Set last_columnnum, pass it to begin_prologue
29847 hook and pass 0 to dwarf2out_begin_prologue.
29848 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29849 to source_line debug hook.
29850 (notice_source_line): Compute last_columnnum and for debug_column_info
29851 return true on column changes.
29852 * debug.h (struct gcc_debug_hooks): Add column argument to
29853 source_line and begin_prologue hooks.
29854 (debug_nothing_int_charstar_int_bool): Remove prototype.
29855 (debug_nothing_int_int_charstar,
29856 debug_nothing_int_int_charstar_int_bool): New prototypes.
29857 (dwarf2out_begin_prologue): Add column argument.
29858 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29859 begin_prologue hooks.
29860 (debug_nothing_int_charstar_int_bool): Remove.
29861 (debug_nothing_int_int_charstar,
29862 debug_nothing_int_int_charstar_int_bool): New functions.
29863 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29864 through to dwarf2out_source_line.
29865 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29866 (dwarf2out_source_line): Add column argument, emit it if requested.
29867 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29869 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29870 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29871 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29872 through to dwarf2out_begin_prologue.
29873 (vmsdbgout_source_line): Add column argument, pass it through to
29874 dwarf2out_source_line.
29875 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29876 dbxout_source_line caller.
29877 (dbxout_source_line): Add column argument.
29879 * common.opt (gno-column-info, gcolumn-info): New options.
29880 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29881 (check_die): Also test for multiple DW_AT_decl_column attributes.
29882 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29883 DW_AT_decl_column if requested.
29884 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29886 (gen_variable_die): Likewise.
29887 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29888 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29891 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29892 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29893 (ix86_handle_option): Handle OPT_m3dnowa.
29894 * doc/invoke.texi (-m3dnowa): Document.
29895 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29896 -m3dnowa instead of -m3dnow -march=athlon.
29899 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29900 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29902 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29905 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29906 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29907 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29908 generator for vsx_xxpermdi_<mode>_be.
29909 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29910 force big-endian semantics.
29911 (vsx_xxpermdi_<mode>_be): New define_expand with same
29912 implementation as previous version of vsx_xxpermdi_<mode>.
29914 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29916 PR tree-optimization/79327
29917 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29918 variable, its initialization and use.
29920 2017-02-17 Julia Koval <julia.koval@intel.com>
29922 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29923 (OPTION_MASK_ISA_PKU_UNSET): New.
29924 (ix86_handle_option): Handle -mrdpid.
29925 * config/i386/cpuid.h (bit_RDPID): New.
29926 * config/i386/driver-i386.c (host_detect_local_cpu):
29927 Detect RDPID feature.
29928 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29929 * config/i386/i386-c.c (ix86_target_macros_internal):
29931 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29932 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29933 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29934 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29935 * config/i386/i386.md (define_insn "rdpid"): New.
29936 * config/i386/i386.opt Add -mrdpid.
29937 * config/i386/immintrin.h (_rdpid_u32): New.
29939 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29941 PR rtl-optimization/79541
29942 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29943 instead of transforming it into USE.
29945 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29947 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29948 If HONOR_SNANS (SFmode) force the input to a register.
29949 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29950 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29951 an frsp or similar insn.
29953 2017-02-17 Martin Liska <mliska@suse.cz>
29955 PR rtl-optimization/79577
29956 * params.def (selsched-max-sched-times): Increase minimum to 1.
29958 2017-02-17 Martin Liska <mliska@suse.cz>
29960 PR rtl-optimization/79574
29961 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29963 2017-02-17 Martin Liska <mliska@suse.cz>
29965 PR tree-optimization/79529
29966 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
29967 ssa_defined_default_def_p to handle cases which are implicitly
29969 * tree-ssa.c (ssa_defined_default_def_p): New function.
29970 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
29971 which are implicitly defined.
29972 * tree-ssa.h (ssa_defined_default_def_p): Declare.
29974 2017-02-17 Richard Biener <rguenther@suse.de>
29976 PR middle-end/79576
29977 * params.def (max-ssa-name-query-depth): Limit to 10.
29979 2017-02-17 Richard Biener <rguenther@suse.de>
29981 PR tree-optimization/79552
29982 * tree-ssa-structalias.c (visit_loadstore): Properly verify
29985 2017-02-17 Richard Biener <rguenther@suse.de>
29988 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
29990 2017-02-17 Marek Polacek <polacek@redhat.com>
29992 PR middle-end/79536
29993 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
29994 (fold_negate_expr): New wrapper.
29996 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
29998 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
29999 Correct terminology and de-emphasize pre-standard behavior.
30001 2017-02-16 Alan Modra <amodra@gmail.com>
30003 PR rtl-optimization/79286
30004 * ira.c (def_dominates_uses): New function.
30005 (update_equiv_regs): Don't create an equivalence for insns that
30006 may trap where the register def does not dominate the use.
30008 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30010 PR rtl-optimization/78127
30011 * lra.c (lra): Call lra_eliminate before finish the loop after
30014 2017-02-16 Richard Biener <rguenther@suse.de>
30016 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30018 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30019 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30020 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30021 (isl_val_int_from_wi): New function.
30022 (extract_affine_gmp): Rename to ...
30023 (extract_affine_wi): ... this, take a widest_int.
30024 (extract_affine_int): Just wrap extract_affine_wi.
30025 (add_param_constraints): Use isl_val_int_from_wi.
30026 (add_loop_constraints): Likewise, and extract_affine_wi.
30028 2017-02-15 Jeff Law <law@redhat.com>
30030 PR middle-end/79521
30031 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30032 ira_init_register_move_cost_if_necessary.
30034 2017-02-15 Martin Sebor <msebor@redhat.com>
30036 PR middle-end/32003
30037 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30038 removed in a prior commit.
30040 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30042 PR tree-optimization/79347
30043 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30044 counters during peeling.
30046 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30048 * Makefile.in (site.exp): Remove "set ISLVER".
30050 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30053 * real.c (real_from_integer): Call real_convert even for decimal.
30055 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30058 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30060 2017-02-14 Andrew Pinski <apinski@cavium.com>
30062 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30063 cores and change the partno/implementer to be correct.
30064 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30065 the 'B" as the implementer.
30066 * config/aarch64/aarch64-tune.md: Regenerate.
30068 2017-02-14 Carl Love <cel@us.ibm.com>
30070 * config/rs6000/rs6000.c: Add case statement entry to make the
30071 xvcvuxdsp built-in argument unsigned.
30072 * config/rs6000/vsx.md: Fix the source and return operand types so they
30073 match the instruction definitions from the ISA document. Fix typo
30074 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30077 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30080 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30081 member early_clobber_alts.
30082 * lra-lives.c (reg_early_clobber_p): New.
30083 (process_bb_lives): Use it.
30084 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30085 (debug_operand_data): Initialize early_clobber_alts.
30086 (setup_operand_alternative): Set up early_clobber_alts.
30087 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30088 alternatives to new_insn_reg.
30089 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30091 (lra_update_insn_regno_info): Pass the new arg.
30093 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30095 PR middle-end/79505
30096 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30097 (new_oacc_loop_raw): Don't clear already cleared fields.
30100 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30101 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30102 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30104 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30107 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30109 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30112 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30113 the extra instruction to the right place to store 128-bit constant
30116 2017-02-14 Martin Sebor <msebor@redhat.com>
30118 PR middle-end/79448
30119 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30120 warning for strings of unknown length.
30122 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30124 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30126 2017-02-14 Jeff Law <law@redhat.com>
30129 * ira-costs.c (scan_one_insn): Initialize register move costs
30130 for pseudos seen in USE/CLOBBER insns.
30132 PR tree-optimization/79095
30133 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30134 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30135 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30136 if the operands are known to be not equal, then the resulting range
30138 (intersect_ranges): If the new range is ~[0,0] and the old range is
30139 wide, then prefer ~[0,0].
30140 * tree-vrp.c (overflow_comparison_p_1): New function.
30141 (overflow_comparison_p): New function.
30142 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30143 if NAME is used in an overflow test.
30144 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30145 overflow check that can be expressed as an equality test, then adjust
30146 ops to be that equality test.
30148 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30150 * config/s390/s390-builtin-types.def: Remove flags argument.
30151 * config/s390/s390.c (s390_init_builtins): Likewise.
30153 2017-02-14 Martin Liska <mliska@suse.cz>
30155 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
30156 vector. Fix trailing white spaces.
30158 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
30160 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
30163 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30165 PR rtl-optimization/68664
30166 * config/arm/arm.c (arm_sched_can_speculate_insn):
30167 New function. Declare prototype.
30168 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30170 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30172 PR rtl-optimization/68664
30173 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
30175 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30177 2017-02-14 Amit Pawar <amit.pawar@amd.com>
30179 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
30180 max skip bytes for function, loop and jump.
30182 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30184 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
30185 ABS_EXPR for gimple dump.
30187 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30190 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
30192 PR tree-optimization/79408
30193 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
30194 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
30195 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
30196 also if rhs1 is INTEGER_CST.
30198 2017-02-14 Richard Biener <rguenther@suse.de>
30200 PR middle-end/79432
30201 * tree-into-ssa.c (insert_phi_nodes): When the function can
30202 have abnormal edges rewrite SSA names with broken use-def
30203 dominance out of SSA and register them for PHI insertion.
30205 2017-02-13 Martin Sebor <msebor@redhat.com>
30207 PR middle-end/79496
30208 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
30209 clearing info.nowrite flag when snprintf size argument is a range.
30211 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30213 * cprop.c (cprop_jump): Add missing space in string literal.
30214 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
30215 (get_constraint_for_component_ref): Likewise.
30216 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
30217 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30218 * lra-constraints.c (process_alt_operands): Likewise.
30219 * ipa-inline.c (inline_small_functions): Likewise.
30220 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30221 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30222 * trans-mem.c (diagnose_tm_1_op): Likewise.
30223 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30224 (grid_parallel_clauses_gridifiable): Likewise.
30226 * config/nvptx/mkoffload.c (process): Add space in between
30229 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30230 "MOD4_SSE_REGS" and "ALL_REGS".
30232 * spellcheck.c (test_data): Add , in between "foo" and "food".
30234 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30237 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30238 boundary crossing check and subsequent code generation agree.
30240 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30242 * config/aarch64/aarch64.c (has_memory_op): Delete.
30243 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30246 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30248 PR rtl-optimization/79388
30249 PR rtl-optimization/79450
30250 * combine.c (distribute_notes): When removing TEM_INSN for which
30251 corresponding dest has last value recorded, invalidate that last
30254 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30256 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30257 of explicit '@'. Add missing assembly comment marker on branch costs
30260 2017-02-13 Nathan Sidwell <nathan@acm.org>
30262 * gengtype-lex.l (<in_struct>): Add '/'.
30264 2017-02-13 Martin Liska <mliska@suse.cz>
30267 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30269 2017-02-13 Richard Biener <rguenther@suse.de>
30271 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30273 * configure: Re-generate.
30274 * config.in: Likewise.
30275 * graphite-dependences.c: Simplify as if
30276 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30277 * graphite-isl-ast-to-gimple.c: Likewise.
30278 * graphite-optimize-isl.c: Likewise.
30279 * graphite-poly.c: Likewise.
30280 * graphite-sese-to-poly.c: Likewise.
30281 * graphite.h: Likewise.
30282 * toplev.c: Include isl/version.h and use isl_version () for
30283 printing the ISL version.
30284 * doc/install.texi: Update ISL requirement.
30286 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30288 * doc/standards.texi (Standards): Update reference to
30291 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30293 * doc/extend.texi (Named Address Spaces): sourceware.org now
30295 * doc/install.texi (Binaries): Ditto.
30298 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30300 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30301 standard terminology "stringize"/"stringizing" throughout.
30302 * doc/cppinternals.texi: Likewise.
30304 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30306 * doc/extend.texi: Fix some spelling mistakes and typos.
30307 * doc/invoke.texi: Likewise.
30309 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30312 * params.def (inline-min-speedup) Change from 10 to 8.
30314 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30316 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30319 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30322 * ipa-inline-analysis.c (get_minimal_bb): New function.
30323 (record_modified): Use it.
30324 (remap_edge_change_prob): Handle also ancestor functions.
30326 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30328 * doc/contrib.texi (Contributors): Remove broken link into
30329 the Mauve CVS repository.
30331 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30333 PR middle-end/79454
30334 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30335 result computation whenever lhs doesn't have vector mode, not
30336 just when it has BLKmode.
30338 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30340 * doc/makefile.texi (profiledbootstrap): Refer to the
30341 installation instructions only in textual form.
30343 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30346 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30348 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30350 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30351 (Specific): Update mingw-w64 reference.
30353 (Specific): Remove broken link to Renesas RX processor.
30355 2017-02-10 Richard Biener <rguenther@suse.de>
30357 * toplev.c (process_options): Do not mention obsolete graphite
30358 options when printing sorry message about missing graphite support.
30359 Mention -floop-nest-optimize.
30361 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30363 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30364 (vtst_p16): Likewise.
30365 (vtstq_p8): Likewise.
30366 (vtstq_p16): Likewise.
30368 (vtstq_p64): Likewise.
30369 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30370 (vset_lane_p64): New.
30371 (vsetq_lane_p64): New.
30373 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30375 PR tree-optimization/79411
30376 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30377 stmt operands are SSA_NAMEs used in abnormal phis.
30378 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30381 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30384 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30387 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30389 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30391 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30393 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30397 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30398 not arbitrary TREE_CONSTANT.
30401 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30402 "omp declare target link" attribute unless is_global_var.
30403 * omp-offload.c (find_link_var_op): Likewise.
30405 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30406 Chung-Lin Tang <cltang@codesourcery.com>
30408 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30410 (gimplify_adjust_omp_clauses): Don't delete TILE.
30411 (gimplify_omp_for): Deal with TILE.
30412 * internal-fn.c (expand_GOACC_TILE): New function.
30413 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30415 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30416 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30418 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30419 avoid DIV for outermost collapse var.
30420 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30421 Remove out of date comments, fix whitespace.
30422 * omp-general.c (omp_extract_for_data): Deal with tiling.
30423 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30424 adjust OLF_DIM_BASE value.
30425 (struct omp_for_data): Add tiling field.
30426 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30427 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30428 for auto loops. Remove default auto determining, moved to
30429 oacc_loop_fixed_partitions.
30430 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30431 stmts, add e_mask field.
30432 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30433 (oacc_thread_numbers): Use oacc_dim_call.
30434 (oacc_xform_tile): New.
30435 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30436 (finish_oacc_loop): Adjust for ifns vector.
30437 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30438 add case for GOACC_TILE fns.
30439 (oacc_loop_xform_loop): Delete.
30440 (oacc_loop_process): Iterate over call list directly, and add
30441 handling for GOACC_TILE fns.
30442 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30444 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30445 vector partitioning to outer loops. Assign 2 partitions to loops
30446 when available. Add TILE handling.
30447 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30448 (execite_oacc_device_lower): Process GOACC_TILE fns,
30449 ignore unknown specs.
30450 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30451 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30452 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30454 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30456 * configure.ac (ACX_BUGURL): Update.
30457 * configure: Regenerate.
30459 2017-02-09 Richard Biener <rguenther@suse.de>
30461 PR tree-optimization/69823
30462 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30463 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30465 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30467 * config/arc/arc-c.def: Add __NPS400__ definition.
30468 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30469 (TARGET_NPS400): Define.
30471 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30473 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30475 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30476 pointer, arch_info.
30477 (arc_cpu_types): Fill the arch_info field with a pointer into the
30478 arc_arch_types table.
30479 (arc_selected_cpu): Declare.
30480 * config/arc/arc.c (arc_selected_cpu): Make global.
30481 (arc_selected_arch): Delete.
30482 (arc_base_cpu): Delete.
30483 (arc_override_options): Remove references to deleted variables,
30484 update access to arch information.
30485 (ARC_OPT): Update access to arch information.
30486 (ARC_OPTX): Likewise.
30487 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30488 (TARGET_ARC600): Update access to arch information.
30489 (TARGET_ARC601): Likewise.
30490 (TARGET_ARC700): Likewise.
30491 (TARGET_EM): Likewise.
30492 (TARGET_HS): Likewise.
30493 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30496 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30499 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30500 condition/operands for integer GE/LE/GEU/LEU operations.
30502 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30504 PR translation/79397
30505 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30508 2017-02-08 Martin Jambor <mjambor@suse.cz>
30511 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30512 whether allocation happened.
30513 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30514 nothing was allocated.
30516 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30518 PR tree-optimization/79408
30519 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30520 constant, but SSA_NAME with a known integer range, use the minimum
30521 of that range instead of op1 to determine if modulo can be replaced
30522 with its first operand.
30524 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30526 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30528 2017-02-08 Richard Biener <rguenther@suse.de>
30530 PR tree-optimization/71824
30531 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30532 Check all loops contained in the merged region.
30534 2017-02-07 Andrew Pinski <apinski@cavium.com>
30536 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30538 2017-02-07 Andrew Pinski <apinski@cavium.com>
30540 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30541 (thunderxt88): Likewise.
30542 (thunderxt81): Disable LSE and change v8.1 to v8.
30543 (thunderxt83): Likewise.
30545 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30546 Richard Biener <rguenther@suse.de>
30548 PR middle-end/79399
30549 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30550 type from int to size_t.
30551 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30553 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30555 PR rtl-optimization/79386
30556 * cprop.c (bypass_conditional_jumps): Initialize
30557 bypass_last_basic_block already before splitting bbs after
30558 unconditional traps...
30559 (bypass_conditional_jumps): ... rather than here.
30562 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30563 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30564 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30565 fix -masm=intel patterns.
30567 2017-02-07 Richard Biener <rguenther@suse.de>
30569 PR tree-optimization/79256
30570 PR middle-end/79278
30571 * builtins.c (get_object_alignment_2): Use min_align_of_type
30572 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30573 and ADJUST_FIELD_ALIGN.
30575 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30577 * doc/tm.texi: Regenerate.
30578 * stor-layout.c (layout_decl): Adjust.
30579 (update_alignment_for_field): Likewise.
30580 (place_field): Likewise.
30581 (min_align_of_type): Likewise.
30582 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30583 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30584 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30585 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30586 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30587 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30588 * config/i386/i386.c (x86_field_alignment): Likewise.
30589 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30590 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30591 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30592 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30593 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30594 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30598 2017-01-30 Richard Biener <rguenther@suse.de>
30600 PR tree-optimization/79256
30601 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30602 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30605 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30607 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30608 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30609 builtins to SImode and emit a zero-extend, if necessary.
30611 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30613 * docs/invoke.texi (RISC-V Options): Alphabetize.
30615 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30617 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30620 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30622 * config/riscv/riscv.c: New file.
30623 * common/config/riscv/riscv-common.c: Likewise.
30624 * config.gcc: Likewise.
30625 * config/riscv/constraints.md: Likewise.
30626 * config/riscv/elf.h: Likewise.
30627 * config/riscv/generic.md: Likewise.
30628 * config/riscv/linux.h: Likewise.
30629 * config/riscv/multilib-generator: Likewise.
30630 * config/riscv/peephole.md: Likewise.
30631 * config/riscv/pic.md: Likewise.
30632 * config/riscv/predicates.md: Likewise.
30633 * config/riscv/riscv-builtins.c: Likewise.
30634 * config/riscv/riscv-c.c: Likewise.
30635 * config/riscv/riscv-ftypes.def: Likewise.
30636 * config/riscv/riscv-modes.def: Likewise.
30637 * config/riscv/riscv-opts.h: Likewise.
30638 * config/riscv/riscv-protos.h: Likewise.
30639 * config/riscv/riscv.h: Likewise.
30640 * config/riscv/riscv.md: Likewise.
30641 * config/riscv/riscv.opt: Likewise.
30642 * config/riscv/sync.md: Likewise.
30643 * config/riscv/t-elf-multilib: Likewise.
30644 * config/riscv/t-linux: Likewise.
30645 * config/riscv/t-linux-multilib: Likewise.
30646 * config/riscv/t-riscv: Likewise.
30647 * configure.ac: Likewise.
30648 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30649 Waterman as RISC-V maintainers.
30650 * doc/install.texi: Add RISC-V entries.
30651 * doc/invoke.texi: Add RISC-V options section.
30652 * doc/md.texi: Add RISC-V constraints section.
30653 * configure: Regenerated.
30655 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
30658 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
30659 false values to be constant vectors with all 0 or all 1 bits set.
30660 (vcondu<mode><mode>): Likewise.
30661 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
30663 (fpmask_comparison_operator): Update comment.
30664 (vecint_comparison_operator): New predicate.
30665 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
30666 vector conditionals when the true and false values are constant
30667 vectors with all 0 bits or all 1 bits set.
30669 2017-02-06 Martin Sebor <msebor@redhat.com>
30671 PR tree-optimization/79376
30672 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30674 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30676 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30677 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30678 to simplify split condition.
30680 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30682 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30683 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30686 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30688 PR rtl-optimization/68664
30689 * target.def (can_speculate_insn): New hook.
30690 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30691 * doc/tm.texi: Regenerate.
30692 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30693 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30694 (rs6000_sched_can_speculate_insn): New function.
30696 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30698 PR tree-optimization/79284
30699 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30700 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30701 vectorizable_mask_load_store, vectorizable_operation,
30702 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30703 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30704 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30705 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30706 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30707 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30708 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30709 is_gimple_assign (stmt). Replace another such test with
30710 is_gimple_assign (stmt).
30712 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30715 * config/avr/avr.c (rtl-iter.h): Include it.
30716 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30717 (avr_legitimate_combined_insn): ...and implementation.
30719 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30721 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30722 * config/s390/s390.c (s390_const_operand_ok)
30723 (s390_canonicalize_comparison, s390_extract_part)
30724 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30725 (s390_contiguous_bitmask_p, s390_rtx_costs)
30726 (legitimize_pic_address): Likewise.
30727 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30728 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30729 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30730 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30731 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30733 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30735 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30736 REGNO($0) == REGNO($1).
30738 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30740 * config/s390/linux.h(SIZE_TYPE): Add comment.
30742 2017-02-06 Julian Brown <julian@codesourcery.com>
30743 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30744 Virendra Pathak <virendra.pathak@broadcom.com>
30746 * config/aarch64/aarch64-cores.def: Change the scheduler
30748 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30749 * config/aarch64/thunderx2t99.md: New file.
30751 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30753 * doc/standards.texi (Go Language): Update link to language
30756 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30758 * tree-eh.c (lower_resx): Sanitize profile.
30759 (cleanup_empty_eh_move_lp): Likewise.
30761 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30764 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30766 * cfgloopmanip.h (loop_version): Update prototype.
30767 * modulo-sched.c (sms_schedule): Update call of loop_version.
30768 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30769 * tree-parloops.c (gen_parallel_loop): Likewise.
30770 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30771 * tree-ssa-loop-split.c (split_loop): Likewise.
30772 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30773 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30775 2017-02-05 Martin Liska <mliska@suse.cz>
30778 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30780 (print_operand_address): Initialize a struct to zero.
30782 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30784 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30785 garbage collector only in textual form.
30787 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30789 * doc/extend.texi (x86 specific memory model extensions for
30790 transactional memory): Simplify a phrase.
30792 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30795 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30796 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30797 (atomic_storedi_1): Likewise.
30799 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30801 PR tree-optimization/79338
30802 * tree-parloops.c (gather_scalar_reductions): Don't call
30803 vect_analyze_loop_form for loop->inner before destroying loop's
30806 2017-02-03 Martin Sebor <msebor@redhat.com>
30808 PR tree-optimization/79327
30809 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30810 when precision has resulted in leading zeros.
30811 (format_integer): Adjust the likely counter to assume an unknown
30812 argument that may be zero is non-zero.
30814 2017-02-03 Jason Merrill <jason@redhat.com>
30817 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30818 avoid copying non-taken branch.
30820 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30822 PR tree-optimization/79340
30823 * tree-vect-loop.c (vectorizable_reduction): Release
30824 vec_defs elements after safe_splicing them into other vectors.
30827 PR tree-optimization/79327
30828 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30829 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30831 (format_integer): Use wide_int_to_tree instead of build_int_cst
30832 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30833 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30834 of shortest and longest sequence.
30836 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30838 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30839 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30841 2017-02-03 Walter Lee <walt@tilera.com>
30844 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30845 after initial stackframe link reg save.
30846 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30848 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30851 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30852 wu for stxssp alternative.
30854 2017-02-03 Martin Sebor <msebor@redhat.com>
30856 PR tree-optimization/79352
30857 * gimple-fold.c (get_range_strlen): Add argument.
30858 (get_range_strlen): Change return type to bool.
30859 (get_maxval_strlen): Pass in a dummy argument.
30860 * gimple-fold.h (get_range_strlen): Change return type to bool.
30861 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30862 * tree.h (array_at_struct_end_p): Add argument.
30863 * tree.c (array_at_struct_end_p): Handle it.
30865 2017-02-03 Martin Liska <mliska@suse.cz>
30868 * multiple_target.c (create_dispatcher_calls): Redirect edge
30869 from a caller of a dispatcher.
30870 (expand_target_clones): Make the clones local.
30871 (ipa_target_clone): Do both target clones and resolvers.
30872 (ipa_dispatcher_calls): Remove the pass.
30873 (pass_dispatcher_calls::gate): Likewise.
30874 (make_pass_dispatcher_calls): Likewise.
30875 * passes.def (pass_target_clone): Put as very first IPA early
30878 2017-02-03 Martin Liska <mliska@suse.cz>
30880 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30881 in case of a function with ifunc attribute.
30883 2017-02-03 Martin Liska <mliska@suse.cz>
30885 * cgraph.c (cgraph_node::dump): Dump function version info.
30886 * symtab.c (symtab_node::dump_base): Add missing new line.
30888 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30890 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30891 (ifcombine_ifandif): Use it.
30893 2017-02-03 Martin Liska <mliska@suse.cz>
30895 * doc/invoke.texi: Document default value for
30896 use-after-scope-direct-emission-threshold.
30898 2017-02-03 Martin Liska <mliska@suse.cz>
30900 PR tree-optimization/79339
30901 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30902 (format_floating): Likewise.
30904 2017-02-03 Martin Liska <mliska@suse.cz>
30907 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30909 (ipa_node_params_t::remove): Likewise.
30910 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30911 initialization from removed ipa_node_params_t::insert.
30912 (ipa_node_params::~ipa_node_params): Move from removed
30913 ipa_node_params_t::release.
30914 * symbol-summary.h (symbol_summary::m_released): New member.
30915 Do not release a summary twice. Do not allow to call finalizer
30916 for types of a summary that live in GGC memory.
30918 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30920 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30923 2017-02-02 Martin Sebor <msebor@redhat.com>
30925 PR middle-end/79275
30926 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30927 (format_string): Tighten up the range of output for non-constant
30928 strings and correct the expected range for wide non-constant strings.
30930 2017-02-02 Martin Sebor <msebor@redhat.com>
30932 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30934 PR middle-end/32003
30935 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30937 (-fdump-tree-@var): Add to index and document how to come up
30938 with pass-specific option and dump file names.
30939 (-fdump-passes): Clarify where to look for output.
30941 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30943 PR middle-end/77445
30944 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30945 statistics of the analyzed path; allow threading for speed when
30946 any of BBs along the path are optimized for speed.
30948 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30950 PR middle-end/78468
30951 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30952 settings of the virtual registers.
30955 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30957 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30958 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30961 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30963 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30964 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30966 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30968 * config/s390/s390.md: Add missing comments with the expanded
30970 * config/s390/vector.md: Likewise.
30971 * config/s390/vx-builtins.md: Likewise.
30973 2017-02-02 Jakub Jelinek <jakub@redhat.com>
30976 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
30977 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
30978 conditions on a single line.
30980 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30982 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30983 __S390_VX__ to __VX__.
30985 2017-02-01 Andrew Pinski <apinski@cavium.com>
30987 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
30988 stmt_info to record_stmt_cost.
30989 (vect_get_known_peeling_cost): Pass stmt_info if known to
30991 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
30992 cpu_vector_cost field into
30993 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
30994 field into vec_int_stmt_cost and vec_fp_stmt_cost.
30995 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
30996 splitting of scalar_stmt_cost and vec_stmt_cost.
30997 (thunderx_vector_cost): Likewise.
30998 (cortexa57_vector_cost): LIkewise.
30999 (exynosm1_vector_cost): Likewise.
31000 (xgene1_vector_cost): Likewise.
31001 (thunderx2t99_vector_cost): Improve after the splitting of the two
31003 (aarch64_builtin_vectorization_cost): Update for the splitting of
31004 scalar_stmt_cost and vec_stmt_cost.
31006 2017-02-01 Torvald Riegel <triegel@redhat.com>
31007 Richard Henderson <rth@redhat.com>
31009 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31010 conditional on existance of a fast atomic load.
31011 * optabs-query.c (can_atomic_load_p): New function.
31012 * optabs-query.h (can_atomic_load_p): Declare it.
31013 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31014 no fast atomic load is available for the particular size of access.
31015 (expand_atomic_compare_and_swap): Likewise.
31016 (expand_atomic_load): Likewise.
31017 (expand_atomic_store): Likewise.
31018 (expand_atomic_fetch_op): Likewise.
31019 * testsuite/lib/target-supports.exp
31020 (check_effective_target_sync_int_128): Remove x86 because it provides
31021 no fast atomic load.
31022 (check_effective_target_sync_int_128_runtime): Likewise.
31024 2017-02-01 Richard Biener <rguenther@suse.de>
31026 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31027 (graphite_transform_loops): Provide opt-info for optimized nests.
31028 * tree-parloop.c (parallelize_loops): Provide opt-info for
31029 parallelized loops.
31031 2017-02-01 Richard Biener <rguenther@suse.de>
31033 PR middle-end/79315
31034 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31035 was not set before.
31037 2017-02-01 Richard Biener <rguenther@suse.de>
31039 PR tree-optimization/71824
31040 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31041 Verify the loops are valid in the merged SESE region.
31042 (scop_detection::can_represent_loop_1): Check analyzing the
31043 evolution of the number of iterations in the region succeeds.
31045 2017-01-31 Ian Lance Taylor <iant@golang.org>
31047 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31048 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31050 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31052 PR preprocessor/79210
31053 * input.c (get_substring_ranges_for_loc): Replace line_width
31054 assertion with error-handling.
31056 2017-01-31 Richard Biener <rguenther@suse.de>
31058 PR tree-optimization/77318
31059 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31060 (create_pw_aff_from_tree): Take loop parameter.
31061 (add_condition_to_pbb): Pass loop of the condition to
31062 create_pw_aff_from_tree.
31064 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31066 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31067 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31069 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31073 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31075 (convert_int_to_float128): Likewise.
31076 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31077 (convert_int_to_float128): Likewise.
31078 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31079 (UNSPEC_IEEE128_CONVERT): Likewise.
31080 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31081 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31082 Use local variables for IBM extended format.
31083 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31084 (fix_trunc<mode>si2_fprs): Likewise.
31085 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31086 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31087 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31088 to know that we can now have integers of all sizes in vector
31090 (fix<uns>_<mode>di2_hw): Likewise.
31091 (float<uns>_<mode>si2_hw): Likewise.
31092 (fix_<mode>si2_hw): Likewise.
31093 (fixuns_<mode>si2_hw): Likewise.
31094 (float<uns>_<mode>di2_hw): Likewise.
31095 (float_<mode>di2_hw): Likewise.
31096 (float_<mode>si2_hw): Likewise.
31097 (floatuns_<mode>di2_hw): Likewise.
31098 (floatuns_<mode>si2_hw): Likewise.
31099 (xscvqp<su>wz_<mode>): Delete, no longer used.
31100 (xscvqp<su>dz_<mode>): Likewise.
31101 (xscv<su>dqp_<mode>): Likewise.
31102 (ieee128_mfvsrd_64bit): Likewise.
31103 (ieee128_mfvsrd_32bit): Likewise.
31104 (ieee128_mfvsrwz): Likewise.
31105 (ieee128_mtvsrw): Likewise.
31106 (ieee128_mtvsrd_64bit): Likewise.
31107 (ieee128_mtvsrd_32bit): Likewise.
31109 2017-01-31 Martin Liska <mliska@suse.cz>
31112 * ipa-prop.c (ipa_free_all_node_params): Call release method
31113 instead of ~sumbol_summary to not to trigger double times
31116 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31118 PR tree-optimization/71691
31119 * bitmap.h (class auto_bitmap): New.
31120 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31121 is_maybe_undefined instead of ssa_undefined_value_p.
31123 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31125 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31126 __S390_ARCH_LEVEL__ to __ARCH__.
31128 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31130 PR tree-optimization/79267
31131 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31132 if should_remove_lhs_p is true.
31134 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31137 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31138 (add_alignment_attribute): New.
31139 (base_type_die): Add alignment attribute.
31140 (subrange_type_die): Likewise.
31141 (modified_type_die): Likewise.
31142 (gen_array_type_die): Likewise.
31143 (gen_descr_array_type_die: Likewise.
31144 (gen_enumeration_type_die): Likewise.
31145 (gen_subprogram_die): Likewise.
31146 (gen_variable_die): Likewise.
31147 (gen_field_die): Likewise.
31148 (gen_ptr_to_mbr_type_die): Likewise.
31149 (gen_struct_or_union_type_die): Likewise.
31150 (gen_subroutine_type_die): Likewise.
31151 (gen_typedef_die): Likewise.
31152 (base_type_cmp): Compare alignment attribute.
31154 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31157 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
31158 (setb_unsigned) New pattern for setb with CCUNS.
31159 * config/rs6000/rs6000.c (expand_block_compare): Use a different
31160 subfc./subfe sequence to avoid overflow problems. Generate a
31161 shorter sequence with cmpld/setb for power9.
31162 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
31163 for generating subfc. instruction.
31164 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
31165 now uses this instruction.
31167 2017-01-30 Ian Lance Taylor <iant@google.com>
31170 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
31171 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
31173 2017-01-30 Martin Sebor <msebor@redhat.com>
31175 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
31176 Move constant to the right of a relational operator.
31177 (get_mpfr_format_length, format_character, format_string): Ditto.
31178 (should_warn_p, maybe_warn): Same.
31180 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
31182 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
31185 * asan.c (get_translation_unit_decl): Remove function.
31186 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
31188 2017-01-30 Martin Liska <mliska@suse.cz>
31190 PR gcov-profile/79259
31191 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
31192 -fprofile-generate.
31194 2017-01-30 Martin Liska <mliska@suse.cz>
31197 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
31198 Initialize variables with NULL value.
31200 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
31203 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
31205 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
31207 2017-01-30 Richard Biener <rguenther@suse.de>
31209 PR tree-optimization/79276
31210 * tree-vrp.c (process_assert_insertions): Properly adjust common
31211 when removing a duplicate.
31213 2017-01-30 Richard Biener <rguenther@suse.de>
31215 PR tree-optimization/79256
31216 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31217 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31219 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31221 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31224 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31225 ("*r<noxa>sbg_<mode>_sll_bitmask")
31226 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31227 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31228 Use contiguous_bitmask_nowrap_operand.
31230 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31233 * config/rs6000/altivec.h (vec_xl): Revise #define.
31234 (vec_xst): Likewise.
31236 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31238 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31240 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31242 PR rtl-optimization/79194
31243 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31244 traps before call to bypass_conditional_jumps.
31246 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31248 PR tree-optimization/71374
31249 * lra-constraints.c (check_conflict_input_operands): New.
31250 (match_reload): Use it.
31252 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31255 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31256 account to calculate conflict_set.
31258 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31260 PR rtl-optimization/78559
31261 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31262 other_insn in combine.
31264 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31266 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31267 uint16_type_node for BT_UINT16.
31269 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31271 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31272 "RTL Tests" to menu.
31273 (GIMPLE Tests): New node.
31274 (RTL Tests): New node.
31276 2017-01-27 Richard Biener <rguenther@suse.de>
31278 PR tree-optimization/79245
31279 * tree-loop-distribution.c (distribute_loop): Apply cost
31280 modeling also to detected patterns.
31282 2017-01-27 Richard Biener <rguenther@suse.de>
31284 PR tree-optimization/71433
31285 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31286 (compare_assert_loc): New function.
31287 (process_assert_insertions): Sort and optimize assert locations
31288 to remove duplicates and push down identical assertions on
31289 edges to their destination block.
31291 2017-01-27 Richard Biener <rguenther@suse.de>
31293 PR tree-optimization/79244
31294 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31295 out SSA names even if abnormal.
31297 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31299 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31300 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31301 instead of MPFR_RNDN.
31303 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31306 * arm.c (arm_option_override): Don't call build_target_option_node
31307 until after doing all option overrides.
31308 (arm_valid_target_attribute_tree): Likewise.
31310 2017-01-27 Martin Liska <mliska@suse.cz>
31312 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31313 for {cd}tors and C++ {cd}tors.
31315 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31317 * config/s390/s390.md ("*setmem_long_and")
31318 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31320 2017-01-26 Martin Sebor <msebor@redhat.com>
31322 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31325 2017-01-26 Martin Sebor <msebor@redhat.com>
31327 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31328 HAVE_DFmode before using XFmode or DFmode.
31329 (parse_directive): Avoid using the z length modifier to avoid
31330 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31332 PR middle-end/78703
31333 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31334 to accept adjustment as an array.
31335 (get_int_range): New function.
31336 (struct directive): Make width and prec arrays.
31337 (directive::set_width, directive::set_precision): Call get_int_range.
31338 (format_integer, format_floating): Handle width and precision ranges.
31339 (format_string, parse_directive): Same.
31341 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31344 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31345 comdat_type_p set, just clone them, but keep the children in the
31349 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31350 which have direct callers with -fvar-tracking-assignments enabled
31352 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31353 inside of type units.
31355 2017-01-26 Martin Sebor <msebor@redhat.com>
31357 PR middle-end/78703
31358 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31360 (struct format_result): Replace number_chars, number_chars_min,
31361 and number_chars_max with a single member of struct result_range.
31363 (format_result::operator+=): Adjust.
31364 (struct fmtresult): Remove bounded. Handle likely and unlikely
31366 (fmtresult::adjust_for_width_or_precision): New function.
31367 (fmtresult:type_max_digits): New function.
31368 (bytes_remaining): Handle likely and unlikely counters.
31369 (min_bytes_remaining): Remove.
31370 (format_percent): Simplify.
31371 (format_integer, format_floating): Set likely and unlikely counters.
31372 (get_string_length, format_character, format_string): Same.
31373 (format_plain, should_warn_p): New function.
31374 (maybe_warn): Call should_warn_p. Update diagnostic messages
31375 and handle those for all directives, including plain strings.
31376 (format_directive): Handle likely and unlikely counters.
31377 Remove unnecessary quoting from diagnostics. Add an informational
31379 (add_bytes): Remove.
31380 (pass_sprintf_length::compute_format_length): Simplify.
31381 (try_substitute_return_value): Handle likely and unlikely counters.
31383 2017-01-26 Carl Love <cel@us.ibm.com>
31385 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31386 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31388 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31391 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31392 endianess for subregs into account.
31393 * lra-constraints.c (lra_constraints): Do risky transformations
31394 always on the first iteration.
31395 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31396 last_call_used_reg_set.
31397 (process_bb_lives): Define and use last_call_used_reg_set.
31398 * lra.c (lra): Always continue after lra_constraints on the first
31401 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31403 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31405 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31407 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31409 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31410 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31411 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31412 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31413 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31414 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31415 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31416 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31417 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31419 2017-01-26 Marek Polacek <polacek@redhat.com>
31422 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31423 for the third operand.
31425 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31427 PR middle-end/79236
31428 * omp-low.c (struct omp_context): Add simt_stmt field.
31429 (scan_omp_for): Return omp_context *.
31430 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31431 context to the _simt_ SIMD stmt.
31432 (lower_omp_for): For combined SIMD with sibling _simt_
31433 SIMD, make sure to use the same decls in _looptemp_
31434 clauses as in the sibling.
31436 2017-01-26 David Sherwood <david.sherwood@arm.com>
31438 PR middle-end/79212
31439 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31442 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31445 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31446 emit fld b; fld a; if possible.
31448 * brig-builtins.def: Update copyright years.
31449 * config/arm/arm_acle_builtins.def: Update copyright years.
31451 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31454 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31455 constraint instead of o for the stxsd instruction.
31457 2017-01-25 Carl Love <cel@us.ibm.com>
31459 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31460 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31462 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31464 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31466 2017-01-25 Richard Biener <rguenther@suse.de>
31468 PR tree-optimization/69264
31469 * target.def (vector_alignment_reachable): Improve documentation.
31470 * doc/tm.texi: Regenerate.
31471 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31473 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31474 earlier changes with respect to TYPE_USER_ALIGN.
31475 (vector_alignment_reachable_p): Likewise. Improve dumping.
31477 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31480 * config/arm/arm.md (xordi3): Force constant operand into a register
31483 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31485 * doc/invoke.texi (-fstore-merging): Correct default optimization
31486 levels at which it is enabled.
31487 (-O): Move -fstore-merging from list to...
31490 2017-01-25 Richard Biener <rguenther@suse.de>
31493 * omp-expand.c: Include debug.h.
31494 (expand_omp_taskreg): Make sure to generate early debug before
31495 outlining anything from a function.
31496 (expand_omp_target): Likewise.
31497 (grid_expand_target_grid_body): Likewise.
31499 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31502 * asan.c (get_translation_unit_decl): New function.
31503 (asan_add_global): Extract modules file name from globals
31504 TRANSLATION_UNIT_DECL name.
31506 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31509 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31510 for long calls with APCS frame and VFP.
31512 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31514 * cfg.c (original_copy_tables_initialized_p): New function.
31515 * cfg.h (original_copy_tables_initialized_p): New decl.
31516 * cfgrtl.c (relink_block_chain): Guard the call to
31517 free_original_copy_tables with a call to
31518 original_copy_tables_initialized_p.
31519 * cgraph.h (symtab_node::native_rtl_p): New decl.
31520 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31521 (symtab_node::needed_p): Don't assert for early assembly output
31522 for __RTL functions.
31523 (cgraph_node::finalize_function): Set "force_output" for __RTL
31525 (cgraph_node::analyze): Bail out early for __RTL functions.
31526 (analyze_functions): Update assertion to support __RTL functions.
31527 (cgraph_node::expand): Bail out early for __RTL functions.
31528 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31530 * function.h (struct function): Update comment for field
31532 * gimple-expr.c: Include "tree-pass.h".
31533 (gimple_has_body_p): Return false for __RTL functions.
31534 * Makefile.in (OBJS): Add run-rtl-passes.o.
31535 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31537 (gcc::pass_manager::get_clean_slate): New accessor.
31538 * passes.c: Include "insn-addr.h".
31539 (should_skip_pass_p): Add logging. Update logic for running
31540 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31541 property-provider override so it is only done for gimple passes.
31543 (skip_pass): New function.
31544 (execute_one_pass): Call skip_pass when skipping passes.
31545 * read-md.c (md_reader::read_char): Support filtering
31546 the input to a subset of line numbers.
31547 (md_reader::md_reader): Initialize fields
31548 m_first_line and m_last_line.
31549 (md_reader::read_file_fragment): New function.
31550 * read-md.h (md_reader::read_file_fragment): New decl.
31551 (md_reader::m_first_line): New field.
31552 (md_reader::m_last_line): New field.
31553 * read-rtl-function.c (function_reader::create_function): Only
31554 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31555 curr_properties. Set DECL_INITIAL to a dummy block.
31556 (read_rtl_function_body_from_file_range): New function.
31557 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31559 * run-rtl-passes.c: New file.
31560 * run-rtl-passes.h: New file.
31562 2017-01-24 Jeff Law <law@redhat.com>
31564 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31567 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31569 PR tree-optimization/79159
31570 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31571 (record_nonwrapping_iv): Improve boundary using above function if no
31572 value range information.
31574 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31575 Martin Jambor <mjambor@suse.cz>
31577 * brig-builtins.def: New file.
31578 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31579 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31580 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31581 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31582 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31583 * builtin-types.def (BT_INT8): New.
31584 (BT_INT16): Likewise.
31585 (BT_UINT8): Likewise.
31586 (BT_UINT16): Likewise.
31587 (BT_FN_ULONG): Likewise.
31588 (BT_FN_UINT_INT): Likewise.
31589 (BT_FN_UINT_ULONG): Likewise.
31590 (BT_FN_UINT_LONG): Likewise.
31591 (BT_FN_UINT_PTR): Likewise.
31592 (BT_FN_ULONG_PTR): Likewise.
31593 (BT_FN_INT8_FLOAT): Likewise.
31594 (BT_FN_INT16_FLOAT): Likewise.
31595 (BT_FN_UINT32_FLOAT): Likewise.
31596 (BT_FN_UINT16_FLOAT): Likewise.
31597 (BT_FN_UINT8_FLOAT): Likewise.
31598 (BT_FN_UINT64_FLOAT): Likewise.
31599 (BT_FN_UINT16_UINT32): Likewise.
31600 (BT_FN_UINT32_UINT16): Likewise.
31601 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31602 (BT_FN_INT_PTR_INT): Likewise.
31603 (BT_FN_UINT_PTR_UINT): Likewise.
31604 (BT_FN_LONG_PTR_LONG): Likewise.
31605 (BT_FN_ULONG_PTR_ULONG): Likewise.
31606 (BT_FN_VOID_UINT64_UINT64): Likewise.
31607 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31608 (BT_FN_INT8_INT8_INT8): Likewise.
31609 (BT_FN_INT16_INT16_INT16): Likewise.
31610 (BT_FN_INT_INT_INT): Likewise.
31611 (BT_FN_UINT_FLOAT_UINT): Likewise.
31612 (BT_FN_FLOAT_UINT_UINT): Likewise.
31613 (BT_FN_ULONG_UINT_UINT): Likewise.
31614 (BT_FN_ULONG_UINT_PTR): Likewise.
31615 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31616 (BT_FN_UINT_UINT_UINT): Likewise.
31617 (BT_FN_VOID_UINT_PTR): Likewise.
31618 (BT_FN_UINT_UINT_PTR: Likewise.
31619 (BT_FN_UINT32_UINT64_PTR): Likewise.
31620 (BT_FN_INT_INT_UINT_UINT): Likewise.
31621 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31622 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31623 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31624 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31625 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31626 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31627 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31628 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31629 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31630 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31631 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31632 * doc/frontends.texi: List BRIG FE.
31633 * doc/install.texi (Testing): Add BRIG tesring requirements.
31634 * doc/invoke.texi (Overall Options): Mention BRIG.
31635 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31637 2017-01-24 Richard Biener <rguenther@suse.de>
31639 PR translation/79208
31640 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31642 2017-01-24 Martin Jambor <mjambor@suse.cz>
31645 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31646 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31647 and known_contexts.
31649 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31651 PR middle-end/79123
31652 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31653 casts from signed to unsigned really don't have a range.
31655 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
31657 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
31658 GMP_RNDx for compatiblity.
31660 2017-01-24 Martin Liska <mliska@suse.cz>
31663 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
31664 that would prevent us to call alloca with -1 as argument.
31666 2017-01-24 Jakub Jelinek <jakub@redhat.com>
31668 * dwarf2out.c (output_compilation_unit_header, output_file_names):
31669 Avoid -Wformat-security warning.
31671 2017-01-23 Andrew Pinski <apinski@cavium.com>
31673 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31676 2017-01-23 Martin Sebor <msebor@redhat.com>
31678 PR middle-end/78703
31679 * gimple-ssa-sprintf.c (warn_level): New global.
31680 (format_integer): Use it here and throughout the rest of the file.
31681 Use the same switch to compute sign as base.
31682 (maybe_warn): New function.
31683 (format_directive): Factor out warnings into maybe_warn.
31684 Add debugging output. Use warn_level.
31685 (add_bytes): Use warn_level.
31686 (pass_sprintf_length::compute_format_length): Add debugging output.
31687 (try_substitute_return_value): Same.
31688 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31690 PR middle-end/78703
31691 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31692 (struct fmtresult, format_integer, format_floating): Adjust.
31693 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31694 (get_string_length, format_string,format_directive): Same.
31695 (pass_sprintf_length::compute_format_length): Same.
31696 (try_substitute_return_value): Simplify slightly.
31698 PR middle-end/78703
31699 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31700 (fmtresult::operator+=): Outlined.
31701 (struct fmtresult): Add ctors.
31702 (struct conversion_spec): Rename...
31703 (struct directive): ...to this. Add and remove data members.
31704 (directive::set_width, directive::set_precision): New functions.
31705 (format_percent): Use fmtresult ctor.
31706 (get_width_and_precision): Remove.
31707 (format_integer): Make naming changes. Avoid computing width and
31709 (format_floating): Same. Adjust indentation.
31710 (format_character, format_none): New functions.
31711 (format_string): Moved character handling to format_character.
31712 (format_directive): Remove arguments, change return type.
31713 (parse_directive): New function.
31714 (pass_sprintf_length::compute_format_length): Move directive
31715 parsing to parse_directive.
31717 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31719 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31720 (assign_assembler_name_if_needed): ... this.
31721 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31722 (assign_assembler_name_if_needed): ... this.
31723 (free_lang_data_in_cgraph): Adjust callers.
31724 * cgraphunit.c (cgraph_node::analyze): Likewise.
31725 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31728 2017-01-23 Richard Biener <rguenther@suse.de>
31730 PR tree-optimization/79088
31731 PR tree-optimization/79188
31732 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31733 resetting loop bounds after last path deletion. Reset loop
31734 bounds of the target loop, make code match the comments.
31735 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31736 Make sure loops need no fixups.
31738 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31740 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31741 exponent support with double type for first argument.
31742 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31743 type returned by __builtin_vec_extract_sig,
31744 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31745 functions from "vector int" to "vector unsigned int" or from
31746 "vector long long int" to "vector unsigned long long int".
31747 Changed type returned by __builtin_vec_extract_exp,
31748 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31749 functions from "vector int" to "vector unsigned int" or from
31750 "vector long long int" to "vector unsigned long long int".
31751 Changed return type of __builtin_vec_test_data_class,
31752 __builtin_vec_test_data_class_sp, and
31753 __builtin_vec_test_data_class_dp from "vector int" to
31754 "vector bool int" or from "vector long long int" to "vector bool
31755 long long int" and changed second argument type from "unsigned
31756 int" to "int". Added new overloaded function forms "vector float
31757 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31758 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31759 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31760 double, vector unsigned long long int)" and "vector double
31761 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31762 long int)". Changed return type of
31763 __builtin_scalar_test_data_class and
31764 __builtin_scalar_test_data_class_sp and
31765 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31766 int" and changed second argument from "unsigned int" to "int".
31767 Changed type returned by __builtin_scalar_test_neg,
31768 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31769 from "int" to "bool int". Added new overloaded function form
31770 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31771 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31772 exponent double-precision with floating point first argument.
31773 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31774 documentation of scalar_test_data_class, scalar_test_neg,
31775 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31776 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31777 vec_test_data_class built-in functions to reflect refinements in
31778 their type signatures.
31780 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31782 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31784 (aarch64_elf_asm_destructor): Likewise.
31786 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31788 PR rtl-optimization/78634
31789 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31790 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31791 * ifcvt.c (noce_try_cmove): Add missing cost check.
31793 PR rtl-optimization/71724
31794 * combine.c (if_then_else_cond): Look for situations where it is
31795 beneficial to undo the work of one of the recursive calls.
31797 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31799 PR tree-optimization/70754
31800 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31801 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31802 combined stmt before it if not NULL.
31803 (combine_chains): Process refs reversely and compute dominance point
31806 2017-01-23 Martin Liska <mliska@suse.cz>
31808 PR tree-optimization/79196
31809 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31810 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31812 (strlen_optimize_stmt): Call the renamed function.
31814 2017-01-23 Michael Matz <matz@suse.de>
31816 PR tree-optimization/78384
31817 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31819 2017-01-23 Richard Biener <rguenther@suse.de>
31821 PR tree-optimization/79186
31822 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31823 both incoming edges before moving an assert.
31825 2017-01-23 Martin Jambor <mjambor@suse.cz>
31827 * ipa-prop.c (load_from_param_1): Removed.
31828 (load_from_unmodified_param): Bits from load_from_param_1 put back
31830 (load_from_param): Removed.
31831 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31832 with stmt. Reverted back to use of load_from_unmodified_param.
31834 2017-01-23 Martin Jambor <mjambor@suse.cz>
31837 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31838 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31839 field a pointer to garbage collected vector, mark lattices and
31840 ipcp_orig_node with GTY((skip)).
31841 (ipa_get_param_count): Adjust to descriptors being a pointer.
31842 (ipa_get_param): Likewise.
31843 (ipa_get_type): Likewise.
31844 (ipa_get_param_move_cost): Likewise.
31845 (ipa_set_param_used): Likewise.
31846 (ipa_get_controlled_uses): Likewise.
31847 (ipa_set_controlled_uses): Likewise.
31848 (ipa_is_param_used): Likewise.
31849 (ipa_node_params_t): Move into garbage collector. New methods insert
31851 (ipa_node_params_sum): Annotate wth GTY(()).
31852 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31854 (ipa_load_from_parm_agg): Adjust declaration.
31855 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31856 * ipa-profile.c (ipa_profile): Likewise.
31857 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31858 (ipa_populate_param_decls): Make descriptors parameter garbage
31860 (ipa_dump_param): Adjust to descriptors being a pointer.
31861 (ipa_alloc_node_params): Likewise.
31862 (ipa_initialize_node_params): Likewise.
31863 (load_from_param_1): Make descriptors parameter garbage collected.
31864 (load_from_unmodified_param): Likewise.
31865 (load_from_param): Likewise.
31866 (ipa_load_from_parm_agg): Likewise.
31867 (ipa_node_params::~ipa_node_params): Removed.
31868 (ipa_free_all_node_params): Remove call to delete operator.
31869 (ipa_node_params_t::insert): New.
31870 (ipa_node_params_t::remove): Likewise.
31871 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31872 copy known_csts and known_contexts vectors.
31873 (ipa_read_node_info): Adjust to descriptors being a pointer.
31874 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31876 (ipcp_transform_function): Make descriptors variable garbage
31879 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31881 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31882 * config/i386/avx512dqintrin.h: Ditto.
31883 * config/i386/avx512fintrin.h: Ditto.
31884 * config/i386/i386.c: Handle new builtins.
31885 * config/i386/i386-builtin.def: Add new builtins.
31886 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31887 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31889 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31890 Martin Liska <mliska@suse.cz>
31892 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31893 * asan.c (asan_expand_poison_ifn): Support stores and use
31894 appropriate ASAN report function.
31895 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31896 * internal-fn.def (ASAN_POISON_USE): Declare.
31897 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31898 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31899 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31900 ASAN_POISON calls w/o LHS.
31901 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31902 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31903 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31904 * gimplify.c (asan_poison_variables): Add attribute
31905 use_after_scope_memory to variables that really needs to live
31907 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31908 having the attribute.
31910 2017-01-23 Martin Liska <mliska@suse.cz>
31912 * asan.c (create_asan_shadow_var): New function.
31913 (asan_expand_poison_ifn): Likewise.
31914 * asan.h (asan_expand_poison_ifn): New declaration.
31915 * internal-fn.c (expand_ASAN_POISON): Likewise.
31916 * internal-fn.def (ASAN_POISON): New builtin.
31917 * sanopt.c (pass_sanopt::execute): Expand
31918 asan_expand_poison_ifn.
31919 * tree-inline.c (copy_decl_for_dup_finish): Make function
31921 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31922 * tree-ssa.c (is_asan_mark_p): New function.
31923 (execute_update_addresses_taken): Rewrite local variables
31924 (identified just by use-after-scope as addressable) into SSA.
31926 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31928 * doc/install.texi (Specific): opensource.apple.com uses https
31929 now. Remove trailing slash.
31931 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31933 * README.Portability: Remove note on an Irix compatibility issue.
31935 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31937 * gcov.c (INCLUDE_ALGORITHM): Define.
31938 (INCLUDE_VECTOR): Define.
31939 No longer include <vector> and <algorithm> directly.
31941 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31943 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31945 * doc/invoke.texi (Code Gen Options): Ditto.
31947 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31950 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31952 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31954 rtl-optimization/79125
31955 * cprop.c (local_cprop_pass): Handle cases where we make an
31956 unconditional trap.
31958 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31962 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31963 read from, for big endian.
31965 2017-01-20 Jiong Wang <jiong.wang@arm.com>
31967 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
31968 register pauth builtins for LP64 only.
31970 2017-01-20 Marek Polacek <polacek@redhat.com>
31973 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
31976 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31978 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
31980 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
31981 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
31982 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
31984 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31987 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
31988 in big-endian mode when they are not a single duplicated value.
31990 2017-01-20 Richard Biener <rguenther@suse.de>
31992 * BASE-VER: Bump to 7.0.1.
31994 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31996 * omp-low.c (omplow_simd_context): New struct. Use it...
31997 (lower_rec_simd_input_clauses): ...here and...
31998 (lower_rec_input_clauses): ...here to hold common data. Adjust all
31999 references to idx, lane, max_vf, is_simt.
32001 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32003 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32006 2017-01-20 Martin Jambor <mjambor@suse.cz>
32008 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32009 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32011 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32012 (GTFILES): Rename hsa.c to hsa-common.c.
32013 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32014 * hsa-dump.c: Likewise.
32015 * hsa-gen.c: Likewise.
32016 * hsa-regalloc.c: Likewise.
32017 * ipa-hsa.c: Likewise.
32018 * omp-expand.c: Likewise.
32019 * omp-low.c: Likewise.
32020 * toplev.c: Likewise.
32022 2017-01-20 Marek Polacek <polacek@redhat.com>
32025 * doc/invoke.texi: Document -Wduplicated-branches.
32026 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32027 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32028 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32029 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32030 return 0 only when not OEP_LEXICOGRAPHIC.
32031 (fold_build_cleanup_point_expr): Use the expression
32032 location when building CLEANUP_POINT_EXPR.
32033 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32034 * tree.c (add_expr): Handle error_mark_node.
32036 2017-01-20 Martin Liska <mliska@suse.cz>
32039 * tree-profile.c (init_ic_make_global_vars): Do not call
32041 (gimple_init_gcov_profiler): Likewise.
32043 2017-01-20 Martin Liska <mliska@suse.cz>
32046 * cgraph.h (maybe_create_reference): Remove argument and
32048 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32050 * ipa-cp.c (create_specialized_node): Likewise.
32051 * symtab.c (symtab_node::maybe_create_reference): Handle
32052 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32054 2017-01-20 Martin Liska <mliska@suse.cz>
32056 * read-rtl-function.c (function_reader::create_function): Use
32057 build_decl instread of build_decl_stat.
32059 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32061 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32062 * config/i386/avx512dqintrin.h: Ditto.
32063 * config/i386/avx512fintrin.h: Ditto.
32064 * config/i386/i386-builtin-types.def: Add new types.
32065 * config/i386/i386.c: Handle new types.
32066 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32067 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32068 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32069 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32070 (__builtin_ia32_kshiftridi): New.
32071 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32073 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32077 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32078 define to rs6000_init_stack_protect_guard.
32079 (rs6000_init_stack_protect_guard): New function.
32081 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32082 Yunqiang Su <yunqiang.su@imgtec.com>
32084 * config.gcc (supported_defaults): Add madd4.
32085 (with_madd4): Add validation.
32086 (all_defaults): Add madd4.
32087 * config/mips/mips.opt (mmadd4): New option.
32088 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32090 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32092 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32093 (ISA_HAS_FUSED_MADD4): Likewise.
32094 * doc/invoke.texi (-mmadd4): Document the new option.
32095 * doc/install.texi (--with-madd4): Document the new option.
32097 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32099 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32100 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32101 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32102 (aarch64_init_pauth_hint_builtins): New.
32103 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32104 (aarch64_expand_builtin): Expand new builtins.
32106 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32108 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32109 * combine-stack-adj.c (no_unhandled_cfa): Handle
32110 REG_CFA_TOGGLE_RA_MANGLE.
32111 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32112 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32113 info for return address signing.
32114 (aarch64_expand_epilogue): Likewise.
32116 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32118 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32119 * config/aarch64/aarch64-protos.h
32120 (aarch64_return_address_signing_enabled): New declaration.
32121 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32123 (aarch64_expand_prologue): Sign return address before it's pushed onto
32125 (aarch64_expand_epilogue): Authenticate return address fetched from
32127 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32128 (aarch64_attributes): New function attributes for "sign-return-address".
32129 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32130 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32131 ("*do_return"): Generate combined instructions according to key index.
32132 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32133 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32135 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32136 * config/aarch64/aarch64.opt (msign-return-address=): New.
32137 * doc/extend.texi (AArch64 Function Attributes): Documents
32138 "sign-return-address=".
32139 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32141 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32143 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32144 overall option summary.
32146 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32148 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32149 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32150 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32151 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32153 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
32155 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
32156 -mpower9-minmax by default for -mcpu=power9.
32157 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
32158 128-bit floating point.
32160 2017-01-20 Alan Modra <amodra@gmail.com>
32162 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
32163 optimizing for size.
32165 2017-01-20 Alan Modra <amodra@gmail.com>
32168 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
32169 for strcmp and strncmp from corresponding builtin decl.
32171 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32173 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
32174 instead of i386/rtems-64.h.
32175 * config/i386/rtems-64.h: Remove.
32177 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32181 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
32183 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
32185 2017-01-19 Tamar Christina <tamar.christina@arm.com>
32187 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
32188 Change int to HOST_WIDE_INT.
32189 * config/aarch64/aarch64-protos.h
32190 (aarch64_simd_gen_const_vector_dup): Likewise.
32191 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
32193 2017-01-19 David Malcolm <dmalcolm@redhat.com>
32195 * langhooks-def.h (lhd_type_for_size): New decl.
32196 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
32197 * langhooks.c (lhd_type_for_size): New function, taken from
32200 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
32202 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
32203 define_bypass for CR latency.
32204 (power9-cracked-alu): Update bypass latency and remove power9-branch.
32205 (power9-alu2): Add define_bypass for CR latency.
32207 (power9-mul): Update insn latency.
32208 (power9-mul-compare): Update insn latency, bypass latency and remove
32211 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32213 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
32215 * config/aarch64/aarch64.md
32216 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
32217 aarch64_nopcrelative_literal_loads.
32218 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32220 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32222 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32223 TARGET_LOONGSON_3A.
32224 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32226 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32229 * config.gcc (supported_defaults): Add lxc1-sxc1.
32230 (with_lxc1_sxc1): Add validation.
32231 (all_defaults): Add lxc1-sxc1.
32232 * config/mips/mips.opt (mlxc1-sxc1): New option.
32233 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32235 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32236 __mips_no_lxc1_sxc1.
32237 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32238 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32239 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32241 2017-01-19 Richard Biener <rguenther@suse.de>
32243 PR tree-optimization/72488
32244 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32245 sure to restore SSA info.
32246 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32248 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32250 PR rtl-optimization/79121
32251 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32252 of the inner type when shifting an extended value.
32254 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32257 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32258 interposable aliases.
32260 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32263 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32264 Use the evmergelohi instruction.
32265 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32266 (mov_sitf_e500_subreg8_2_be): Likewise.
32267 (mov_sitf_e500_subreg12_2_le): Likewise.
32268 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32269 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32270 (mov_sitf_e500_subreg8_2_le): Likewise.
32271 (mov_sitf_e500_subreg12_2_be): Likewise.
32273 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32275 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32276 attribute from vecsimple to vecperm.
32277 (altivec_vbpermq2): Likewise.
32279 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32282 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32284 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32285 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32286 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32287 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32288 case where N arg is SIZE_MAX.
32289 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32290 (cmpstrsi): Add pattern.
32292 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32294 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32295 __builtin_vec_revb builtins.
32296 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32297 built-in functions to support generation of the ISA 3.0 XXBR<x>
32298 vector byte reverse instructions.
32299 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32300 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32301 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32302 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32303 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32304 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32305 (P9V_BUILTIN_VEC_REVB): Likewise.
32306 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32307 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32308 (p9_xxbrq_v16qi): Likewise.
32309 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32310 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32311 (p9_xxbrh_v8hi): Likewise.
32312 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32313 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32314 vec_revb built-in functions.
32316 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32318 PR rtl-optimization/78952
32319 * config/i386/i386.md (any_extract): New code iterator.
32320 (*insvqi_2): Use any_extract for source operand.
32321 (*insvqi_3): Use any_shiftrt for source operand.
32323 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32325 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32327 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32329 2017-01-18 Matthias Klose <doko@ubuntu.com>
32331 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32333 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32335 * config/rs6000/altivec.h (vec_bperm): Change #define.
32336 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32337 (altivec_vbpermq2): New define_insn.
32338 (altivec_vbpermd): Likewise.
32339 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32340 function interface.
32341 (VBPERMD): Likewise.
32342 (VBPERM): New polymorphic function interface.
32343 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32344 Add entries for P9V_BUILTIN_VEC_VBPERM.
32345 * doc/extend.texi: Add interfaces for vec_bperm.
32347 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32349 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32350 first letter of error messages.
32351 (s390_resolve_overloaded_builtin): Likewise.
32352 * config/s390/s390.c (s390_expand_builtin): Likewise.
32353 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32354 (s390_valid_target_attribute_inner_p): Likewise.
32355 * config/s390/s390.md ("tabort"): Likewise.
32357 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32359 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32360 (ISA_AVOID_DIV_HILO): New macro.
32361 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32362 (ISA_HAS_DDIV): Likewise.
32364 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32366 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32368 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32370 * doc/invoke.texi (fabi-version): Spelling fix.
32372 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32375 * doc/invoke.texi (fabi-version): Mention mangling fix for
32378 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32381 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32383 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32386 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32387 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32389 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32391 (stack_protect_test): Ditto.
32392 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32393 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32395 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32396 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32397 -mstack-protector-guard-offset=.
32398 (RS/6000 and PowerPC Options): Ditto.
32400 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32402 * config/i386/i386.h (MASK_CLASS_P): New define.
32403 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32404 there are no registers from different register sets also when
32405 mask registers are used. Update function comment.
32406 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32407 to (*k/*r) and (*k/*km) alternatives.
32409 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32411 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32412 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32413 (EH_RETURN_HANDLER_RTX): New define.
32414 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32415 Force frame pointer in EH return functions.
32416 (aarch64_expand_epilogue): Add barrier for eh_return.
32417 (aarch64_final_eh_return_addr): Remove.
32418 (aarch64_eh_return_handler_rtx): New function.
32419 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32421 (aarch64_eh_return_handler_rtx): New prototype.
32423 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32425 * config/rs6000/altivec.h (vec_rlmi): New #define.
32426 (vec_vrlnm): Likewise.
32427 (vec_rlnm): Likewise.
32428 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32429 (UNSPEC_VRLNM): Likewise.
32430 (VIlong): New mode iterator.
32431 (altivec_vrl<VI_char>mi): New define_insn.
32432 (altivec_vrl<VI_char>nm): Likewise.
32433 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32435 (VRLDNM): Likewise.
32436 (RLNM): New polymorphic function entry.
32437 (VRLWMI): New monomorphic function entry.
32438 (VRLDMI): Likewise.
32439 (RLMI): New polymorphic function entry.
32440 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32441 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32442 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32445 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32448 * dwarf2out.c (field_byte_offset): Restore the
32449 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32450 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32451 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32454 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32457 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32459 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32462 * dwarf2out.c (add_data_member_location_attribute): For constant
32463 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32464 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32465 DW_AT_byte_size attributes.
32467 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32469 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32470 after forcing to constant memory when the code model is medium.
32472 2017-01-17 Julia Koval <julia.koval@intel.com>
32475 * config/i386/avx512fintrin.h
32476 (_mm512_i32gather_ps): Change __addr type to void const*.
32477 (_mm512_mask_i32gather_ps): Ditto.
32478 (_mm512_i32gather_pd): Ditto.
32479 (_mm512_mask_i32gather_pd): Ditto.
32480 (_mm512_i64gather_ps): Ditto.
32481 (_mm512_mask_i64gather_ps): Ditto.
32482 (_mm512_i64gather_pd): Ditto.
32483 (_mm512_mask_i64gather_pd): Ditto.
32484 (_mm512_i32gather_epi32): Ditto.
32485 (_mm512_mask_i32gather_epi32): Ditto.
32486 (_mm512_i32gather_epi64): Ditto.
32487 (_mm512_mask_i32gather_epi64): Ditto.
32488 (_mm512_i64gather_epi32): Ditto.
32489 (_mm512_mask_i64gather_epi32): Ditto.
32490 (_mm512_i64gather_epi64): Ditto.
32491 (_mm512_mask_i64gather_epi64): Ditto.
32492 (_mm512_i32scatter_ps): Change __addr type to void*.
32493 (_mm512_mask_i32scatter_ps): Ditto.
32494 (_mm512_i32scatter_pd): Ditto.
32495 (_mm512_mask_i32scatter_pd): Ditto.
32496 (_mm512_i64scatter_ps): Ditto.
32497 (_mm512_mask_i64scatter_ps): Ditto.
32498 (_mm512_i64scatter_pd): Ditto.
32499 (_mm512_mask_i64scatter_pd): Ditto.
32500 (_mm512_i32scatter_epi32): Ditto.
32501 (_mm512_mask_i32scatter_epi32): Ditto.
32502 (_mm512_i32scatter_epi64): Ditto.
32503 (_mm512_mask_i32scatter_epi64): Ditto.
32504 (_mm512_i64scatter_epi32): Ditto.
32505 (_mm512_mask_i64scatter_epi32): Ditto.
32506 (_mm512_i64scatter_epi64): Ditto.
32507 (_mm512_mask_i64scatter_epi64): Ditto.
32508 * config/i386/avx512pfintrin.h
32509 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32510 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32511 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32512 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32513 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32514 (_mm512_prefetch_i32scatter_ps): Ditto.
32515 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32516 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32517 (_mm512_prefetch_i64scatter_pd): Ditto.
32518 (_mm512_prefetch_i64scatter_ps): Ditto.
32519 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32520 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32521 * config/i386/avx512vlintrin.h
32522 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32523 (_mm_mmask_i32gather_ps): Ditto.
32524 (_mm256_mmask_i32gather_pd): Ditto.
32525 (_mm_mmask_i32gather_pd): Ditto.
32526 (_mm256_mmask_i64gather_ps): Ditto.
32527 (_mm_mmask_i64gather_ps): Ditto.
32528 (_mm256_mmask_i64gather_pd): Ditto.
32529 (_mm_mmask_i64gather_pd): Ditto.
32530 (_mm256_mmask_i32gather_epi32): Ditto.
32531 (_mm_mmask_i32gather_epi32): Ditto.
32532 (_mm256_mmask_i32gather_epi64): Ditto.
32533 (_mm_mmask_i32gather_epi64): Ditto.
32534 (_mm256_mmask_i64gather_epi32): Ditto.
32535 (_mm_mmask_i64gather_epi32): Ditto.
32536 (_mm256_mmask_i64gather_epi64): Ditto.
32537 (_mm_mmask_i64gather_epi64): Ditto.
32538 (_mm256_i32scatter_ps): Change __addr type to void*.
32539 (_mm256_mask_i32scatter_ps): Ditto.
32540 (_mm_i32scatter_ps): Ditto.
32541 (_mm_mask_i32scatter_ps): Ditto.
32542 (_mm256_i32scatter_pd): Ditto.
32543 (_mm256_mask_i32scatter_pd): Ditto.
32544 (_mm_i32scatter_pd): Ditto.
32545 (_mm_mask_i32scatter_pd): Ditto.
32546 (_mm256_i64scatter_ps): Ditto.
32547 (_mm256_mask_i64scatter_ps): Ditto.
32548 (_mm_i64scatter_ps): Ditto.
32549 (_mm_mask_i64scatter_ps): Ditto.
32550 (_mm256_i64scatter_pd): Ditto.
32551 (_mm256_mask_i64scatter_pd): Ditto.
32552 (_mm_i64scatter_pd): Ditto.
32553 (_mm_mask_i64scatter_pd): Ditto.
32554 (_mm256_i32scatter_epi32): Ditto.
32555 (_mm256_mask_i32scatter_epi32): Ditto.
32556 (_mm_i32scatter_epi32): Ditto.
32557 (_mm_mask_i32scatter_epi32): Ditto.
32558 (_mm256_i32scatter_epi64): Ditto.
32559 (_mm256_mask_i32scatter_epi64): Ditto.
32560 (_mm_i32scatter_epi64): Ditto.
32561 (_mm_mask_i32scatter_epi64): Ditto.
32562 (_mm256_i64scatter_epi32): Ditto.
32563 (_mm256_mask_i64scatter_epi32): Ditto.
32564 (_mm_i64scatter_epi32): Ditto.
32565 (_mm_mask_i64scatter_epi32): Ditto.
32566 (_mm256_i64scatter_epi64): Ditto.
32567 (_mm256_mask_i64scatter_epi64): Ditto.
32568 (_mm_i64scatter_epi64): Ditto.
32569 (_mm_mask_i64scatter_epi64): Ditto.
32570 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32571 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32572 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32573 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32574 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32575 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32576 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32577 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32578 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32579 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32580 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32581 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32582 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32583 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32584 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32585 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32586 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32587 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32588 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32589 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32590 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32591 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32592 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32593 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32594 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32595 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32596 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32597 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32598 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32599 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32600 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32601 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32602 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32603 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32604 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32605 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32606 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32607 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32608 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32609 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32610 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32611 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32612 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32613 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32614 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32615 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32616 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32617 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32618 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32619 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32620 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32621 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32622 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32623 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32624 definitions accordingly.
32626 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32627 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32630 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32633 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32636 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32637 hard regs for inner regno.
32639 2017-01-17 Martin Liska <mliska@suse.cz>
32642 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32643 assumption and add comment.
32645 2017-01-17 Nathan Sidwell <nathan@acm.org>
32647 * ipa-visibility.c (localize_node): New function, broken out of ...
32648 (function_and_variable_visibility): ... here. Call it.
32650 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32652 PR middle-end/77445
32653 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32654 correctly set frequency of oudgoing edge.
32655 (duplicate_thread_path): Fix profile updating.
32657 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32660 * configure.ac: Add GCC_BASE_VER.
32661 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
32662 version from BASE-VER file.
32663 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
32664 (gcc.o): Depend on $(BASEVER).
32665 * common.opt (dumpfullversion): New option.
32666 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
32667 * doc/invoke.texi: Document -dumpfullversion.
32668 * doc/install.texi: Document --with-gcc-major-version-only.
32669 * configure: Regenerated.
32671 2017-01-17 Richard Biener <rguenther@suse.de>
32673 PR tree-optimization/71433
32674 * tree-vrp.c (register_new_assert_for): Merge same asserts
32675 on all incoming edges.
32676 (process_assert_insertions_for): Handle insertions at the
32679 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32681 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32682 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32684 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32687 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32690 2017-01-17 Alan Modra <amodra@gmail.com>
32693 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32694 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32695 symbolic stack limit when pic.
32697 2017-01-16 Martin Sebor <msebor@redhat.com>
32699 PR tree-optimization/78608
32700 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32702 2017-01-16 Jeff Law <law@redhat.com>
32705 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32706 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32707 for several include directories that may be relative to sysroot.
32708 * config/i386/x-mingw32 (gplus_includedir): Define.
32709 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32710 (native_system_includedir): Likewise.
32711 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32712 override if TARGET_SYSTEM_ROOT is defined.
32713 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32715 PR tree-optimization/79090
32716 PR tree-optimization/33562
32717 PR tree-optimization/61912
32718 PR tree-optimization/77485
32719 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32720 and computed trims into the dump file.
32722 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32724 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32726 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32729 * gimplify.c (gimplify_init_constructor): If want_value and
32730 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32734 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32735 sequence. Formatting fixes.
32736 (doloop_optimize): Formatting fixes.
32739 * gcc.c (debug_level_greater_than_spec_func): New function.
32740 (static_spec_functions): Add debug-level-gt spec function.
32741 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32743 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32744 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32745 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32746 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32747 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32748 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32750 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32752 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32753 QImode fixups to general and mask registers only.
32755 2017-01-16 Carl Love <cel@us.ibm.com>
32757 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32758 for built-in functions
32759 vector signed char vec_nabs (vector signed char)
32760 vector signed short vec_nabs (vector signed short)
32761 vector signed int vec_nabs (vector signed int)
32762 vector signed long long vec_nabs (vector signed long long)
32763 vector float vec_nabs (vector float)
32764 vector double vec_nabs (vector double)
32765 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32767 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32768 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32769 * doc/extend.texi: Update the documentation file for the new built-in
32772 2017-01-16 Martin Sebor <msebor@redhat.com>
32774 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32777 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32779 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32780 UNSPEC_VSX__XXSPLTD to require special splat handling.
32782 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32785 * system.h: Poison strndup.
32787 2017-01-16 Alan Modra <amodra@gmail.com>
32790 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32793 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32795 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32797 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32799 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32800 call recog here. Assert that INSN_CODE (insn) is non-negative.
32802 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32805 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32807 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32808 in the currently scheduled RTL fragment.
32810 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32812 PR rtl-optimization/78751
32813 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32816 2017-01-14 Jeff Law <law@redhat.com>
32818 PR tree-optimization/79090
32819 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32820 variable length stores.
32821 (compute_trims): Delete dead assignment to *trim_tail.
32822 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32825 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32827 PR rtl-optimization/78626
32828 PR rtl-optimization/78727
32829 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32830 of a block, and split such blocks after everything else is finished.
32832 2017-01-14 Alan Modra <amodra@gmail.com>
32835 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32836 target legitimate_combined_insn.
32837 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32838 (rs6000_legitimate_combined_insn): New function.
32839 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32841 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32842 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32843 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32845 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32847 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32849 2017-01-13 Jeff Law <law@redhat.com>
32851 PR tree-optimization/33562
32852 PR tree-optimization/61912
32853 PR tree-optimization/77485
32854 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32856 (delete_dead_assignment): Likewise.
32857 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32858 statement to delete_dead_call and delete_dead_assignment.
32860 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32863 * substring-locations.c (format_warning_va): Strengthen case 1 so
32864 that both endpoints of the substring must be within the format
32865 range for just the substring to be printed.
32867 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32869 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32870 * config/i386/i386.c (ix86_target_string): Add missing options
32871 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32872 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32873 flags_other and ix86_target_other to flags2_other. Display unknown
32875 (ix86_valid_target_attribute_inner_p): Add missing options and
32876 reorder options by implied ISAs, as in ix86_target_string.
32878 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32880 * hash-table.h (hash_table::too_empty_p): New function.
32881 (hash_table::expand): Use it.
32882 (hash_table::traverse): Likewise.
32883 (hash_table::empty_slot): Use sizeof (value_type) instead of
32884 sizeof (PTR) to convert bytes to elements. Shrink the table
32885 if the current size is excessive for the current number of
32888 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32890 * ira-costs.c (record_reg_classes): Break from the inner loop
32891 early once alt_fail is known to be true. Update outer loop
32892 handling accordingly.
32894 2017-01-13 Jeff Law <law@redhat.com>
32896 * tree-ssa-dse.c (decrement_count): New function.
32897 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32898 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32899 when we know the partially dead statement is a mem* function.
32901 PR tree-optimization/61912
32902 PR tree-optimization/77485
32903 * tree-ssa-dse.c: Include expr.h.
32904 (maybe_trim_constructor_store): New function.
32905 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32907 PR tree-optimization/33562
32908 PR tree-optimization/61912
32909 PR tree-optimization/77485
32910 * doc/invoke.texi: Document new dse-max-object-size param.
32911 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32912 * tree-ssa-dse.c: Include params.h.
32913 (dse_store_status): New enum.
32914 (initialize_ao_ref_for_dse): New, partially extracted from
32916 (valid_ao_ref_for_dse, normalize_ref): New.
32917 (setup_live_bytes_from_ref, compute_trims): Likewise.
32918 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32919 (maybe_trim_partially_dead_store): Likewise.
32920 (maybe_trim_complex_store): Likewise.
32921 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32922 Track what bytes live from the original store. Return tri-state
32923 for dead, partially dead or live.
32924 (dse_dom_walker): Add constructor, destructor and new private members.
32925 (delete_dead_call, delete_dead_assignment): New extracted from
32927 (dse_optimize_stmt): Make a member of dse_dom_walker.
32928 Use initialize_ao_ref_for_dse.
32930 PR tree-optimization/33562
32931 PR tree-optimization/61912
32932 PR tree-optimization/77485
32933 * sbitmap.h (bitmap_count_bits): Prototype.
32934 (bitmap_clear_range, bitmap_set_range): Likewise.
32935 * sbitmap.c (bitmap_clear_range): New function.
32936 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32938 2017-01-13 Martin Liska <mliska@suse.cz>
32941 * function.c (set_cfun): Add new argument force.
32942 * function.h (set_cfun): Likewise.
32943 * ipa-inline-transform.c (inline_call): Use the function when
32944 strict alising from is dropped for function we inline to.
32946 2017-01-13 Richard Biener <rguenther@suse.de>
32948 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32949 for dumping GIMPLE INTEGER_CSTs.
32951 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32953 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32954 to 201112L since C++17.
32956 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32959 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32960 if -fsanitize=kernel-address is present.
32962 2017-01-13 Richard Biener <rguenther@suse.de>
32964 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32965 as _Literal ( type ) number in case usual suffixes do not
32966 preserve all information.
32968 2017-01-13 Richard Biener <rguenther@suse.de>
32970 PR tree-optimization/77283
32971 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
32972 and ssa-iterators.h.
32973 (is_feasible_trace): Implement a cost model based on joiner
32976 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
32979 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
32980 char or short to __float128/_Float128 directly.
32982 2017-01-12 Martin Sebor <msebor@redhat.com>
32984 to -Wformat-overflow.
32985 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
32986 (min_bytes_remaining): Same.
32987 (get_string_length): Same.
32988 (format_string): Same.
32989 (format_directive): Same.
32991 (pass_sprintf_length::handle_gimple_call): Same.
32993 2017-01-12 Jakub Jelinek <jakub@redhat.com>
32995 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
32996 info.nowrite calls with no lhs that can't throw. Return bool
32997 whether gsi_remove has been called or not.
32998 (pass_sprintf_length::handle_gimple_call): Return bool whether
32999 try_substitute_return_value called gsi_remove. Formatting fix.
33000 (pass_sprintf_length::execute): Don't use gsi_remove if
33001 handle_gimple_call returned true.
33004 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33005 be removed due to side-effects, don't remove following barrier nor
33006 turn the successor edge into fallthru edge.
33008 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33011 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33012 element-reversing loads and stores as not swappable.
33014 2017-01-12 Nathan Sidwell <nathan@acm.org>
33015 Nicolai Stange <nicstange@gmail.com>
33017 * combine.c (try_combine): Don't ignore result of overlap checking
33018 loop. Combine overlap & asm check into single loop.
33020 2017-01-12 Richard Biener <rguenther@suse.de>
33022 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33023 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33025 2017-01-12 Richard Biener <rguenther@suse.de>
33027 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33028 and TS_TARGET_OPTION directly derive from TS_BASE.
33029 * tree-core.h (tree_optimization_option): Derive from tree_base.
33030 (tree_target_option): Likewise.
33032 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33034 * config/i386/i386.c (memory_address_length): Increase len
33035 only when rip_relative_addr_p returns false.
33037 2017-01-11 Julia Koval <julia.koval@intel.com>
33039 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33040 (OPTION_MASK_ISA_SGX_SET): New.
33041 (ix86_handle_option): Handle OPT_msgx.
33042 * config.gcc: Added sgxintrin.h.
33043 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33044 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33045 * config/i386/i386.c (ix86_target_string): Add -msgx.
33047 (ix86_option_override_internal): Handle new options.
33048 (ix86_valid_target_attribute_inner_p): Add sgx.
33049 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33050 * config/i386/i386.opt: Add msgx.
33051 * config/i386/sgxintrin.h: New file.
33052 * config/i386/x86intrin.h: Add sgxintrin.h.
33054 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33057 * fold-const.c (maybe_nonzero_address): Return 1 for function
33059 (tree_single_nonzero_warnv_p): Don't handle function local objects
33063 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33066 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33069 * opts.c: Include "spellcheck.h"
33070 (struct string_fragment): New struct.
33071 (struct edit_distance_traits<const string_fragment &>): New
33073 (get_closest_sanitizer_option): New function.
33074 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33076 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33078 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33080 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33081 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33082 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33083 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33084 for initial die_offset if dwarf_split_debug_info.
33085 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33086 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33088 (output_skeleton_debug_sections): Formatting fix. Use
33089 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33090 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33092 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33094 * config/arm/cortex-a53.md: Add bypasses for
33095 cortex_a53_r2f_cvt.
33096 (cortex_a53_r2f): Only use for transfers.
33097 (cortex_a53_f2r): Likewise.
33098 (cortex_a53_r2f_cvt): Add reservation for conversions.
33099 (cortex_a53_f2r_cvt): Likewise.
33101 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33103 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33104 to all inlined functions, change static to extern.
33106 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33109 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33111 (arm_assemble_integer): Likewise.
33113 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33115 * config.gcc: Use new awk script to check CPU, FPU and architecture
33116 parameters for --with-... options.
33117 * config/arm/parsecpu.awk: New file
33118 * config/arm/arm-cpus.in: New file.
33119 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33121 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33123 * config/arm/t-arm: Update dependency rules.
33124 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33125 of processing .def files.
33126 * config/arm/genopt.sh: Deleted.
33127 * config/arm/gentune.sh: Deleted.
33128 * config/arm/arm-cores.def: Deleted.
33129 * config/arm/arm-arches.def: Deleted.
33130 * config/arm/arm-fpus.def: Deleted.
33131 * config/arm/arm-tune.md: Regenerated.
33132 * config/arm/arm-tables.opt: Regenerated.
33133 * config/arm/arm-cpu.h: New generated file.
33134 * config/arm/arm-cpu-data.h: New generated file.
33135 * config/arm/arm-cpu-cdata.h: New generated file.
33137 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33140 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33142 (input_varpool_node): Unpack dynamically_initialized bit.
33144 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33146 PR rtl-optimization/79032
33147 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33148 the alignment of the adjusted memory reference against that of MODE,
33149 instead of the alignment of the original memory reference.
33151 2017-01-11 Martin Jambor <mjambor@suse.cz>
33153 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
33155 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
33156 decorated functions.
33158 2017-01-11 Richard Biener <rguenther@suse.de>
33160 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
33161 set range/nonnull info for PHI results. Do not set it on
33162 stmts marked for removal.
33164 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
33166 * expr.c (store_field): In the bitfield case, fetch the return value
33167 from the registers before applying a single big-endian adjustment.
33168 Always do a final load for a BLKmode value not larger than a word.
33170 2017-01-10 David Malcolm <dmalcolm@redhat.com>
33173 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33174 that we correctly handle column numbers greater than
33175 LINE_MAP_MAX_COLUMN_NUMBER.
33177 2017-01-10 Martin Sebor <msebor@redhat.com>
33179 PR middle-end/78245
33180 * gimple-ssa-sprintf.c (get_destination_size): Call
33181 {init,fini}object_sizes.
33182 * tree-object-size.c (addr_object_size): Adjust.
33183 (pass_through_call): Adjust.
33184 (pass_object_sizes::execute): Adjust.
33185 * tree-object-size.h (fini_object_sizes): Declare.
33187 2017-01-10 Martin Sebor <msebor@redhat.com>
33189 PR tree-optimization/78775
33190 * builtins.c (get_size_range): Move...
33191 * calls.c: ...to here.
33192 (alloc_max_size): Accept zero argument.
33193 (operand_signed_p): Remove.
33194 (maybe_warn_alloc_args_overflow): Call get_size_range.
33195 * calls.h (get_size_range): Declare.
33197 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
33199 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
33200 from TI's devices.csv file as of September 2016.
33201 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
33203 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
33205 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
33206 * doc/invoke.texi: Likewise.
33207 * doc/md.texi: Likewise.
33208 * doc/objc.texi: Likewise.
33210 2017-01-10 Joshua Conner <joshconner@google.com>
33212 * config/arm/fuchsia-elf.h: New file.
33213 * config/fuchsia.h: New file.
33214 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
33215 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
33217 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33219 2016-01-10 Richard Biener <rguenther@suse.de>
33221 PR tree-optimization/79034
33222 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33223 Propagate out degenerate PHIs in the joiner.
33225 2017-01-10 Martin Liska <mliska@suse.cz>
33227 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33228 (sort_congruence_classes_by_decl_uid): Likewise.
33229 (sort_congruence_class_groups_by_decl_uid): Likewise.
33230 (sem_item_optimizer::merge_classes): Sort class, groups in these
33231 classes and members in the groups by DECL_UID of declarations.
33232 This would make merge operations stable.
33234 2017-01-10 Martin Liska <mliska@suse.cz>
33236 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33237 usage of m_classes_vec.
33238 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33239 (sem_item_optimizer::get_group_by_hash): Likewise.
33240 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33241 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33242 (sem_item_optimizer::verify_classes): Likewise.
33243 (sem_item_optimizer::process_cong_reduction): Likewise.
33244 (sem_item_optimizer::dump_cong_classes): Likewise.
33245 (sem_item_optimizer::merge_classes): Likewise.
33246 * ipa-icf.h (congruence_class_hash): Rename from
33247 congruence_class_group_hash. Remove declaration of m_classes_vec.
33249 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33251 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33252 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33253 * config.gcc: Add avx512vpopcntdqintrin.h.
33254 * config/i386/avx512vpopcntdqintrin.h: New.
33255 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33256 * config/i386/i386-builtin-types.def: Add new types.
33257 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33258 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33259 __builtin_ia32_vpopcountq_v8di_mask): New.
33260 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33261 __AVX512VPOPCNTDQ__.
33262 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33263 (PTA_AVX512VPOPCNTDQ): Define.
33264 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33265 TARGET_AVX512VPOPCNTDQ_P): Define.
33266 * config/i386/i386.opt: Add mavx512vpopcntdq.
33267 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33268 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33270 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33272 PR middle-end/77484
33273 * predict.def (PRED_CALL): Set to 67.
33275 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33277 * expr.c (store_field): In the bitfield case, if the value comes from
33278 a function call and is of an aggregate type returned in registers, do
33279 not modify the field mode; extract the value in all cases if the mode
33280 is BLKmode and the size is not larger than a word.
33282 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33285 * config/i386/cpuid.h: Fix undefined behavior.
33287 2017-01-04 Jeff Law <law@redhat.com>
33289 PR tree-optimization/79007
33290 PR tree-optimization/67955
33291 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33292 conservative for pt.null when flag_non_call_exceptions is on.
33294 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33296 PR translation/79019
33297 PR translation/79020
33298 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33299 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33300 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33302 * config/avr/avr.opt (maccumulate-args): Likewise.
33303 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33304 * common.opt (freport-bug): Likewise.
33305 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33306 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33307 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33308 translatable string.
33309 * config/i386/i386.c (function_value_32): Likewise.
33310 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33311 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33313 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33314 * common/config/msp430/msp430-common.c (msp430_handle_option):
33316 * symtab.c (symtab_node::verify_base): Likewise.
33317 * opts.c (set_debug_level): Likewise.
33318 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33319 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33320 missing whitespace to translatable strings.
33321 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33322 * config/sh/superh.h: Likewise.
33323 * config/i386/xopintrin.h: Likewise.
33324 * config/i386/znver1.md: Likewise.
33325 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33326 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33327 * double-int.h (struct double_int): Likewise.
33328 * double-int.c (div_and_round_double): Likewise.
33329 * wide-int.cc: Likewise.
33330 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33331 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33332 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33333 (crossjumps_occurred): ... this.
33334 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33337 PR tree-optimization/78899
33338 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33339 returning bool return struct loop *, NULL for failure and the new
33341 (versionable_outer_loop_p): Don't version outer loop if it has
33342 dont_vectorized bit set.
33343 (tree_if_conversion): When versioning outer loop, ensure
33344 tree_if_conversion is performed also on the inner loop of the
33345 non-vectorizable outer loop copy.
33346 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33347 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33348 prevent vectorization of it.
33349 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33350 the outer loop vectorization of the non-scalar version is attempted
33351 before vectorization of the inner loop in scalar version. If
33352 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33353 vectorization of its inner loop.
33354 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33355 has 2 inner loops, rename also on edges from bb whose single pred
33356 is outer_loop->header. Fix typo in function comment.
33358 2017-01-09 Martin Sebor <msebor@redhat.com>
33361 * asan.c (asan_emit_stack_protection): Increase local buffer size
33362 to avoid snprintf truncation warning.
33364 2017-01-09 Andrew Pinski <apinski@cavium.com>
33366 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33367 to reference thunderx2t99 for the tuning structure
33368 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33370 (thunderx2t99_extra_costs): This.
33371 * config/aarch64/aarch64-tune.md: Regenerate.
33372 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33373 (vulcan_addrcost_table): This.
33374 (vulcan_regmove_cost): Rename to ...
33375 (thunderx2t99_regmove_cost): This.
33376 (vulcan_vector_cost): Rename to ...
33377 (thunderx2t99_vector_cost): this.
33378 (vulcan_branch_cost): Rename to ...
33379 (thunderx2t99_branch_cost): This.
33380 (vulcan_tunings): Rename to ...
33381 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33382 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33384 2017-01-09 Martin Jambor <mjambor@suse.cz>
33388 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33389 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33390 (propagate_vr_accross_jump_function): Use the above function for all
33391 value range computations for pass-through jump functions and type
33392 converasion from explicit value range values.
33393 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33394 parameters from TYPE_ARG_TYPES.
33395 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33396 (ipa_write_node_info): Stream type of the actual argument.
33397 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33399 2017-01-09 Martin Liska <mliska@suse.cz>
33402 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33403 (lookup_compiler): Do not show error message with have_E.
33405 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33407 PR tree-optimization/78938
33408 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33409 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33410 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33411 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33414 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33416 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33419 2017-01-09 Richard Biener <rguenther@suse.de>
33421 PR tree-optimization/78997
33422 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33423 name condition properly.
33425 2017-01-09 Richard Biener <rguenther@suse.de>
33428 * dwarf2out.c (is_cxx): New overload with context.
33429 (is_naming_typedef_decl): Use it.
33431 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33433 * invoke.texi (Option Summary): Correct spacing in option lists
33434 and add line breaks to fix over-long lines.
33436 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33438 PR middle-end/17660
33440 * extend.texi (Common Variable Attributes): Add xref to GCC
33441 Internals manual to explain mode attribute keywords.
33443 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33446 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33447 and Preprocessor Options.
33448 (Options for Linking): Document -pthread here....
33449 (RS/6000 and PowerPC Options): ...not here.
33450 (Solaris 2 Options): ...or here.
33451 * doc/cppopts.texi: Document -pthread.
33453 2017-01-08 Martin Sebor <msebor@redhat.com>
33455 PR middle-end/77708
33456 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33457 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33458 New member functions.
33459 (format_directive): Used them.
33461 (pass_sprintf_length::handle_gimple_call): Same.
33462 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33463 to avoid truncation for any argument.
33464 (extract_affine_mul): Same.
33465 * tree.c (get_file_function_name): Same.
33467 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33469 PR middle-end/77484
33470 * predict.def (PRED_INDIR_CALL): Set to 86.
33472 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33474 PR preprocessor/54124
33475 * doc/cppopts.texi: Reformat -d subtable to list the full name
33476 of the options. Add cross-reference to the docs for the general
33477 compiler -d options.
33478 * doc/invoke.texi (Developer Options): Add cross-reference to the
33479 preprocessor-specific -d option documentation.
33481 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33483 PR preprocessor/13498
33484 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33485 redudant material, and reflect new command-line options.
33486 (System Headers): Likewise.
33488 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33490 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33491 -isystem, and -idirafter. Copy-edit.
33492 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33493 default for -ftrack-macro-expansion. Delete obsolete and
33494 badly-formatted implementation details about -fdebug-cpp output.
33495 * doc/cppwarnopts.texi: Copy-edit.
33497 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33500 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33501 that the transition from a max line width >= 1<<10 to narrower
33502 lines works correctly.
33504 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33506 * doc/options.texi (PerFunction): New.
33507 * opt-functions.awk (switch_flags): Map both Optimization and
33508 PerFunction to CL_OPTIMIZATION.
33509 * opth-gen.awk: Test for PerFunction flag along with
33511 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33512 it only when the latter is present. Skip those that don't in
33513 the hash function generator.
33514 * common.opt (fvar-tracking): Mark as PerFunction instead of
33516 (fvar-tracking-assignments): Likewise.
33517 (fvar-tracking-assignments-toggle): Likewise.
33518 (fvar-tracking-uninit): Likewise.
33520 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33522 PR translation/79018
33523 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33526 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33529 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33530 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33531 TARGET_LONG_JUMP_TABLE_OFFSETS.
33532 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33533 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33534 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33535 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33536 * config/m68k/m68k.md (tablejump expander): Likewise.
33537 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33538 TARGET_LONG_JUMP_TABLE_OFFSETS.
33539 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33540 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33542 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33543 David Holsgrove <david.holsgrove@xilinx.com>
33545 * common/config/microblaze/microblaze-common.c
33546 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33547 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33549 * config/microblaze/microblaze.c (microblaze_must_save_register)
33550 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33552 (microblaze_eh_return): New function.
33553 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33554 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33555 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33556 * config/microblaze/microblaze.md (eh_return): New pattern.
33558 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33560 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33561 GCC_DIAGNOSTIC_STRINGIFY): Define.
33563 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33565 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33567 * config/arm/arm.md (<mcrr>): New.
33569 * config/arm/arm.c (arm_arch5te): New.
33570 (arm_option_override): Set arm_arch5te.
33571 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33573 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33574 (arm_mcrr_qualifiers): ... this. New.
33575 (MRRC_QUALIFIERS): Define to...
33576 (arm_mrrc_qualifiers): ... this. New.
33577 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33579 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33580 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33581 (MRRCI, mrrc, MRRC): New.
33582 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33583 VUNSPEC_MRRC2): New.
33585 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33587 * config/arm/arm.md (<mcr>): New.
33589 * config/arm/arm.c (arm_coproc_builtin_available): Add
33590 support for mcr, mrc, mcr2 and mrc2.
33591 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33592 (arm_mcr_qualifiers): ... this. New.
33593 (MRC_QUALIFIERS): Define to ...
33594 (arm_mrc_qualifiers): ... this. New.
33595 (MCR_QUALIFIERS): Define to ...
33596 (arm_mcr_qualifiers): ... this. New.
33597 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33599 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33600 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33601 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33602 VUNSPEC_MRC2): New.
33604 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33606 * config/arm/arm.md (*ldc): New.
33610 * config/arm/arm.c (arm_coproc_builtin_available): Add
33611 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33612 (arm_coproc_ldc_stc_legitimate_address): New.
33613 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33614 'qualifier_const_pointer'.
33615 (LDC_QUALIFIERS): Define to...
33616 (arm_ldc_qualifiers): ... this. New.
33617 (STC_QUALIFIERS): Define to...
33618 (arm_stc_qualifiers): ... this. New.
33619 * config/arm/arm-protos.h
33620 (arm_coproc_ldc_stc_legitimate_address): New.
33621 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33622 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33623 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33624 stc2, stcl, stc2l): New.
33625 * config/arm/constraints.md (Uz): New.
33626 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33627 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33628 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33629 VUNSPEC_STC2L): New.
33631 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33633 * config/arm/arm.md (<cdp>): New.
33634 * config/arm/arm.c (neon_const_bounds): Rename this ...
33635 (arm_const_bounds): ... this.
33636 (arm_coproc_builtin_available): New.
33637 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33638 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33639 (CDP_QUALIFIERS): Define to...
33640 (arm_cdp_qualifiers): ... this. New.
33642 (arm_expand_builtin_args): Add case for 6 arguments.
33643 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33644 (arm_const_bounds): ... this.
33645 (arm_coproc_builtin_available): New.
33646 * config/arm/arm_acle.h (__arm_cdp): New.
33648 * config/arm/arm_acle_builtins.def (cdp): New.
33650 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33651 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33652 'arm_const_bounds'.
33653 * config/arm/types.md (coproc): New.
33654 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
33655 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
33656 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
33657 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
33659 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33661 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
33662 (UBINOP_QUALIFIERS): New.
33664 (acle_builtin_data): New. Change comment.
33665 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
33666 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
33667 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
33668 arm_acle_builtins.def.
33669 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
33670 (arm_init_acle_builtins): New.
33671 (CRC32_BUILTIN): Remove.
33672 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33673 crc32cb, crc32ch and crc32cw.
33674 (arm_init_crc32_builtins): Remove.
33675 (arm_init_builtins): Use arm_init_acle_builtins rather
33676 than arm_init_crc32_builtins.
33677 (arm_expand_acle_builtin): New.
33678 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33679 * config/arm/arm_acle_builtins.def: New.
33681 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33683 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33684 (arm_builtin_datum): ... this.
33685 (arm_init_neon_builtin): Rename to ...
33686 (arm_init_builtin): ... this. Add a new parameters PREFIX
33687 and USE_SIG_IN_NAME.
33688 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33689 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33690 'arm_builtin_datum'.
33691 (arm_init_vfp_builtins): Likewise.
33692 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33693 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33694 (arm_expand_neon_args): Rename to ...
33695 (arm_expand_builtin_args): ... this. Rename builtin_arg
33696 enum values and differentiate between ARG_BUILTIN_MEMORY
33697 and ARG_BUILTIN_NEON_MEMORY.
33698 (arm_expand_neon_builtin_1): Rename to ...
33699 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33700 values, arm_expand_builtin_args and add bool parameter NEON.
33701 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33702 (arm_expand_vfp_builtin): Likewise.
33703 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33705 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33707 PR middle-end/77484
33708 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33709 * predict.c (tree_estimate_probability_bb): Reverse direction of
33710 polymorphic call predictor.
33712 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33714 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33715 (determine_pass_name_match): ...this new function and...
33716 (should_skip_pass_p): ...this new function.
33718 2017-01-06 Nathan Sidwell <nathan@acm.org>
33720 * ipa-visibility.c (function_and_variable_visibility): Reformat
33721 comments and long lines. Remove extrneous if.
33722 * symtab.c (symtab_node::make_decl_local): Fix code format.
33723 (symtab_node::set_section_for_node): Fix comment typo.
33725 2017-01-06 Martin Liska <mliska@suse.cz>
33728 * lra-constraints.c: Rename invariant to lra_invariant.
33729 * predict.c (set_even_probabilities): Initialize e to NULL.
33731 2017-01-05 Martin Sebor <msebor@redhat.com>
33733 PR tree-optimization/78910
33734 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33735 (format_integer): Correct off-by-one error in the handling
33736 of precision with negative numbers in signed conversions..
33738 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33740 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33742 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33744 PR tree-optimization/71016
33745 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33746 factor_out_conditional_conversion. Formatting fix.
33747 (factor_out_conditional_conversion): Add cond_stmt argument.
33748 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33749 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33752 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33754 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33755 read-rtl-function.o, and selftest-rtl.o.
33756 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33757 (selftest::aarch64_test_loading_full_dump): New function.
33758 (selftest::aarch64_run_selftests): New function.
33759 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33760 selftest::aarch64_run_selftests.
33761 * config/i386/i386.c
33762 (selftest::ix86_test_loading_dump_fragment_1): New function.
33763 (selftest::ix86_test_loading_call_insn): New function.
33764 (selftest::ix86_test_loading_full_dump): New function.
33765 (selftest::ix86_test_loading_unspec): New function.
33766 (selftest::ix86_run_selftests): Call the new functions.
33767 * emit-rtl.c (maybe_set_max_label_num): New function.
33768 * emit-rtl.h (maybe_set_max_label_num): New decl.
33769 * function.c (instantiate_decls): Guard call to
33770 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33771 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33773 * gensupport.c (gen_reader::gen_reader): Pass "false"
33774 for new "compact" param of rtx_reader.
33775 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33776 rather than an empty string for NULL strings.
33777 * read-md.c: Potentially include config.h rather than bconfig.h.
33778 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33779 (have_error): New global, copied from errors.c.
33780 (md_reader::read_name): Rename to...
33781 (md_reader::read_name_1): ...this, adding "out_loc" param,
33782 and converting "missing name or number" to returning false, rather
33784 (md_reader::read_name): Reimplement in terms of read_name_1.
33785 (md_reader::read_name_or_nil): New function.
33786 (md_reader::read_string): Handle "(nil)" by returning NULL.
33787 (md_reader::md_reader): Add new param "compact".
33788 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33789 (md_reader::read_file): New method.
33790 * read-md.h (md_reader::md_reader): Add new param "compact".
33791 (md_reader::read_file): New method.
33792 (md_reader::is_compact): New accessor.
33793 (md_reader::read_name): Convert return type from void to file_location.
33794 (md_reader::read_name_or_nil): New decl.
33795 (md_reader::read_name_1): New decl.
33796 (md_reader::m_compact): New field.
33797 (noop_reader::noop_reader): Pass "false" for new "compact" param
33799 (rtx_reader::rtx_reader): Add new "compact" param.
33800 (rtx_reader::read_rtx_operand): Make virtual and convert return
33801 type from void to rtx.
33802 (rtx_reader::read_until): New decl.
33803 (rtx_reader::handle_any_trailing_information): New virtual function.
33804 (rtx_reader::postprocess): New virtual function.
33805 (rtx_reader::finalize_string): New virtual function.
33806 (rtx_reader::m_in_call_function_usage): New field.
33807 (rtx_reader::m_reuse_rtx_by_id): New field.
33808 * read-rtl-function.c: New file.
33809 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33810 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33811 (selftest::verify_three_block_rtl_cfg): New decl.
33812 * read-rtl-function.h: New file.
33813 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33814 For host, include function.h, memmodel.h, and emit-rtl.h.
33815 (one_time_initialization): New function.
33816 (struct compact_insn_name): New struct.
33817 (compact_insn_names): New array.
33818 (find_code): Handle insn codes in compact dumps.
33819 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33820 (bind_subst_iter_and_attr): Likewise.
33821 (add_condition_to_string): Likewise.
33822 (add_condition_to_rtx): Likewise.
33823 (apply_attribute_uses): Likewise.
33824 (add_current_iterators): Likewise.
33825 (apply_iterators): Likewise.
33826 (initialize_iterators): Guard usage of apply_subst_iterator with
33827 #ifdef GENERATOR_FILE.
33828 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33829 (md_reader::read_mapping): Likewise.
33830 (add_define_attr_for_define_subst): Likewise.
33831 (add_define_subst_attr): Likewise.
33832 (read_subst_mapping): Likewise.
33833 (check_code_iterator): Likewise.
33834 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33836 (one_time_initialization): New function.
33837 (rtx_reader::read_until): New method.
33838 (read_flags): New function.
33839 (parse_reg_note_name): New function.
33840 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33841 Handle reuse_rtx ids.
33842 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33843 Add parsing support for RTL dumps, mirroring the special-cases in
33844 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33845 values, and calling handle_any_trailing_information.
33846 (rtx_reader::read_rtx_operand): Convert return type from void
33847 to rtx, returning return_rtx. Handle case 'e'. Call
33848 finalize_string on XSTR and XTMPL fields.
33849 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33850 "(nil)" values were omitted. Call the postprocess vfunc on the
33852 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33853 class ctor. Initialize m_in_call_function_usage. Call
33854 one_time_initialization.
33855 * rtl-tests.c (selftest::test_uncond_jump): Call
33856 set_new_first_and_last_insn.
33857 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33858 * selftest-rtl.c: New file.
33859 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33860 (selftest::get_insn_by_uid): New decl.
33861 * selftest-run-tests.c (selftest::run_tests): Call
33862 read_rtl_function_c_tests.
33863 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33864 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33867 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33869 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33870 operands in a special way. Assert that pos+len <= mode precision.
33872 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33874 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33875 3 argument Alias with unlimited for the negative form.
33876 (fno-vect-cost-model): Removed.
33878 2017-01-05 Martin Liska <mliska@suse.cz>
33880 * hsa-gen.c (gen_hsa_divmod): New function.
33881 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33883 2017-01-05 Martin Liska <mliska@suse.cz>
33886 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33889 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33891 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33892 small constant length operands.
33894 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33896 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33897 between loop iterations.
33899 2017-01-05 Martin Liska <mliska@suse.cz>
33902 * gimplify.c (gimplify_decl_expr): Compare to
33903 asan_poisoned_variables instread of checking flags.
33904 (gimplify_target_expr): Likewise.
33905 (gimplify_expr): Likewise.
33906 (gimplify_function_tree): Conditionally initialize
33907 asan_poisoned_variables.
33909 2017-01-04 Jeff Law <law@redhat.com>
33911 PR tree-optimizatin/78812
33912 * rtl.h (contains_mem_rtx_p): Prototype.
33913 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33914 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33915 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33916 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33917 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33919 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33921 * input.c (assert_char_at_range): Default-initialize actual_range.
33923 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33925 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33926 to match the caller.
33928 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33930 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33931 insns after final jump in test to emit dummy move.
33933 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33935 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33936 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33938 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33940 * multiple_target.c (create_dispatcher_calls): Init e_next.
33941 * tree-ssa-loop-split.c (split_loop): Init border.
33942 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33945 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33950 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33951 (altivec_register_operand): Do not return true if the operand
33952 contains a SUBREG mixing SImode and SFmode.
33953 (vsx_register_operand): Likewise.
33954 (vsx_reg_sfsubreg_ok): New predicate.
33955 (vfloat_operand): Do not return true if the operand contains a
33956 SUBREG mixing SImode and SFmode.
33957 (vint_operand): Likewise.
33958 (vlogical_operand): Likewise.
33959 (gpc_reg_operand): Likewise.
33960 (int_reg_operand): Likewise.
33961 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33962 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33963 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33965 (rs6000_emit_move_si_sf_subreg): New helper function.
33966 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
33967 fixup SUBREGs involving SImode and SFmode.
33968 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
33969 numbers for the new peephole2 optimization.
33970 (peephole2 for SFmode unions): New peephole2 to optimize cases in
33971 the GLIBC math library that do AND/IOR/XOR operations on single
33972 precision floating point.
33973 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
33974 target macros to say whether we need to avoid SUBREGs mixing
33976 (TARGET_ALLOW_SF_SUBREG): Likewise.
33977 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
33978 (UNSPEC_SI_FROM_SF): Likewise.
33979 (iorxor): Change spacing.
33980 (and_ior_xor): New iterator for AND, IOR, and XOR.
33981 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
33982 (movdi_from_sf_zero_ext): Likewise.
33983 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
33984 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
33985 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
33986 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
33987 (fms<mode>4): Likewise.
33988 (fnma<mode>4): Likewise.
33989 (fnms<mode>4): Likewise.
33990 (nfma<mode>4): Likewise.
33991 (nfms<mode>4): Likewise.
33993 2017-01-04 Marek Polacek <polacek@redhat.com>
33996 * doc/invoke.texi: Document -Wpointer-compare.
33998 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34000 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34003 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34004 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34007 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34010 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34011 documentation of the powerpc_popcntb_ok attribute.
34012 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34013 code to issue warning messages if a requested CPU configuration is
34014 not supported by the binary (assembler and loader) toolchain.
34015 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34016 made to define a built-in function that has been disabled.
34017 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34018 made to define a built-in function that has been disabled.
34019 (altivec_init_builtins): Add comment explaining why definition
34020 of the DST built-in functions is not preceded by an assertion
34021 check. Add assertions to prevent ICE if attempts are made to
34022 define an altivec predicate or an abs* built-in function that has
34024 (htm_init_builtins): Add comment explaining why definition of the
34025 htm built-in functions is not preceded by an assertion check.
34027 2017-01-04 Jeff Law <law@redhat.com>
34029 PR tree-optimizatin/67955
34030 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34031 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34032 the points-to solution does not include pt_null. Use DECL_PT_UID
34035 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34037 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34038 Use gen_int_mode instead of gen_lopwart for const_int operands.
34040 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34042 PR tree-optimization/71563
34043 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34044 out of range value - has low bits known to be zero.
34046 2017-01-04 Alan Modra <amodra@gmail.com>
34048 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34049 * configure: Regenerate.
34050 * config.in: Regenerate.
34052 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34055 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34056 a substring of the message, but strcmp with the whole message. Ifdef
34057 ENABLE_NLS, translate the message first using dgettext.
34059 2017-01-03 Jeff Law <law@redhat.com>
34061 PR tree-optimizatin/78856
34062 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34063 (mark_threaded_blocks): Remove code to truncate thread paths that
34064 cross multiple loop headers. Instead invalidate the cached loop
34065 iteration information and handle case of a thread path walking
34066 into an irreducible region.
34068 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34071 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34072 assertions. Add support for doing the signbit if the IEEE 128-bit
34073 floating point value is in a GPR.
34074 * config/rs6000/rs6000.md (Fsignbit): Delete.
34075 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34076 Update the length attribute if the value is in a GPR.
34077 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34078 the sign or zero extension instruction, since the value is always 0/1.
34079 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34082 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34083 extracting SImode to a GPR register so that we can generate a
34084 store, limit the vector to be in a traditional Altivec register
34085 for the vextuwrx instruction.
34087 2017-01-03 Ian Lance Taylor <iant@google.com>
34089 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34091 2017-01-03 Martin Sebor <msebor@redhat.com>
34093 PR tree-optimization/78696
34094 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34095 precision. Use MPFR for %f for greater fidelity. Correct handling
34097 (pass_sprintf_length::compute_format_length): Set width and precision
34098 specified by asrerisk to void_node for vararg functions.
34099 (try_substitute_return_value): Adjust dump output.
34101 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34103 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34105 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34107 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34108 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34109 -mlra/-mno-lra was passed to the compiler.
34111 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34113 PR rtl-optimization/65618
34114 * emit-rtl.c (try_split): Move initialization of "before" and
34115 "after" to just before the call to emit_insn_after_setloc.
34117 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34119 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34121 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34123 * dwarf2out.c (gen_enumeration_type_die): When
34124 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34126 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34128 PR tree-optimization/78965
34129 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34130 Change first argument from const call_info & to call_info &. For %n
34131 set info.nowrite to false.
34133 PR middle-end/78901
34134 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34135 possibly throwing calls.
34137 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34138 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34139 and fns handling, rather than in a separate case SSA_NAME.
34141 2017-01-02 Jeff Law <law@redhat.com>
34143 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34144 fixes for first_period and second_period variables.
34146 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34149 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34150 (*insvqi_1): New insn pattern.
34151 (*insvqi_1_mem_rex64): Ditto.
34152 (*insvqi_2): Ditto.
34153 (*insvqi_3): Rename from *insvqi.
34155 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
34157 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
34159 * doc/cfg.texi (Edges): Remove reference to Java.
34160 (Maintaining the CFG): Ditto.
34162 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34164 PR middle-end/77674
34165 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
34166 transparent aliases.
34168 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34170 PR middle-end/77484
34171 * predict.def (PRED_CALL): Update hitrate.
34172 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
34173 * predict.c (tree_estimate_probability_bb): Split CALL predictor
34174 into direct/indirect/polymorphic variants.
34176 2017-01-01 Jakub Jelinek <jakub@redhat.com>
34178 Update copyright years.
34180 * gcc.c (process_command): Update copyright notice dates.
34181 * gcov-dump.c (print_version): Ditto.
34182 * gcov.c (print_version): Ditto.
34183 * gcov-tool.c (print_version): Ditto.
34184 * gengtype.c (create_file): Ditto.
34185 * doc/cpp.texi: Bump @copying's copyright year.
34186 * doc/cppinternals.texi: Ditto.
34187 * doc/gcc.texi: Ditto.
34188 * doc/gccint.texi: Ditto.
34189 * doc/gcov.texi: Ditto.
34190 * doc/install.texi: Ditto.
34191 * doc/invoke.texi: Ditto.
34193 Copyright (C) 2017 Free Software Foundation, Inc.
34195 Copying and distribution of this file, with or without modification,
34196 are permitted in any medium without royalty provided the copyright
34197 notice and this notice are preserved.