1 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * doc/sourcebuild.texi (vect_masked_store): Document.
7 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8 Alan Hayward <alan.hayward@arm.com>
9 David Sherwood <david.sherwood@arm.com>
11 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
13 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
14 Alan Hayward <alan.hayward@arm.com>
15 David Sherwood <david.sherwood@arm.com>
17 * doc/sourcebuild.texi (vect_variable_length): Document.
19 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
20 Alan Hayward <alan.hayward@arm.com>
21 David Sherwood <david.sherwood@arm.com>
23 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
25 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
26 Alan Hayward <alan.hayward@arm.com>
27 David Sherwood <david.sherwood@arm.com>
29 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
31 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
32 Alan Hayward <alan.hayward@arm.com>
33 David Sherwood <david.sherwood@arm.com>
35 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
36 previously undocumented selectors.
37 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
39 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
41 * doc/rtl.texi (const_vector): Say that elements can be
43 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
44 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
45 (gen_vec_duplicate): Use it instead of CONSTANT_P.
46 * optabs.c (expand_vector_broadcast): Likewise.
48 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
49 Alan Hayward <alan.hayward@arm.com>
50 David Sherwood <david.sherwood@arm.com>
52 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
53 scaled index even if the unscaled address was invalid.
54 Don't increase the complexity of using a scale in that case.
56 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
57 Alan Hayward <alan.hayward@arm.com>
58 David Sherwood <david.sherwood@arm.com>
60 * doc/rtl.texi: Rewrite the subreg rules so that they partition
61 the inner register into REGMODE_NATURAL_SIZE bytes rather than
63 * emit-rtl.c (validate_subreg): Divide subregs into blocks
64 based on REGMODE_NATURAL_SIZE of the inner mode.
65 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
66 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
67 * expmed.c (lowpart_bit_field_p): Divide the value up into
68 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
69 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
70 whether something is likely to occupy more than one register.
72 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
75 * ipa-inline-transform.c (update_noncloned_frequencies): Use
76 profile_count::adjust_for_ipa_scaling.
77 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
78 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
80 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
82 2017-11-09 Jakub Jelinek <jakub@redhat.com>
84 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
85 (split_group): Add total_orig and total_new arguments, estimate the
86 number of statements related to the store group without store merging
87 and with store merging.
88 (imm_store_chain_info::output_merged_store): Adjust split_group
89 callers, punt if estimated number of statements with store merging
90 is not smaller than estimated number of statements without it.
92 (handled_load): Remove has_single_use checks.
93 (pass_store_merging::process_store): Likewise.
95 2017-11-09 Richard Biener <rguenther@suse.de>
97 PR tree-optimization/82902
98 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
100 2017-11-09 Martin Liska <mliska@suse.cz>
103 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
106 2017-11-09 Martin Liska <mliska@suse.cz>
108 PR tree-optimization/82669
109 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
111 2017-11-09 Martin Liska <mliska@suse.cz>
113 PR gcov-profile/48463
114 * coverage.c (coverage_begin_function): Output also end locus
115 of a function and information whether the function is
117 * gcov-dump.c (tag_function): Parse and print the information.
118 * gcov.c (INCLUDE_MAP): Add include.
119 (INCLUDE_SET): Likewise.
120 (struct line_info): Move earlier in the source file because
121 of vector<line_info> in function_info structure.
122 (line_info::line_info): Likewise.
123 (line_info::has_block): Likewise.
124 (struct source_info): Add new member index.
125 (source_info::get_functions_at_location): New function.
126 (function_info::group_line_p): New function.
127 (output_intermediate_line): New function.
128 (output_intermediate_file): Use the mentioned function.
129 (struct function_start): New.
130 (struct function_start_pair_hash): Likewise.
131 (process_file): Add code that identifies group functions.
132 Assign lines either to global or function scope.
133 (generate_results): Skip artificial functions.
134 (find_source): Assign index for each source file.
135 (read_graph_file): Read new flag artificial and end_line.
136 (add_line_counts): Assign it either to global of function scope.
137 (accumulate_line_counts): Isolate core of the function to
138 accumulate_line_info and call it for both function and global
140 (accumulate_line_info): New function.
141 (output_line_beginning): Fix GNU coding style.
142 (print_source_line): New function.
143 (output_line_details): Likewise.
144 (output_function_details): Likewise.
145 (output_lines): Iterate both source (global) scope and function
147 (struct function_line_start_cmp): New class.
148 * doc/gcov.texi: Reflect changes in documentation.
150 2017-11-09 Jakub Jelinek <jakub@redhat.com>
153 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
154 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
155 and similarly for not instead of neg.
157 2017-11-08 Andi Kleen <ak@linux.intel.com>
159 * config/i386/i386.opt: Add -mforce-indirect-call.
160 * config/i386/predicates.md: Check for flag_force_indirect_call.
161 * doc/invoke.texi: Document -mforce-indirect-call
163 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
165 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
167 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
168 config/riscv/riscv.c (predict.h): New include.
169 (riscv_slow_unaligned_access_p): No longer static.
170 (riscv_block_move_straight): Add require.
171 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
173 2017-11-08 Jakub Jelinek <jakub@redhat.com>
176 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
177 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
178 nonimmediate_operand predicate for operand 1 instead of
181 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
183 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
185 * config/aarch64/constraints.md (Uml): New constraint.
186 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
189 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
191 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
192 of two vec_duplicates into a vec_concat.
194 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
196 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
197 Simplify vec_merge of vec_duplicate and vec_concat.
198 * config/aarch64/constraints.md (Utq): New constraint.
199 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
202 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
204 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
205 Simplify vec_merge of vec_duplicate and const_vector.
206 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
208 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
209 mode iterator. Update predicate on operand 1 to
210 handle non-const_vec constants. Delete constraints.
211 (*aarch64_combinez_be<mode>): Likewise for operand 2.
213 2017-11-08 Jakub Jelinek <jakub@redhat.com>
215 PR tree-optimization/78821
216 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
218 (store_operand_info::store_operand_info): Initialize it to false.
219 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
220 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
221 store in the group, and if chain_info is non-NULL, to ignore altogether
223 (compatible_load_p): Fail if bit_not_p does not match.
224 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
225 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
226 (pass_store_merging::process_store): Adjust
227 terminate_all_aliasing_chains calls to pass NULL in all current spots,
228 call terminate_all_aliasing_chains newly when adding a store into
229 a chain with non-NULL chain_info.
231 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
233 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
235 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
237 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
239 (aarch64_layout_frame): Initialise emit_frame_chain.
240 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
241 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
243 2017-11-08 Martin Liska <mliska@suse.cz>
245 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
246 of gimple_call_internal_p.
248 2017-11-07 Tom de Vries <tom@codesourcery.com>
250 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
252 2017-11-07 Tom de Vries <tom@codesourcery.com>
254 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
257 2017-11-08 Martin Liska <mliska@suse.cz>
260 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
262 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
264 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
266 2017-11-07 Jakub Jelinek <jakub@redhat.com>
269 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
270 (*cmp<mode>_ccz_1): New insn with $k alternative.
273 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
274 RTX_COMM_COMPARE as commutative as well.
275 (ix86_binary_operator_ok): Formatting fix.
276 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
277 *<code><mode>3<mask_name><round_saeonly_name>,
278 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
279 *<s>mul<mode>3_highpart<mask_name>,
280 *vec_widen_umult_even_v16si<mask_name>,
281 *vec_widen_umult_even_v8si<mask_name>,
282 *vec_widen_umult_even_v4si<mask_name>,
283 *vec_widen_smult_even_v16si<mask_name>,
284 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
285 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
286 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
287 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
288 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
289 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
290 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
291 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
292 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
293 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
294 ix86_binary_operator_ok. Formatting fixes.
295 (*<plusminus_insn><mode>3<mask_name><round_name>,
296 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
299 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
301 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
302 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
303 eq and ne if TARGET_ISEL.
304 (cmp): New code_iterator.
305 (UNS, UNSU_, UNSIK): New code_attrs.
306 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
307 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
309 ("eq<mode>3"): ... this.
310 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
312 ("ne<mode>3"): ... this.
314 2017-11-07 Julia Koval <julia.koval@intel.com>
317 * common/config/i386/i386-common.c
318 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
319 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
320 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
321 * config/i386/i386.opt: Ditto.
322 * config/i386/i386.c (ix86_target_string): Ditto.
323 (ix86_option_override_internal): Ditto.
324 (ix86_init_mpx_builtins): Move MPX to args2.
325 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
326 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
327 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
328 __builtin_ia32_vgf2p8affineinvqb_v32qi,
329 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
330 __builtin_ia32_vgf2p8affineinvqb_v16qi,
331 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
333 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
336 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
337 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
338 (zero-extendsidi peephole2): Remove peephole.
340 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
343 * stor-layout.c: Include attribs.h.
344 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
345 explicit lookup of "aligned" attribute.
347 2017-11-07 Andrew Waterman <andrew@sifive.com>
349 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
350 (riscv_expand_block_move): Likewise.
351 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
353 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
354 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
355 * config/riscv/riscv.c (riscv_block_move_straight): New function.
356 (riscv_adjust_block_mem): Likewise.
357 (riscv_block_move_loop): Likewise.
358 (riscv_expand_block_move): Likewise.
359 gcc/config/riscv/riscv.md (movmemsi): New pattern.
361 2017-11-07 Michael Clark <michaeljclark@mac.com>
363 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
364 (MUSL_DYNAMIC_LINKER): Likewise.
366 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
368 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
369 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
372 2017-11-07 Richard Biener <rguenther@suse.de>
374 * match.pd: Fix build.
376 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
377 Jackson Woodruff <jackson.woodruff@arm.com>
379 PR tree-optimization/71026
380 * match.pd: Canonicalize negate in division.
382 2017-11-07 Sudakshina Das <sudi.das@arm.com>
385 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
387 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
389 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
390 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
392 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
394 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
395 non-scalar integral types.
396 * match.pd (negate_expr_p): Handle MINUS_EXPR.
397 (-(A-B), -(~A)): New transformations.
399 2017-11-07 Tom de Vries <tom@codesourcery.com>
401 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
402 semicolon after "do {} while (0)".
403 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
404 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
405 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
406 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
407 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
409 2017-11-07 Tom de Vries <tom@codesourcery.com>
411 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
412 after "do {} while (0)".
413 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
414 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
415 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
416 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
417 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
419 2017-11-07 Tom de Vries <tom@codesourcery.com>
422 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
424 (arm_rtx_costs_internal): Add missing semicolon after
425 HANDLE_NARROW_SHIFT_ARITH call.
427 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
429 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
430 disable isel if it was not set explicitly.
432 2017-11-06 James Bowman <james.bowman@ftdichip.com>
434 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
436 (add_type_attribute) likewise.
438 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
440 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
441 of struct ix86_frame.
442 (ix86_initial_elimination_offset): Likewise.
443 (ix86_expand_split_stack_prologue): Likewise.
445 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
447 * tree-vrp.h (enum value_range_type): Update stale comment.
449 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
450 Alan Hayward <alan.hayward@arm.com>
451 David Sherwood <david.sherwood@arm.com>
453 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
454 (aarch64_expand_vec_perm_const): Take the number of units too.
455 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
456 (aarch64_expand_vec_perm_const): Likewise.
457 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
458 (vec_perm<mode>): Update accordingly.
460 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
461 Alan Hayward <alan.hayward@arm.com>
462 David Sherwood <david.sherwood@arm.com>
464 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
465 Take the number of units too.
466 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
467 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
468 but check for a vector mode before rather than after the call.
469 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
470 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
471 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
472 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
473 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
474 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
475 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
476 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
477 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
478 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
479 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
480 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
481 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
482 (aarch64_sqdmull2_n<mode>): Update accordingly.
484 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
485 Alan Hayward <alan.hayward@arm.com>
486 David Sherwood <david.sherwood@arm.com>
488 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
489 the number of units too.
490 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
491 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
492 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
493 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
494 (vec_store_lanesxi<mode>): Update accordingly.
496 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
497 Alan Hayward <alan.hayward@arm.com>
498 David Sherwood <david.sherwood@arm.com>
500 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
501 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
502 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
503 of units rather than the mode.
504 * config/aarch64/iterators.md (nunits): New mode attribute.
505 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
506 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
507 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
508 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
509 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
510 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
511 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
512 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
513 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
514 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
515 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
516 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
517 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
518 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
519 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
520 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
521 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
522 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
523 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
524 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
525 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
526 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
527 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
528 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
529 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
530 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
531 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
532 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
533 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
534 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
535 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
536 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
537 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
538 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
539 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
540 (aarch64_simd_vec_setv2di): Likewise.
542 2017-11-06 Carl Love <cel@us.ibm.com>
544 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
546 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
547 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
549 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
551 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
552 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
553 Add power 8 macro expansions.
554 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
555 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
556 power 8 instructions. (VSX_XXBR): Add iterator.
558 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
560 * config/arm/arm.md (predicable_short_it): Change default to "no",
561 improve documentation, remove uses that are identical to the default.
562 (enabled_for_depr_it): Rename to enabled_for_short_it.
563 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
564 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
565 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
566 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
567 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
569 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
572 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
573 float128 helper macros, which are no longer used after deleting
574 the old 'q' built-in functions, and moving the round to odd
575 built-in functions to being special built-in functions.
576 (BU_FLOAT128_2): Likewise.
577 (BU_FLOAT128_1_HW): Likewise.
578 (BU_FLOAT128_2_HW): Likewise.
579 (BU_FLOAT128_3_HW): Likewise.
580 (FABSQ): Delete old 'q' built-in functions.
581 (COPYSIGNQ): Likewise.
582 (SQRTF128_ODD): Move round to odd built-in functions to be
583 special built-in functions, so that we can handle
584 -mabi=ieeelongdouble.
585 (TRUNCF128_ODD): Likewise.
586 (ADDF128_ODD): Likewise.
587 (SUBF128_ODD): Likewise.
588 (MULF128_ODD): Likewise.
589 (DIVF128_ODD): Likewise.
590 (FMAF128_ODD): Likewise.
591 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
592 built-in names to 'f128'.
593 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
594 old 'q' built-in functions, as the machine independent code for
595 'f128' built-in functions handles this.
596 (rs6000_expand_builtin): Add expansion for float128 round to odd
597 functions, keying off on -mabi=ieeelongdouble of whether to use
598 the KFmode or TFmode variant.
599 (rs6000_init_builtins): Initialize the _Float128 round to odd
601 * doc/extend.texi (PowerPC Built-in Functions): Document the old
602 _Float128 'q' built-in functions are now mapped into the new
603 'f128' built-in functions.
605 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
607 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
608 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
610 2017-11-06 David Malcolm <dmalcolm@redhat.com>
613 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
614 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
615 * toplev.c: Include "ipa-fnsummary.h".
616 (toplev::finalize): Call ipa_fnsummary_c_finalize.
618 2017-11-06 Jakub Jelinek <jakub@redhat.com>
620 PR tree-optimization/82838
621 * gimple-ssa-store-merging.c
622 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
623 on a separate gimple_seq which is then appended to seq.
625 2017-11-06 Jeff Law <law@redhat.com>
628 * config/i386/i386.c (PROBE_INTERVAL): Remove.
629 (get_probe_interval): New functions.
630 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
631 (ix86_adjust_stack_and_probe): Likewise.
632 (output_adjust_stack_and_probe): Likewise.
633 (ix86_emit_probe_stack_range): Likewise.
634 (ix86_expand_prologue): Likewise.
636 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
638 PR tree-optimization/82816
639 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
640 if the modes of the two types are the same.
641 (convert_plusminus_to_widen): Likewise.
643 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
645 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
647 (usadv16qi): New define_expand.
648 (usadv8hi): New define_expand.
650 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
653 * ipa-inline-transform.c (update_noncloned_frequencies): Always
655 (inline_transform): Likewise.
656 * predict.c (counts_to_freqs): Remove useless conditional.
657 * profile-count.h (profile_count::apply_scale): Move sanity check.
658 * tree-inline.c (copy_bb): Always scale.
659 (copy_cfg_body): Likewise.
661 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
664 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
666 (*cmp_ite0): Add enabled_for_depr_it attribute.
667 (*cmp_ite1): Likewise.
669 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
671 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
674 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
676 * tree-vrp.c (vrp_int_const_binop): Return true on success and
677 return the value by pointer.
678 (extract_range_from_multiplicative_op_1): Update accordingly.
679 Return as soon as an operation fails.
681 2017-11-05 Tom de Vries <tom@codesourcery.com>
684 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
685 (DEF_SANITIZER_BUILTIN): ... here.
686 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
687 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
689 2017-11-05 Tom de Vries <tom@codesourcery.com>
692 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
694 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
695 ASM_OUTPUT_BEFORE_CASE_LABEL call.
696 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
698 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
699 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
700 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
702 2017-11-05 Tom de Vries <tom@codesourcery.com>
705 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
708 2017-11-04 Michael Clark <michaeljclark@mac.com>
710 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
711 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
713 (*addsi3_extended): Likewise.
714 (*addsi3_extended2): Likewise.
715 (<optab>si3): Likewise.
716 (<optab>di3): Likewise.
717 (<optab><mode>3): Likewise.
718 (<*optabe>si3_internal): Likewise.
719 (zero_extendqi<SUPERQI:mode>2): Likewise.
720 (*add<mode>hi3): Likewise.
721 (*xor<mode>hi3): Likewise.
722 (<optab>di3): Likewise.
723 (*<optab>si3_extend): Likewise.
724 (*sge<u>_<X:mode><GPR:mode>): Likewise.
725 (*slt<u>_<X:mode><GPR:mode>): Likewise.
726 (*sle<u>_<X:mode><GPR:mode>): Likewise.
728 2017-11-04 Andrew Waterman <andrew@sifive.com>
730 * config/riscv/riscv.c (riscv_option_override): Conditionally set
731 TARGET_STRICT_ALIGN based upon -mtune argument.
733 2017-11-04 Andrew Waterman <andrew@sifive.com>
735 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
737 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
739 config/i386/i386.c (choose_basereg): Use optional scratch
740 register and add assertion.
741 (x86_emit_outlined_ms2sysv_save): Use scratch register when
742 needed, and don't allocate stack.
743 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
744 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
745 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
747 2017-11-03 Jeff Law <law@redhat.com>
749 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
750 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
751 to probe at the start of a noreturn function.
753 2017-11-03 Jakub Jelinek <jakub@redhat.com>
755 PR tree-optimization/78821
756 * gimple-ssa-store-merging.c: Update the file comment.
757 (MAX_STORE_ALIAS_CHECKS): Define.
758 (struct store_operand_info): New type.
759 (store_operand_info::store_operand_info): New constructor.
760 (struct store_immediate_info): Add rhs_code and ops data members.
761 (store_immediate_info::store_immediate_info): Add rhscode, op0r
762 and op1r arguments to the ctor, initialize corresponding data members.
763 (struct merged_store_group): Add load_align_base and load_align
765 (merged_store_group::merged_store_group): Initialize them.
766 (merged_store_group::do_merge): Update them.
767 (merged_store_group::apply_stores): Pick the constant for
768 encode_tree_to_bitpos from one of the two operands, or skip
769 encode_tree_to_bitpos if neither operand is a constant.
770 (class pass_store_merging): Add process_store method decl. Remove
771 bool argument from terminate_all_aliasing_chains method decl.
772 (pass_store_merging::terminate_all_aliasing_chains): Remove
773 var_offset_p argument and corresponding handling.
774 (stmts_may_clobber_ref_p): New function.
775 (compatible_load_p): New function.
776 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
777 if there is overlap and rhs_code is not INTEGER_CST. For
778 non-overlapping stores terminate group if rhs is not mergeable.
779 (get_alias_type_for_stmts): Change first argument from
780 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
781 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
782 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
784 (get_location_for_stmts): Change first argument from
785 auto_vec<gimple *> & to vec<gimple *> &.
786 (struct split_store): Remove orig_stmts data member, add orig_stores.
787 (split_store::split_store): Create orig_stores rather than orig_stmts.
788 (find_constituent_stmts): Renamed to ...
789 (find_constituent_stores): ... this. Change second argument from
790 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
791 to info structures rather than the statements.
792 (split_group): Rename ALLOW_UNALIGNED argument to
793 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
794 it. Adjust find_constituent_stores caller.
795 (imm_store_chain_info::output_merged_store): Handle rhs_code other
796 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
797 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
798 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
799 (mem_valid_for_store_merging): New function.
800 (handled_load): New function.
801 (pass_store_merging::process_store): New method.
802 (pass_store_merging::execute): Use process_store method. Adjust
803 terminate_all_aliasing_chains caller.
805 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
807 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
808 Return true for more constants, symbols and label references.
809 (aarch64_valid_floating_const): Remove unused function.
811 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
814 * config/aarch64/aarch64.c (aarch64_layout_frame):
815 Undo forcing of LR at bottom of frame.
817 2017-11-03 Jeff Law <law@redhat.com>
820 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
821 for int_registers_saved.
823 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
824 extracted from tree-ssa-dom.c.
825 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
826 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
827 (record_equivalences_from_incoming_edge): Add additional argument
828 to single_pred_edge_ignoring_loop_edges call.
829 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
830 (uncprop_dom_walker::before_dom_children): Add additional argument
831 to single_pred_edge_ignoring_loop_edges call.
832 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
833 single_pred_edge_ignoring_loop_edges rather than open coding.
834 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
836 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
838 * match.pd (-(-A)): Rewrite.
840 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
842 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
843 (rs6000_emit_int_cmove): New declaration.
844 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
845 (rs6000_emit_sISEL): Delete.
846 (rs6000_emit_int_cmove): Make non-static.
847 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
848 instead of rs6000_emit_sISEL.
850 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
852 * asan.c (create_cond_insert_point): Maintain profile.
853 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
855 * basic-block.h (struct basic_block_def): Remove frequency.
856 (EDGE_FREQUENCY): Use to_frequency
857 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
859 (find_traces): Update to use to_frequency.
860 (find_traces_1_round): Likewise; use only IPA counts.
861 (bb_to_key): Likewise.
862 (connect_traces): Use IPA counts only.
863 (copy_bb_p): Update to use to_frequency.
864 (fix_up_crossing_landing_pad): Likewise.
865 (sanitize_hot_paths): Likewise.
866 * bt-load.c (basic_block_freq): Likewise.
867 * cfg.c (init_flow): Set count_max to uninitialized.
868 (check_bb_profile): Remove frequencies; check counts.
869 (dump_bb_info): Do not dump frequencies.
870 (update_bb_profile_for_threading): Update counts only.
871 (scale_bbs_frequencies_int): Likewise.
872 (MAX_SAFE_MULTIPLIER): Remove.
873 (scale_bbs_frequencies_gcov_type): Update counts only.
874 (scale_bbs_frequencies_profile_count): Update counts only.
875 (scale_bbs_frequencies): Update counts only.
876 * cfg.h (struct control_flow_graph): Add count-max.
877 (update_bb_profile_for_threading): Update prototype.
878 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
879 (find_many_sub_basic_blocks): Likewise.
880 * cfgcleanup.c (try_forward_edges): Likewise.
881 (try_crossjump_to_edge): Likewise.
882 * cfgexpand.c (expand_gimple_cond): Likewise.
883 (expand_gimple_tailcall): Likewise.
884 (construct_init_block): Likewise.
885 (construct_exit_block): Likewise.
886 * cfghooks.c (verify_flow_info): Check consistency of counts.
887 (dump_bb_for_graph): Do not dump frequencies.
888 (split_block_1): Do not update frequencies.
889 (split_edge): Do not update frequencies.
890 (make_forwarder_block): Do not update frequencies.
891 (duplicate_block): Do not update frequencies.
892 (account_profile_record): Do not update frequencies.
893 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
894 for global heuristics.
895 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
896 (expected_loop_iterations_unbounded): Use counts only.
897 * cfgloopmanip.c (scale_loop_profile): Simplify.
898 (create_empty_loop_on_edge): Simplify
900 (duplicate_loop_to_header_edge): Simplify
901 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
902 (update_br_prob_note): Take care of removing note when profile
904 (relink_block_chain): Do not dump frequency.
905 (rtl_account_profile_record): Use to_frequency.
906 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
907 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
908 (cgraph_update_edges_for_call_stmt_node): Likewise.
909 (cgraph_edge::verify_count_and_frequency): Update.
910 (cgraph_node::verify_node): Temporarily disable frequency verification.
911 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
913 (cgraph_edge::rebuild_edges): Convert to ipa counts.
914 * cgraphunit.c (init_lowered_empty_function): Do not initialize
916 (cgraph_node::expand_thunk): Update profile.
917 * except.c (dw2_build_landing_pads): Do not update frequency.
918 * final.c (compute_alignments): Use to_frequency.
919 (dump_basic_block_info): Do not dump frequency.
920 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
921 (dump_gimple_bb_header): Do not dump frequency.
922 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
924 * gimple-streamer-in.c (input_bb): Do not stream frequency.
925 * gimple-streamer-out.c (output_bb): Do not stream frequency.
926 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
927 (init_before_recovery): Do not update frequency.
928 (sched_create_recovery_edges): Do not update frequency.
929 * hsa-gen.c (convert_switch_statements): Do not update frequency.
930 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
931 (ipa_cp_c_finalize): Set max_count to uninitialized.
932 * ipa-fnsummary.c (get_minimal_bb): Use counts.
933 (param_change_prob): Use counts.
934 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
936 * ipa-split.c (consider_split): Use to_frequency.
937 (split_function): Use to_frequency.
938 * ira-build.c (loop_compare_func): Likewise.
939 (mark_loops_for_removal): Likewise.
940 (mark_all_loops_for_removal): Likewise.
941 * loop-doloop.c (doloop_modify): Do not update frequency.
942 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
944 * lto-streamer-in.c (input_function): Update count_max.
945 * omp-expand.c (expand_omp_taskreg): Update count_max.
946 * omp-simd-clone.c (simd_clone_adjust): Update profile.
947 * predict.c (maybe_hot_frequency_p): Use to_frequency.
948 (maybe_hot_count_p): Use ipa counts only.
949 (maybe_hot_bb_p): Simplify.
950 (maybe_hot_edge_p): Simplify.
951 (probably_never_executed): Do not take frequency argument.
952 (probably_never_executed_bb_p): Do not pass frequency.
953 (probably_never_executed_edge_p): Likewise.
954 (combine_predictions_for_bb): Check that profile is nonzero.
955 (propagate_freq): Do not set frequency.
956 (drop_profile): Simplify.
957 (counts_to_freqs): Simplify.
958 (expensive_function_p): Use to_frequency.
959 (propagate_unlikely_bbs_forward): Simplify.
960 (determine_unlikely_bbs): Simplify.
961 (estimate_bb_frequencies): Add hack to silence graphite issues.
962 (compute_function_frequency): Use ipa counts.
963 (pass_profile::execute): Update.
964 (rebuild_frequencies): Use counts only.
965 (force_edge_cold): Use counts only.
966 * profile-count.c (profile_count::dump): Dump new count types.
967 (profile_count::differs_from_p): Check compatiblity.
968 (profile_count::to_frequency): New function.
969 (profile_count::to_cgraph_frequency): New function.
970 * profile-count.h (struct function): Declare.
971 (enum profile_quality): Add profile_guessed_local and
972 profile_guessed_global0.
973 (class profile_proability): Decrease number of bits to 29;
974 update from_reg_br_prob_note and to_reg_br_prob_note.
975 (class profile_count: Update comment; decrease number of bits
976 to 61. Check compatibility.
977 (profile_count::compatible_p): New private member function.
978 (profile_count::ipa_p): New member function.
979 (profile_count::operator<): Handle global zero correctly.
980 (profile_count::operator>): Handle global zero correctly.
981 (profile_count::operator<=): Handle global zero correctly.
982 (profile_count::operator>=): Handle global zero correctly.
983 (profile_count::nonzero_p): New member function.
984 (profile_count::force_nonzero): New member function.
985 (profile_count::max): New member function.
986 (profile_count::apply_scale): Handle IPA scalling.
987 (profile_count::guessed_local): New member function.
988 (profile_count::global0): New member function.
989 (profile_count::ipa): New member function.
990 (profile_count::to_frequency): Declare.
991 (profile_count::to_cgraph_frequency): Declare.
992 * profile.c (OVERLAP_BASE): Delete.
993 (compute_frequency_overlap): Delete.
994 (compute_branch_probabilities): Do not use compute_frequency_overlap.
995 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
996 * sched-ebb.c (rank): Use counts only.
997 * shrink-wrap.c (handle_simple_exit): Use counts only.
998 (try_shrink_wrapping): Use counts only.
999 (place_prologue_for_one_component): Use counts only.
1000 * tracer.c (find_best_predecessor): Use to_frequency.
1001 (find_trace): Use to_frequency.
1002 (tail_duplicate): Use to_frequency.
1003 * trans-mem.c (expand_transaction): Do not update frequency.
1004 * tree-call-cdce.c: Do not update frequency.
1005 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1006 (gimple_merge_blocks): Likewise.
1007 (gimple_split_edge): Likewise.
1008 (gimple_duplicate_sese_region): Likewise.
1009 (gimple_duplicate_sese_tail): Likewise.
1010 (move_sese_region_to_fn): Likewise.
1011 (gimple_account_profile_record): Likewise.
1012 (insert_cond_bb): Likewise.
1013 * tree-complex.c (expand_complex_div_wide): Likewise.
1014 * tree-eh.c (lower_resx): Update profile.
1015 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1017 (initialize_cfun): Do not initialize frequencies
1018 (freqs_to_counts): Delete.
1019 (copy_cfg_body): Ignore count parameter.
1020 (copy_body): Update.
1021 (expand_call_inline): Update count_max.
1022 (optimize_inline_calls): Update count_max.
1023 (tree_function_versioning): Update count_max.
1024 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1025 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1027 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1028 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1029 (try_peel_loop): Likewise.
1030 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1032 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1033 (tree_transform_and_unroll_loop): Do not use frequencies
1034 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1035 Use reliable prediction only.
1036 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1037 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1038 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1039 probability scaling.
1040 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1041 not update frequency
1042 (any_remaining_duplicated_blocks): Likewise.
1043 (update_profile): Likewise.
1044 (estimated_freqs_path): Delete.
1045 (freqs_to_counts_path): Delete.
1046 (clear_counts_path): Delete.
1047 (ssa_fix_duplicate_block_edges): Likewise.
1048 (duplicate_thread_path): Likewise.
1049 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1050 * tree-tailcall.c (decrease_profile): Do not update frequency.
1051 (eliminate_tail_call): Likewise.
1052 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1053 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1054 (optimize_mask_stores): Likewise.
1055 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1056 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1057 (ubsan_expand_ptr_ifn): Update profile.
1058 * value-prof.c (gimple_ic): Simplify.
1059 * value-prof.h (gimple_ic): Update prototype.
1060 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1061 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1063 (want_inline_self_recursive_call_p): Likewise.
1064 (resolve_noninline_speculation): Only cummulate defined counts.
1065 (inline_small_functions): Use nonzero_p.
1066 (ipa_inline): Do not access freed node.
1068 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1070 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1071 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1073 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1075 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1076 non-legitimate address.
1078 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1080 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1081 (*lt0_<mode>di, *lt0_<mode>si): New.
1083 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1085 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1086 TARGET_PAIRED_FLOAT.
1088 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1089 Jim Wilson <jim.wilson@linaro.org>
1091 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1092 * config/aarch64/aarch64-tune.md: Regenerated.
1093 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1094 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1096 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1098 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
1099 (arc_expand_prologue): Restore blink for millicode.
1100 * config/arc/linux.h (LINK_EH_SPEC): Defined.
1102 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
1105 * config/i386/i386.c (ix86_vector_duplicate_value): Use
1106 gen_vec_duplicate after forcing the scalar into a register.
1108 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
1110 * combine (try_combine): Print the insns input to try_combine to the
1113 2017-11-02 Steve Ellcey <sellcey@cavium.com>
1116 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
1117 Remove second argument from aarch64_process_target_attr call.
1118 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
1120 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
1122 (aarch64_handle_attr_arch): Remove second argument.
1123 (aarch64_handle_attr_cpu): Ditto.
1124 (aarch64_handle_attr_tune): Ditto.
1125 (aarch64_handle_attr_isa_flags): Ditto.
1126 (aarch64_process_one_target_attr): Ditto.
1127 (aarch64_process_target_attr): Ditto.
1128 (aarch64_option_valid_attribute_p): Remove second argument.
1129 on aarch64_process_target_attr call.
1131 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1133 * diagnostic.c: Include "selftest-diagnostic.h".
1134 (selftest::assert_location_text): New function.
1135 (selftest::test_diagnostic_get_location_text): New function.
1136 (selftest::diagnostic_c_tests): Call it.
1138 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1140 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
1141 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
1142 (class selftest::test_diagnostic_context): Move to...
1143 * selftest-diagnostic.c: New file.
1144 * selftest-diagnostic.h: New file.
1146 2017-11-02 James Bowman <james.bowman@ftdichip.com>
1148 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
1149 offset range for FT32B.
1150 * config/ft32/ft32.h: option "mcompress" enables relaxation.
1151 * config/ft32/ft32.md: Add TARGET_NOPM.
1152 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
1153 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
1155 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
1157 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
1159 2017-11-02 Jeff Law <law@redhat.com>
1161 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1162 virtual keyword on FINAL OVERRIDE members.
1164 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1165 virtuals together. Add virtual destructor.
1166 (substitute_and_fold_engine): Similarly.
1168 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1170 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1172 2017-11-02 Richard Biener <rguenther@suse.de>
1174 PR tree-optimization/82795
1175 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1177 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1179 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1181 (gcc_GAS_CHECK_FEATURE): Remove.
1182 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1183 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1184 * configure: Regenerate.
1186 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1188 * config/arc/arc.c (hwloop_optimize): Account for empty
1191 2017-11-02 Richard Biener <rguenther@suse.de>
1194 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1195 Truncate ARRAY_REF index and element size.
1197 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1199 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1201 2017-11-01 Jeff Law <law@redhat.com>
1203 * tree-ssa-ccp.c (ccp_folder): New class derived from
1204 substitute_and_fold_engine.
1205 (ccp_folder::get_value): New member function.
1206 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1207 (ccp_fold_stmt): Remove prototype.
1208 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1209 * tree-ssa-copy.c (copy_folder): New class derived from
1210 substitute_and_fold_engine.
1211 (copy_folder::get_value): Renamed from get_value.
1212 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1213 * tree-vrp.c (vrp_folder): New class derived from
1214 substitute_and_fold_engine.
1215 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1216 (vrp_folder::get_value): New member function.
1217 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1218 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1219 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1220 provide a class interface to folder/substitute routines.
1221 (ssa_prop_fold_stmt_fn): Remove typedef.
1222 (ssa_prop_get_value_fn): Likewise.
1223 (subsitute_and_fold): Remove prototype.
1224 (replace_uses_in): Likewise.
1225 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1226 Renamed from replace_uses_in. Call the virtual member function
1227 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1228 (substitute_and_fold_dom_walker): Remove initialization of
1229 data member entries for calbacks. Add substitute_and_fold_engine
1230 member and initialize it.
1231 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1232 member functions for get_value, replace_phi_args_in c
1233 replace_uses_in, and fold_stmt calls.
1234 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1235 substitute_and_fold. Remove assert. Update ctor call.
1237 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1238 (ssa_prop_visit_phi_fn): Likewise.
1239 (class ssa_propagation_engine): New class to provide an interface
1241 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1243 (ssa_prop_visit_phi): Likewise.
1244 (ssa_propagation_engine::simulate_stmt): Moved into class.
1245 Call visit_phi/visit_stmt from the class rather than via
1246 file scoped static variables.
1247 (ssa_propagation_engine::simulate_block): Moved into class.
1248 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1249 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1250 set file scoped statics for the visit_stmt/visit_phi callbacks.
1251 * tree-complex.c (complex_propagate): New class derived from
1252 ssa_propagation_engine.
1253 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1254 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1255 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1257 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1258 ssa_propagation_engine.
1259 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1260 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1261 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1262 * tree-ssa-copy.c (copy_prop): New class derived from
1263 ssa_propagation_engine.
1264 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1265 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1266 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1267 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1268 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1269 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1270 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1272 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1274 PR rtl-optimization/82778
1275 PR rtl-optimization/82597
1276 * compare-elim.c (struct comparison): Add in_a_setter field.
1277 (find_comparison_dom_walker::before_dom_children): Remove killed
1278 bitmap and df_simulate_find_defs call, instead walk the defs.
1279 Compute last_setter and initialize in_a_setter. Merge definitions
1280 with first initialization for a few variables.
1281 (try_validate_parallel): Use insn_invalid_p instead of
1282 recog_memoized. Return insn rather than just the pattern.
1283 (try_merge_compare): Fix up comment. Don't uselessly test if
1284 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1286 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1289 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1290 Alan Hayward <alan.hayward@arm.com>
1291 David Sherwood <david.sherwood@arm.com>
1293 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1294 aarch64_hard_regno_nregs to get the number of registers
1297 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1298 Alan Hayward <alan.hayward@arm.com>
1299 David Sherwood <david.sherwood@arm.com>
1301 * config/aarch64/constraints.md (Upl): Rename to...
1303 * config/aarch64/aarch64.md
1304 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1307 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1308 Alan Hayward <alan.hayward@arm.com>
1309 David Sherwood <david.sherwood@arm.com>
1311 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1312 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1315 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1316 Alan Hayward <alan.hayward@arm.com>
1317 David Sherwood <david.sherwood@arm.com>
1319 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1320 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1321 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1323 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1324 of a single element.
1325 * config/aarch64/iterators.md: Add a comment above the permute
1326 unspecs to say that they are generated directly by
1327 aarch64_expand_vec_perm_const.
1328 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1330 2017-11-01 Nathan Sidwell <nathan@acm.org>
1332 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1334 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1336 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1337 and medany code models, and describe what they do.
1339 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1341 Revert accidental duplicate:
1343 * combine.c (can_change_dest_mode): Reject changes in
1344 REGMODE_NATURAL_SIZE.
1346 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1348 PR rtl-optimization/64682
1349 PR rtl-optimization/69567
1350 PR rtl-optimization/69737
1351 PR rtl-optimization/82683
1352 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1353 register mentioned in the note, drop the note, unless it came from I3,
1354 in which case it should go to I3 again.
1356 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1358 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1359 and return false if not.
1360 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1362 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1364 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1365 range is known to be empty.
1367 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1368 Alan Hayward <alan.hayward@arm.com>
1369 David Sherwood <david.sherwood@arm.com>
1371 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1372 and CONST_VECTOR_NUNITS instead of computing the number of units from
1373 the byte sizes of the vector and element.
1374 (simplify_binary_operation_1): Likewise.
1375 (simplify_const_binary_operation): Likewise.
1376 (simplify_ternary_operation): Likewise.
1378 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1379 Alan Hayward <alan.hayward@arm.com>
1380 David Sherwood <david.sherwood@arm.com>
1382 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1383 (int_mem_offset): ...this new function.
1384 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1385 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1386 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1389 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1390 Alan Hayward <alan.hayward@arm.com>
1391 David Sherwood <david.sherwood@arm.com>
1393 * lower-subreg.c (interesting_mode_p): New function.
1394 (compute_costs, find_decomposable_subregs, decompose_register)
1395 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1396 (resolve_clobber, dump_choices): Use it.
1398 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1399 Alan Hayward <alan.hayward@arm.com>
1400 David Sherwood <david.sherwood@arm.com>
1402 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
1404 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1405 Alan Hayward <alan.hayward@arm.com>
1406 David Sherwood <david.sherwood@arm.com>
1408 * alias.c (find_base_value, find_base_term): Only process integer
1409 truncations. Check the precision rather than the size.
1411 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1412 Alan Hayward <alan.hayward@arm.com>
1413 David Sherwood <david.sherwood@arm.com>
1415 * machmode.h (is_narrower_int_mode): New function
1416 * optabs.c (expand_float, expand_fix): Use it.
1417 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1419 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1420 Alan Hayward <alan.hayward@arm.com>
1421 David Sherwood <david.sherwood@arm.com>
1423 * rtl.h (narrower_subreg_mode): New function.
1424 * ira-color.c (update_costs_from_allocno): Use it.
1426 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1427 Alan Hayward <alan.hayward@arm.com>
1428 David Sherwood <david.sherwood@arm.com>
1430 * optabs-query.h (convert_optab_p): New function, split out from...
1431 (convert_optab_handler): ...here.
1432 (widening_optab_handler): Delete.
1433 (find_widening_optab_handler): Remove permit_non_widening parameter.
1434 (find_widening_optab_handler_and_mode): Likewise. Provide an
1435 override that operates on mode class wrappers.
1436 * optabs-query.c (widening_optab_handler): Delete.
1437 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1438 parameter. Assert that the two modes are the same class and that
1439 the "from" mode is narrower than the "to" mode. Use
1440 convert_optab_handler instead of widening_optab_handler.
1441 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1442 instead of widening_optab_handler.
1443 * expr.c (expand_expr_real_2): Update calls to
1444 find_widening_optab_handler.
1445 * optabs.c (expand_widen_pattern_expr): Likewise.
1446 (expand_binop_directly): Take the insn_code as a parameter.
1447 (expand_binop): Only call find_widening_optab_handler for
1448 conversion optabs; use optab_handler otherwise. Update calls
1449 to find_widening_optab_handler and expand_binop_directly.
1450 Use convert_optab_handler instead of widening_optab_handler.
1451 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1452 find_widening_optab_handler and use scalar_mode rather than
1454 (convert_plusminus_to_widen): Likewise.
1456 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1457 Alan Hayward <alan.hayward@arm.com>
1458 David Sherwood <david.sherwood@arm.com>
1460 * machmode.h (fixed_size_mode): New class.
1461 * rtl.h (get_pool_mode): Return fixed_size_mode.
1462 * gengtype.c (main): Add fixed_size_mode.
1463 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1464 (get_raw_arg_mode): Likewise.
1465 * doc/tm.texi: Regenerate.
1466 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1467 * targhooks.c (default_get_reg_raw_mode): Likewise.
1468 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1469 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1470 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1471 (msp430_get_raw_result_mode): Likewise.
1472 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1473 * dbxout.c (dbxout_parms): Require fixed-size modes.
1474 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1475 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1476 * omp-low.c (lower_oacc_reductions): Likewise.
1477 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1478 (simplify_subreg): Update accordingly.
1479 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1480 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1481 that aren't fixed-size.
1482 (get_pool_mode): Return a fixed_size_mode.
1483 (output_constant_pool_2): Take a fixed_size_mode.
1485 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1486 Alan Hayward <alan.hayward@arm.com>
1487 David Sherwood <david.sherwood@arm.com>
1489 * doc/rtl.texi (vec_series): Document.
1490 (const): Say that the operand can be a vec_series.
1491 * rtl.def (VEC_SERIES): New rtx code.
1492 * rtl.h (const_vec_series_p_1): Declare.
1493 (const_vec_series_p): New function.
1494 * emit-rtl.h (gen_const_vec_series): Declare.
1495 (gen_vec_series): Likewise.
1496 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1497 (gen_vec_series): Likewise.
1498 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1499 * simplify-rtx.c (simplify_unary_operation): Handle negations
1501 (simplify_binary_operation_series): New function.
1502 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1503 (test_vector_ops_series): New function.
1504 (test_vector_ops): Call it.
1505 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1506 gen_const_vec_series.
1507 (altivec_lvsr): Likewise.
1508 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1510 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1511 Alan Hayward <alan.hayward@arm.com>
1512 David Sherwood <david.sherwood@arm.com>
1514 * doc/rtl.texi (const): Update description of address constants.
1515 Say that vector constants are allowed too.
1516 * common.md (E, F): Use CONSTANT_P instead of checking for
1518 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1519 checking for CONST_VECTOR.
1520 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1522 * expr.c (expand_expr_real_2): Check for vector modes instead
1523 of checking for CONST_VECTOR.
1524 * rtl.h (const_vec_p): New function.
1525 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1526 (unwrap_const_vec_duplicate): Handle them here too.
1528 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1529 David Malcolm <dmalcolm@redhat.com>
1530 Alan Hayward <alan.hayward@arm.com>
1531 David Sherwood <david.sherwood@arm.com>
1533 * rtl.h (vec_duplicate_p): New function.
1534 * selftest-rtl.c (assert_rtx_eq_at): New function.
1535 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1536 (assert_rtx_eq_at): Declare.
1537 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1538 * selftest-run-tests.c (selftest::run_tests): Call it.
1539 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1540 (simplify_unary_operation_1): Recursively handle vector duplicates.
1541 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1543 (simplify_subreg): Handle subregs of vector duplicates.
1544 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1545 (selftest::simplify_rtx_c_tests): New functions.
1547 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1548 Alan Hayward <alan.hayward@arm.com>
1549 David Sherwood <david.sherwood@arm.com>
1551 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1552 (gen_vec_duplicate): Likewise.
1553 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1555 (gen_const_vector): ...here.
1556 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1557 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1558 whose elements are all equal.
1559 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1560 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1561 (simplify_relational_operation): Likewise.
1562 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1564 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1565 (aarch64_expand_vector_init): Likewise.
1566 * config/arm/arm.c (neon_vdup_constant): Likewise.
1567 (neon_expand_vector_init): Likewise.
1568 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1569 (arm_block_set_unaligned_vect): Likewise.
1570 (arm_block_set_aligned_vect): Likewise.
1571 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1572 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1573 (expand_vec_perm_even_odd_pack): Likewise.
1574 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1575 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1576 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1577 gen_const_vec_duplicate.
1578 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1579 * config/mips/mips.c (mips_gen_const_int_vector): Use
1580 gen_const_vec_duplicate.
1581 (mips_expand_vector_init): Use CONST0_RTX.
1582 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1583 (define_split): Use gen_const_vec_duplicate.
1584 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1585 (define_split): Use gen_const_vec_duplicate.
1586 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1587 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1588 * config/spu/spu.c (spu_const): Likewise.
1590 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1591 Alan Hayward <alan.hayward@arm.com>
1592 David Sherwood <david.sherwood@arm.com>
1594 * combine.c (can_change_dest_mode): Reject changes in
1595 REGMODE_NATURAL_SIZE.
1597 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
1599 * configure.ac (--enable-libssp): New.
1600 (gcc_cv_libc_provides_ssp): Check for explicit setting before
1601 trying to determine target-specific default. Adjust indentation.
1602 * configure: Regenerated.
1603 * doc/install.texi (Configuration): Expand --disable-libssp
1606 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
1608 config/i386/i386.c (ix86_expand_epilogue): Correct stack
1611 2017-10-31 Martin Jambor <mjambor@suse.cz>
1614 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1616 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1618 * auto-profile.c (autofdo_source_profile::read): Use
1619 UNKNOWN_LOCATION rather than 0.
1620 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1621 (warning_at): ...this overload.
1622 (warning_at_rich_loc_n): Rename to...
1623 (warning_n): ...this overload.
1624 (error_at_rich_loc): Rename to...
1625 (error_at): ...this overload.
1626 (pedwarn_at_rich_loc): Rename to...
1627 (pedwarn): ...this overload.
1628 (permerror_at_rich_loc): Rename to...
1629 (permerror): ...this overload.
1630 (inform_at_rich_loc): Rename to...
1631 (inform): ...this overload.
1632 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1633 (diagnostic_n_impl_richloc): Rename to...
1634 (diagnostic_n_impl): ...this rich_location *-based decl.
1635 (inform_at_rich_loc): Rename to...
1636 (inform): ...this, and add an assertion.
1637 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1638 (warning_at_rich_loc): Rename to...
1639 (warning_at): ...this, and add an assertion.
1640 (warning_at_rich_loc_n): Rename to...
1641 (warning_n): ...this, and add an assertion.
1642 (warning_n): Update location_t-based implementation for removal of
1643 location_t-based diagnostic_n_impl.
1644 (pedwarn_at_rich_loc): Rename to...
1645 (pedwarn): ...this, and add an assertion.
1646 (permerror_at_rich_loc): Rename to...
1647 (permerror): ...this, and add an assertion.
1648 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1649 (error_at_rich_loc): Rename to...
1650 (error_at): ...this, and add an assertion.
1651 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1652 (driver::do_spec_on_infiles): Likewise.
1653 * substring-locations.c (format_warning_va): Update for renaming
1654 of inform_at_rich_loc.
1656 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1658 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1659 _Float<N>X built-in functions so that the variant without the
1660 "__builtin_" prefix is only enabled for the GNU C and Objective C
1661 languages when they are in non-strict ANSI/ISO mode.
1662 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1663 * target.def (floatn_builtin_p): Add a target hook to control
1664 whether _Float<N> and _Float<N>X built-in functions without the
1665 "__builtin_" prefix are enabled, and return true for C and
1666 Objective C in the default hook. Include langhooks.h in
1668 * targhooks.h (default_floatn_builtin_p): Likewise.
1669 * targhooks.c (default_floatn_builtin_p): Likewise.
1670 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1671 floatn_builtin_p target hook.
1672 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1674 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1675 Eric Botcazou <ebotcazou@adacore.com>
1677 PR rtl-optimization/81803
1678 * lra-constraints.c (curr_insn_transform): Also reload the whole
1679 register for a strict subreg no wider than a word if this is for
1680 a WORD_REGISTER_OPERATIONS target.
1682 2017-10-31 Jason Merrill <jason@redhat.com>
1684 * gdbinit.in: Skip over inlines from timevar.h.
1686 2017-10-31 Martin Liska <mliska@suse.cz>
1688 * doc/gcov.texi: Document new option.
1689 * gcov.c (print_usage): Likewise print it.
1690 (process_args): Support the argument.
1691 (format_count): New function.
1692 (format_gcov): Use the function.
1694 2017-10-31 Martin Liska <mliska@suse.cz>
1696 * gcov.c (struct name_map): do not use typedef.
1697 Define operator== and operator<.
1698 (name_search): Remove.
1699 (name_sort): Remove.
1700 (main): Do not allocate names.
1701 (process_file): Add vertical space.
1702 (generate_results): Use std::find.
1703 (release_structures): Do not release memory.
1704 (find_source): Use std::find.
1706 2017-10-31 Martin Liska <mliska@suse.cz>
1708 * gcov.c (struct line_info): Remove it's typedef.
1709 (line_info::line_info): Add proper ctor.
1710 (line_info::has_block): Do not use a typedef.
1711 (struct source_info): Do not use typedef.
1712 (circuit): Likewise.
1713 (get_cycles_count): Likewise.
1714 (output_intermediate_file): Iterate via vector iterator.
1715 (add_line_counts): Use std::vector methods.
1716 (accumulate_line_counts): Likewise.
1717 (output_lines): Likewise.
1719 2017-10-31 Martin Liska <mliska@suse.cz>
1721 * gcov.c (struct source_info): Remove typedef.
1722 (source_info::source_info): Add proper ctor.
1723 (accumulate_line_counts): Use struct, not it's typedef.
1724 (output_gcov_file): Likewise.
1725 (output_lines): Likewise.
1726 (main): Do not allocate an array.
1727 (output_intermediate_file): Use size of vector container.
1728 (process_file): Resize the vector.
1729 (generate_results): Do not preallocate, use newly added vector
1731 (release_structures): Do not release sources.
1732 (find_source): Use vector methods.
1733 (add_line_counts): Do not use typedef.
1735 2017-10-31 Martin Liska <mliska@suse.cz>
1737 * doc/gcov.texi: Document that.
1738 * gcov.c (add_line_counts): Mark lines with a non-executed
1740 (output_line_beginning): Handle such lines.
1741 (output_lines): Pass new argument.
1742 (output_intermediate_file): Print it in intermediate format.
1744 2017-10-31 Martin Liska <mliska@suse.cz>
1746 * color-macros.h: New file.
1747 * diagnostic-color.c: Factor out color related to macros to
1749 * doc/gcov.texi: Document -k option.
1750 * gcov.c (INCLUDE_STRING): Include string.h.
1751 (print_usage): Add -k option.
1752 (process_args): Parse it.
1753 (pad_count_string): New function.
1754 (output_line_beginning): Likewise.
1755 (DEFAULT_LINE_START): New macro.
1756 (output_lines): Support color output.
1758 2017-10-31 Martin Liska <mliska@suse.cz>
1760 PR gcov-profile/82633
1761 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1762 their interaction with GCOV infrastructure.
1763 * configure.ac: Add -fkeep-{inline,static}-functions to
1765 * configure: Regenerate.
1767 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1770 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1772 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1775 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1776 into a register if it does not fit into an immediate offset field.
1778 2017-10-31 Olivier Hainque <hainque@adacore.com>
1780 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1782 2017-10-31 Julia Koval <julia.koval@intel.com>
1784 * config.gcc: Add gfniintrin.h.
1785 * config/i386/gfniintrin.h: New.
1786 * config/i386/i386-builtin-types.def
1787 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1788 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1789 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1790 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1791 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1792 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1793 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1794 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1795 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1796 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1797 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1798 * config/i386/immintrin.h: Include gfniintrin.h.
1799 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1801 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1803 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1805 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1807 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1808 (arm_ashldi3_1bit): Remove pattern.
1809 (ashrdi3): Remove shift by 1 expansion.
1810 (arm_ashrdi3_1bit): Remove pattern.
1811 (lshrdi3): Remove shift by 1 expansion.
1812 (arm_lshrdi3_1bit): Remove pattern.
1813 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1814 cost of ashldi3 by 1.
1815 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1816 (<shift>di3_neon): Likewise.
1818 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1820 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1821 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1822 and (*aarch64_simd_mov<VQ:mode>).
1823 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1824 pattern alternative.
1825 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1826 attributes to match pattern alternative.
1828 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1830 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1831 * config/rs6000/emmintrin.h: New file.
1832 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1834 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1836 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1837 * (movdi_vfp_cortexa8): Remove pattern.
1839 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1841 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1843 (Specific, alpha*-dec-osf5.1): Remove.
1844 (Specific, mips-sgi-irix5): Remove.
1845 (Specific, mips-sgi-irix6): Remove.
1847 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1850 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1851 arguments to clear_bit_region_be.
1853 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1855 * gimplify.c: Include memmodel.h.
1857 2017-10-30 Martin Jambor <mjambor@suse.cz>
1859 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1860 condition whether loop should be executed at all.
1862 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1864 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1865 gimple folding of vec_madd() intrinsics.
1866 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1867 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1868 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1870 2017-10-30 Richard Biener <rguenther@suse.de>
1872 PR tree-optimization/82762
1874 2017-10-23 Richard Biener <rguenther@suse.de>
1876 PR tree-optimization/82129
1878 2017-08-01 Richard Biener <rguenther@suse.de>
1880 PR tree-optimization/81181
1881 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1882 (compute_antic): ... end of iteration here.
1884 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1886 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1887 -std=iso9899:2017 and -std=gnu17.
1888 * doc/standards.texi (C Language): Document C17 support.
1889 * doc/cpp.texi (Overview): Mention -std=c17.
1890 (Standard Predefined Macros): Document C11 and C17 values of
1891 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1892 * doc/extend.texi (Inline): Do not list individual options for
1893 standards newer than C99.
1894 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1896 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1899 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1901 * asan.c (asan_finish_file): Align asan globals array by shadow
1904 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1907 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1908 (struct store_immediate_info): Add bitregion_start and bitregion_end
1910 (store_immediate_info::store_immediate_info): Add brs and bre
1911 arguments and initialize bitregion_{start,end} from those.
1912 (struct merged_store_group): Add bitregion_start, bitregion_end,
1913 align_base and mask fields. Drop unnecessary struct keyword from
1914 struct store_immediate_info. Add do_merge method.
1915 (clear_bit_region_be): Use memset instead of loop storing zeros.
1916 (merged_store_group::do_merge): New method.
1917 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
1918 stores as long as the surrounding bitregions have no gaps.
1919 (merged_store_group::merge_overlapping): Use do_merge.
1920 (merged_store_group::apply_stores): Test that bitregion_{start,end}
1921 is byte aligned, rather than requiring that start and width are
1922 byte aligned. Drop unnecessary struct keyword from
1923 struct store_immediate_info. Allocate and populate also mask array.
1924 Make start of the arrays relative to bitregion_start rather than
1925 start and size them according to bitregion_{end,start} difference.
1926 (struct imm_store_chain_info): Drop unnecessary struct keyword from
1927 struct store_immediate_info.
1928 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
1929 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
1930 struct keyword from struct store_immediate_info.
1931 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
1932 between stores as long as the surrounding bitregions have no gaps.
1934 (struct split_store): Add orig non-static data member.
1935 (split_store::split_store): Initialize orig to false.
1936 (find_constituent_stmts): Return store_immediate_info *, non-NULL
1937 if there is exactly a single original stmt. Change stmts argument
1938 to pointer from reference, if NULL, don't push anything to it. Add
1939 first argument, use it to optimize skipping over orig stmts that
1940 are known to be before bitpos already. Simplify.
1941 (split_group): Return unsigned int count how many stores are or
1942 would be needed rather than a bool. Add allow_unaligned argument.
1943 Change split_stores argument from reference to pointer, if NULL,
1944 only do a dry run computing how many stores would be produced.
1945 Rewritten algorithm to use both alignment and misalign if
1946 !allow_unaligned and handle bitfield stores with gaps.
1947 (imm_store_chain_info::output_merged_store): Set start_byte_pos
1948 from bitregion_start instead of start. Compute allow_unaligned
1949 here, if true, do 2 split_group dry runs to compute which one
1950 produces fewer stores and prefer aligned if equal. Punt if
1951 new count is bigger or equal than original before emitting any
1952 statements, rather than during that. Remove no longer needed
1953 new_ssa_names tracking. Replace num_stmts with
1954 split_stores.length (). Use 32-bit stack allocated entries
1955 in split_stores auto_vec. Try to reuse original store lhs/rhs1
1956 if possible. Handle bitfields with gaps.
1957 (pass_store_merging::execute): Ignore bitsize == 0 stores.
1958 Compute bitregion_{start,end} for the stores and construct
1959 store_immediate_info with that. Formatting fixes.
1961 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
1964 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
1965 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
1967 2017-10-29 Jim Wilson <wilson@tuliptree.org>
1969 * gimplify.c: Include tm_p.h.
1971 * common.opt (gcoff): Re-add as ignored option.
1972 (gcoff1, gcoff2, gcoff3): Likewise.
1974 * Makefile.in (OBJS): Delete sdbout.o.
1975 (GTFILES): Delete $(srcdir)/sdbout.c.
1976 * debug.h: Delete sdb_debug_hooks.
1977 * final.c: Delete sdbout.h include.
1978 (final_scan_insn): Delete SDB_DEBUG check.
1979 (rest_of_clean_state): Likewise.
1980 * output.h: Delete sdb_begin_function_line.
1983 * toplev.c: Delete sdbout.h include.
1984 (process_options): Delete SDB_DEBUG check.
1985 * tree-core.h (tree_type_common): Delete pointer field of
1987 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
1988 TYPE_SYMTAB_POINTER.
1989 * tree.h (TYPE_SYMTAB_POINTER): Delete.
1990 (TYPE_SYMTAB_IS_POINTER): Delete.
1991 (TYPE_SYMTAB_IS_DIE): Renumber.
1992 * xcoffout.c: Refer to former sdbout.c file.
1993 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
1995 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
1996 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
1997 (Debugging Options): Delete -gcoff.
1998 (-gstabs): Delete SDB reference.
2000 (-gcoff@var{level}): Delete.
2001 * doc/passes.texi (Debugging information output): Delete SDB and
2002 sdbout.c references.
2003 * doc/tm.texi: Regenerate.
2004 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2005 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2007 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2008 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2009 (SDB_DEBUGGING_INFO): Delete.
2010 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2011 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2012 * target.def (output_source_filename): Delete COFF reference.
2014 * common.opt (gcoff): Delete.
2015 (gxcoff+): Update Negative chain.
2016 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2018 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2019 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2020 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2022 (expand_function_start): Change sdb reference to past tense.
2023 (expand_function_end): Change sdb reference to past tense.
2024 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2025 * opts.c (debug_type_names): Delete coff entry.
2026 (common_handle_option): Delete OPT_gcoff case.
2027 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2029 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2030 * config/cris/cris.h: Delete SDB reference in comment.
2031 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2032 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2033 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2034 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2036 (ix86_expand_prologue): Likewise.
2037 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2038 * config/ia64/ia64.h: Likewise.
2039 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2040 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2042 * config/mmix/mmix.h: Likewise.
2043 * config/nds32/nds32.c: Likewise.
2044 * config/stormy/storym16.h: Likewise.
2045 * config/visium/visium.h: Likewise.
2046 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2048 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2050 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2051 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2052 FRAME_POINTER_REGNUM point at high end of local var area.
2054 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2056 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2057 Move comment around. Do not reset best_edge for a copiable
2058 destination if the copy would cause a partition change.
2059 (better_edge_p): Remove redundant check.
2061 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2063 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2065 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2067 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2068 for math functions that have _Float<N> and _Float<N>X variants.
2069 (mathfn_built_in_2): Add support for math functions that have
2070 _Float<N> and _Float<N>X variants.
2071 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2072 (expand_builtin_mathfn_ternary): Add support for fma with
2073 _Float<N> and _Float<N>X variants.
2074 (expand_builtin): Likewise.
2075 (fold_builtin_3): Likewise.
2076 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2077 create math function _Float<N> and _Float<N>X variants as external
2079 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2080 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2081 the __builtin_ prefix and if not strict ansi, without the prefix.
2082 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2083 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2084 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2085 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2086 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2087 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2088 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2089 function signatures for fma _Float<N> and _Float<N>X variants.
2090 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2091 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2092 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2093 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2094 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2095 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2096 * gencfn-macros.c (print_case_cfn): Add support for math functions
2097 that have _Float<N> and _Float<N>X variants.
2098 (print_define_operator_list): Likewise.
2099 (fltfn_suffixes): Likewise.
2101 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
2102 for math functions that have _Float<N> and _Float<N>X variants.
2103 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
2104 and _Float<N>X variants.
2105 (COPYSIGN): Likewise.
2108 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
2109 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
2111 (integer_valued_read_call_p): Likewise.
2112 * fold-const-call.c (fold_const_call_ss): Likewise.
2113 (fold_const_call_sss): Add support for copysign, fmin, and fmax
2114 _Float<N> and _Float<N>X variants.
2115 (fold_const_call_ssss): Add support for fma _Float<N> and
2116 _Float<N>X variants.
2117 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
2118 support for copysign and fma _Float<N> and _Float<N>X variants.
2119 (backprop::process_builtin_call_use): Likewise.
2120 * tree-call-cdce.c (can_test_argument_range); Add support for
2121 sqrt _Float<N> and _Float<N>X variants.
2122 (edom_only_function): Likewise.
2123 (get_no_error_domain): Likewise.
2124 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
2125 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
2126 copysign _Float<N> and _Float<N>X variants.
2127 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
2128 handled by machine independent code.
2129 (FMAF128): Likewise.
2130 * doc/cpp.texi (Common Predefined Macros): Document defining
2131 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
2132 fma _Float<N> and _Float<N>X variants.
2134 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2137 * config/i386/i386-modes.def (CCFPU): Remove definition.
2138 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
2139 (ix86_cc_modes_compatible): Ditto.
2140 (ix86_expand_carry_flag_compare): Ditto.
2141 (ix86_expand_int_movcc): Ditto.
2142 (ix86_expand_int_addcc): Ditto.
2143 (ix86_reverse_condition): Ditto.
2144 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
2145 Return true/false for unordered/ordered fp comparisons.
2146 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
2147 (ix86_prepare_fp_compare_args): Update for rename.
2148 (ix86_expand_fp_compare): Update for rename. Generate unordered
2149 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
2150 (ix86_expand_sse_compare_and_jump): Ditto.
2151 * config/i386/predicates.md (fcmov_comparison_operator):
2152 Remove CCFPU mode handling.
2153 (ix86_comparison_operator): Ditto.
2154 (ix86_carry_flag_operator): Ditto.
2155 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
2156 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
2157 (*cmpu<mode>_cc_i387): Ditto.
2158 (FPCMP): Remove mode iterator.
2159 (unord): Remove mode attribute.
2160 (unord_subst): New define_subst transformation
2161 (unord): New define_subst attribute.
2163 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2164 (*cmpi<unord>xf_i387): Ditto.
2165 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2166 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2167 using unord_subst transformation.
2168 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2169 (round_saeonly): Also handle CCFP mode.
2170 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2171 Remove UNSPEC_SAHF unspec handling.
2173 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2175 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2177 2017-10-27 Jeff Law <law@redhat.com>
2179 * gimple-ssa-sprintf.c: Include domwalk.h.
2180 (class sprintf_dom_walker): New class, derived from dom_walker.
2181 (sprintf_dom_walker::before_dom_children): New function.
2182 (struct call_info): Moved into sprintf_dom_walker class
2183 (compute_formath_length, handle_gimple_call): Likewise.
2184 (sprintf_length::execute): Call the dominator walker rather
2185 than walking the statements.
2187 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2188 gimple statement locations.
2189 (check_array_bounds): Corresponding changes. Get the statement's
2190 location directly from wi->stmt.
2192 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2195 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2197 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2199 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2200 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2202 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2205 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2206 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2207 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2208 maybe_get_pool_constant.
2209 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2212 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2214 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2215 2.26 caveat. Update gas and gld versions.
2216 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2219 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2221 * cgraph.h (set_malloc_flag): Declare.
2222 * cgraph.c (set_malloc_flag_1): New function.
2223 (set_malloc_flag): Likewise.
2224 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2225 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2227 (read_ipa_call_summary): Add support for reading is_return_callee.
2228 (write_ipa_call_summary): Stream is_return_callee.
2229 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2230 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2231 ipa-prop.h, ipa-fnsummary.h.
2232 (pure_const_names): Change to static.
2233 (malloc_state_e): Define.
2234 (malloc_state_names): Define.
2235 (funct_state_d): Add field malloc_state.
2236 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2237 (check_retval_uses): New function.
2238 (malloc_candidate_p): Likewise.
2239 (analyze_function): Add support for malloc attribute.
2240 (pure_const_write_summary): Stream malloc_state.
2241 (pure_const_read_summary): Add support for reading malloc_state.
2242 (dump_malloc_lattice): New function.
2243 (propagate_malloc): New function.
2244 (warn_function_malloc): New function.
2245 (ipa_pure_const::execute): Call propagate_malloc and
2246 ipa_free_fn_summary.
2247 (pass_local_pure_const::execute): Add support for malloc attribute.
2248 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2249 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2251 2017-10-27 Martin Liska <mliska@suse.cz>
2253 PR gcov-profile/82457
2254 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2255 for fork-like functions to be properly instrumented.
2257 2017-10-27 Richard Biener <rguenther@suse.de>
2260 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2261 info when we redirected EH.
2263 2017-10-26 Michael Collison <michael.collison@arm.com>
2265 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2267 (<optab>_trunchf<GPI:mode>2: New pattern.
2268 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2269 * config/aarch64/iterators.md (wv): New mode attribute.
2270 (vf, VF): New mode attributes.
2271 (vgp, VGP): New mode attributes.
2272 (s): Update attribute with SImode and DImode prefixes.
2274 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2276 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2277 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2278 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2279 (nios2_option_overide): Initialize it. Don't allow R0-relative
2280 addressing with PIC.
2281 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2282 (nios2_symbolic_constant_p): Likewise.
2283 (nios2_legitimate_address_p): Likewise.
2284 (nios2_r0rel_section_name_p): New.
2285 (nios2_symbol_ref_in_r0rel_data_p): New.
2286 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2287 (r0rel_constant_p): New.
2288 (nios2_print_operand_address): Handle r0rel_constant_p.
2289 (nios2_cdx_narrow_form_p): Likewise.
2290 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2291 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2292 (Nios II Options): Document -mr0rel-sec.
2294 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2296 * config/nios2/nios2.c: Include xregex.h.
2297 (nios2_gprel_sec_regex): New.
2298 (nios2_option_overide): Initialize it. Don't allow GP-relative
2299 addressing with PIC.
2300 (nios2_small_section_name_p): Check for regex match.
2301 * config/nios2/nios2.opt (mgprel-sec=): New option.
2302 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2303 (Nios II Options): Document -mgprel-sec.
2305 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2307 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2309 2017-10-26 Tom de Vries <tom@codesourcery.com>
2311 PR tree-optimization/82707
2312 * gimple.c (gimple_copy): Fix unsharing of
2313 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2315 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2317 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2318 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2319 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2320 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2321 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2322 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2323 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2324 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2325 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2326 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2327 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2328 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2329 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2330 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2331 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2332 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2333 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2335 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2337 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2339 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2340 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2341 warning to rs6000.c. Remove the Undocumented flag, since it has
2343 (-mabi=ibmlongdouble): Likewise.
2344 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2345 already set, set the default format for long double.
2346 (rs6000_debug_reg_global): Print whether long double is IBM or
2348 (rs6000_option_override_internal): Rework setting long double
2349 format. Only warn if the user is changing the long double default
2350 and they did not use -Wno-psabi.
2351 * doc/invoke.texi (PowerPC options): Update the documentation for
2352 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2354 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2355 Alan Hayward <alan.hayward@arm.com>
2356 David Sherwood <david.sherwood@arm.com>
2358 * rtl.h (wider_subreg_mode): New function.
2359 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2360 rather than an unsigned int *.
2361 * ira-color.c (regno_max_ref_width): Replace with...
2362 (regno_max_ref_mode): ...this new variable.
2363 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2364 Use wider_subreg_mode.
2365 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2366 rather than an unsigned int *.
2367 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2368 (process_alt_operands): Likewise.
2369 (invariant_p): Likewise.
2370 * lra-spills.c (assign_mem_slot): Likewise.
2371 (add_pseudo_to_slot): Likewise.
2372 * lra.c (collect_non_operand_hard_regs): Likewise.
2373 (add_regs_to_insn_regno_info): Likewise.
2374 * reload1.c (regno_max_ref_width): Replace with...
2375 (regno_max_ref_mode): ...this new variable.
2376 (reload): Update accordingly. Update call to
2377 ira_sort_regnos_for_alter_reg.
2378 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2379 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2380 (scan_paradoxical_subregs): Likewise.
2382 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2384 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2385 (aarch64_frame): Add emit_frame_chain boolean.
2386 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2387 Move eh_return case to aarch64_layout_frame.
2388 (aarch64_layout_frame): Initialize emit_frame_chain.
2389 (aarch64_expand_prologue): Use emit_frame_chain.
2391 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2393 * config/aarch64/aarch64.c (aarch64_layout_frame):
2394 Ensure LR is always stored at the bottom of the callee-saves.
2395 Remove rarely used frame layout which saves callee-saves at top of
2396 frame, so the store of LR can be used as a valid probe in all cases.
2398 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2400 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2401 Improve unaligned TImode/TFmode base/offset split.
2403 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2404 Alan Hayward <alan.hayward@arm.com>
2405 David Sherwood <david.sherwood@arm.com>
2407 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2408 * combine.c (find_single_use_1): Likewise.
2409 (expand_field_assignment): Likewise.
2410 (move_deaths): Likewise.
2411 * lra-constraints.c (simplify_operand_subreg): Likewise.
2412 (curr_insn_transform): Likewise.
2413 * lra.c (collect_non_operand_hard_regs): Likewise.
2414 (add_regs_to_insn_regno_info): Likewise.
2415 * rtlanal.c (reg_referenced_p): Likewise.
2416 (covers_regno_no_parallel_p): Likewise.
2418 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2420 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2421 Don't print any bits outside the precision of the value.
2422 * wide-int.cc (test_printing): Add some new tests.
2424 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2426 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2427 supports -xbrace_comment option.
2428 * configure: Regenerate.
2429 * config.in: Regenerate.
2430 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2431 (ASM_CPU_SPEC): Use it.
2433 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2435 * target.def (static_rtx_alignment): New hook.
2436 * targhooks.h (default_static_rtx_alignment): Declare.
2437 * targhooks.c (default_static_rtx_alignment): New function.
2438 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2439 * doc/tm.texi: Regenerate.
2440 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2441 instead of targetm.constant_alignment. Remove call to
2443 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2444 (cris_preferred_mininum_alignment): New function, split out from...
2445 (cris_constant_alignment): ...here.
2446 (cris_static_rtx_alignment): New function.
2447 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2449 (ix86_constant_alignment): ...here.
2450 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2451 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2452 (mmix_static_rtx_alignment): New function.
2453 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2454 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2456 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2459 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2460 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2462 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2464 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2465 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2466 mark as a sign-extending load.
2467 (local_pic_load_u): Define.
2469 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2472 * fold-const.c (operand_equal_for_comparison_p): Also return true
2473 if ARG0 is a simple variant of ARG1 with narrower precision.
2474 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2476 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2478 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2480 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2481 scatter_static, scatter_per_elt.
2482 * x86-tune-costs.h: Add new cost entries.
2484 2017-10-25 Richard Biener <rguenther@suse.de>
2486 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2487 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2488 class pass_fre): Move to ...
2489 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2490 class pass_fre): ... here and adjust for statistics.
2492 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2495 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2496 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2497 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2498 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2501 2017-10-25 Richard Biener <rguenther@suse.de>
2503 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2504 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2505 eliminate_push_avail, eliminate_insert): Move inside...
2506 (class eliminate_dom_walker): ... this class in preparation
2508 (fini_eliminate): Remove by merging with ...
2509 (eliminate): ... this function. Adjust for class changes.
2510 (pass_pre::execute): Remove fini_eliminate call.
2511 (pass_fre::execute): Likewise.
2513 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2516 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2517 (VPERMI2, VPERMI2I): New mode iterators.
2518 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2519 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2521 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2522 mode iterator. Remove 3 old define_insn patterns.
2523 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2524 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2525 VPERMI2 mode iterator, remove the other two expanders.
2526 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2527 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2528 instructions, remove the other two patterns.
2529 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2530 mode iterator, remove the other two patterns.
2531 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2532 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2533 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2534 and adjust argument order accordingly.
2535 (ix86_expand_vec_perm): Adjust caller.
2536 (expand_vec_perm_1): Likewise.
2537 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2538 (expand_vec_perm_vpermt2_vpshub2): ... this.
2539 (ix86_expand_vec_perm_const_1): Adjust caller.
2540 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2543 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2544 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2545 (vec_shl_<mode>): Remove unused expander.
2546 (avx512bw_<shift_insn><mode>3): New define_insn.
2547 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2548 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2550 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2553 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2556 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2558 PR rtl-optimization/82396
2559 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2560 (autopref_multipass_init): Simplify initialization.
2561 (autopref_rank_data): Simplify sort order.
2562 * gcc/sched-int.h (autopref_multipass_data_): Remove
2563 multi_mem_insn_p, min_offset and max_offset.
2565 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2568 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2569 Check special value of flag_omit_frame_pointer.
2570 (aarch64_can_eliminate): Likewise.
2571 (aarch64_override_options_after_change_1): Simplify handling of
2572 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2574 2017-10-24 Richard Biener <rguenther@suse.de>
2576 PR tree-optimization/82697
2577 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2578 zero for conditional load and unconditional store.
2580 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2582 * doc/install.texi: Document bootstrap-cet.
2584 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2587 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2588 ENDBR instruction at function entrance if function is only
2591 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2594 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2595 patterns to better describe from which operation the CF is computed.
2596 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2597 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2598 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
2599 is 0, use _0 suffixed expanders instead of emitting a comparison
2602 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
2604 * config/i386/i386.md(*movsf_internal, *movdf_internal):
2605 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
2607 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
2610 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
2611 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
2612 * loop-iv.c (iv_get_reaching_def): Likewise.
2613 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
2614 variable is promoted and the partition contains undefined values.
2616 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2618 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2620 (nios2_address_cost): Define.
2621 (nios2_legitimize_address): Recognize (exp + constant) directly.
2622 (TARGET_ADDRESS_COST): Define.
2624 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2626 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2627 (nios2_symbolic_memory_operand_p): Declare.
2628 (nios2_split_large_constant): Declare.
2629 (nios2_split_symbolic_memory_operand): Declare.
2630 * config/nios2/nios2.c: Adjust includes.
2631 (nios2_symbolic_constant_allowed): New.
2632 (nios2_symbolic_constant_p): New.
2633 (nios2_plus_symbolic_constant_p): New.
2634 (nios2_valid_addr_expr_p): Recognize addresses involving
2636 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2637 (nios2_symbolic_memory_operand_p): New.
2638 (nios2_large_constant_p): New.
2639 (nios2_split_large_constant): New.
2640 (nios2_split_plus_large_constant): New.
2641 (nios2_split_symbolic_memory_operand): New.
2642 (nios2_legitimize_address): Code refactoring. Handle addresses
2643 involving symbolic constants.
2644 (nios2_emit_move_sequence): Likewise.
2645 (nios2_print_operand): Improve error output.
2646 (nios2_print_operand_address): Handle LO_SUM.
2647 (nios2_cdx_narrow_form_p): Likewise.
2648 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2649 operands involving symbolic constants.
2650 (movhi_internal, movsi_internal): Likewise.
2651 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2652 (extendhisi2, extendqi<mode>2): Likewise.
2654 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2656 * tree-pass.h (PROP_rtl_split_insns): Define.
2657 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2659 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2661 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2663 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2666 * target.def (const_not_ok_for_debug_p): Default to
2667 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2668 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2669 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2670 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2671 which targetm.const_not_ok_for_debug_p returned true.
2672 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2674 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2676 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2677 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2678 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2680 (ix86_const_not_ok_for_debug_p): New function.
2681 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2682 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2684 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2687 * system.h: Conditionally include "unique-ptr.h" if
2688 INCLUDE_UNIQUE_PTR is defined.
2689 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2690 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2692 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2694 * config/rl78/rl78.md: New define_expand "subdi3".
2696 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2699 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2700 DF_REF_INSN if DF_REF_INSN_INFO is false.
2702 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2704 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2705 xmm_move instead of sse_move.
2706 (sse_store_index): New function.
2707 (ix86_register_move_cost): Be more sensible about mismatch stall;
2708 model AVX moves correctly; make difference between sse->integer and
2710 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2711 moves; make difference between SSE and AVX.
2712 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2713 and zmm_move. Increase size of sse load and store tables;
2714 add unaligned load and store tables; add ssemmx_to_integer.
2715 * x86-tune-costs.h: Update all entries according to real
2716 move latencies from Agner Fog's manual and chip documentation.
2718 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2721 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2722 * config/i386/constraints.md (Wf): New constraint.
2723 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2724 (cmp<dwi>_doubleword): Removed.
2725 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2726 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2727 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2728 expand with cmp<dwi>_doubleword. For LTU and GEU use
2729 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2731 * common.opt (gcolumn-info): Enable by default.
2732 * doc/invoke.texi (gcolumn-info): Document new default.
2734 2017-10-23 Richard Biener <rguenther@suse.de>
2736 PR tree-optimization/82672
2737 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2738 Fold the stmt if we propagated into it.
2740 2017-10-23 Richard Biener <rguenther@suse.de>
2742 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2743 (bitmap_remove_expr_from_set): ... this. All callers call this
2744 for non-constant values.
2745 (bitmap_set_subtract): Rename to...
2746 (bitmap_set_subtract_expressions): ... this. Adjust and
2748 (bitmap_set_contains_value): Remove superfluous check.
2749 (bitmap_set_replace_value): Inline into single caller ...
2750 (bitmap_value_replace_in_set): ... here and simplify.
2751 (dependent_clean): Merge into ...
2752 (clean): ... this using an overload. Adjust.
2753 (prune_clobbered_mems): Adjust.
2754 (compute_antic_aux): Likewise.
2755 (compute_partial_antic_aux): Likewise.
2757 2017-10-23 Richard Biener <rguenther@suse.de>
2759 PR tree-optimization/82129
2761 2017-08-01 Richard Biener <rguenther@suse.de>
2763 PR tree-optimization/81181
2764 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2765 (compute_antic): ... end of iteration here.
2767 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2769 * target.def (starting_frame_offset): New hook.
2770 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2771 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2772 * doc/tm.texi.in: Regenerate.
2773 * hooks.h (hook_hwi_void_0): Declare.
2774 * hooks.c (hook_hwi_void_0): New function.
2775 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2776 STARTING_FRAME_OFFSET.
2777 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2778 * reload1.c (reload): Likewise.
2779 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2780 instead of STARTING_FRAME_OFFSET.
2781 * function.c (try_fit_stack_local): Likewise.
2782 (assign_stack_local_1): Likewise
2783 (instantiate_virtual_regs): Likewise.
2784 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2785 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2786 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2787 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2788 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2789 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2790 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2791 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2792 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2793 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2794 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2795 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2796 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2797 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2798 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2799 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2800 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2801 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2802 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2803 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2804 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2805 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2806 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2807 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2808 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2809 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2810 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2811 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2812 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2813 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2814 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2815 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2816 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2817 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2818 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2819 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2820 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2821 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2822 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2823 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2824 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2825 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2826 return a HOST_WIDE_INT.
2827 (avr_builtin_setjmp_frame_value): Use it instead of
2828 STARTING_FRAME_OFFSET.
2829 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2830 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2831 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2833 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2834 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2835 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2836 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2837 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2838 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2839 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2840 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2841 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2842 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2843 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2844 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2846 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2847 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2848 * config/mips/mips.c (mips_compute_frame_info): Refer to
2849 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2850 (mips_starting_frame_offset): New function.
2851 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2852 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2853 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2854 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2855 and return a HOST_WIDE_INT.
2856 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2857 (mmix_initial_elimination_offset): Refer to
2858 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2859 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2860 * config/pa/pa.c (pa_starting_frame_offset): New function.
2861 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2862 (pa_expand_prologue): Likewise.
2863 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2864 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2865 !FRAME_GROWS_DOWNWARD handling to...
2866 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2867 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2868 !FRAME_GROWS_DOWNWARD handling to...
2869 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2870 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2871 !FRAME_GROWS_DOWNWARD handling to...
2872 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2873 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2875 (rs6000_starting_frame_offset): New function.
2876 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2877 !FRAME_GROWS_DOWNWARD handling to...
2878 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2879 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2880 !FRAME_GROWS_DOWNWARD handling to...
2881 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2882 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2883 !FRAME_GROWS_DOWNWARD handling to...
2884 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2885 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2886 (rs6000_starting_frame_offset): New function.
2887 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2888 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2889 * config/vax/vax.c (vax_starting_frame_offset): New function.
2890 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2891 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2892 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2893 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2894 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2895 * system.h (STARTING_FRAME_OFFSET): Poison.
2897 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2899 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2900 SCALAR_TYPE_MODE instead of TYPE_MODE.
2902 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2903 Alan Hayward <alan.hayward@arm.com>
2904 David Sherwood <david.sherwood@arm.com>
2906 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2908 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2909 Alan Hayward <alan.hayward@arm.com>
2910 David Sherwood <david.sherwood@arm.com>
2912 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
2914 2017-10-23 Richard Biener <rguenther@suse.de>
2916 PR tree-optimization/82129
2917 * tree-ssa-pre.c (bitmap_set_and): Remove.
2918 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
2919 canonicalizing expressions in the set to those with lowest
2920 ID rather than taking that from the first edge.
2922 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2924 * combine.c (rtx_equal_for_field_assignment_p): Use
2925 byte_lowpart_offset.
2927 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2928 Alan Hayward <alan.hayward@arm.com>
2929 David Sherwood <david.sherwood@arm.com>
2931 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
2932 to a SUBREG_PROMOTED_VAR.
2934 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2935 Alan Hayward <alan.hayward@arm.com>
2936 David Sherwood <david.sherwood@arm.com>
2938 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
2939 (expand_debug_source_expr): Likewise.
2940 * combine.c (combine_simplify_rtx): Likewise.
2941 * cse.c (fold_rtx): Likewise.
2942 * optabs.c (expand_float): Likewise.
2943 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2944 (simplify_binary_operation_1): Likewise.
2946 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2947 Alan Hayward <alan.hayward@arm.com>
2948 David Sherwood <david.sherwood@arm.com>
2950 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
2951 (record_promoted_value): Likewise.
2952 * expr.c (expand_expr_real_2): Likewise.
2953 * ree.c (update_reg_equal_equiv_notes): Likewise.
2954 (combine_set_extension): Likewise.
2955 * rtlanal.c (low_bitmask_len): Likewise.
2956 * simplify-rtx.c (neg_const_int): Likewise.
2957 (simplify_binary_operation_1): Likewise.
2959 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2960 Alan Hayward <alan.hayward@arm.com>
2961 David Sherwood <david.sherwood@arm.com>
2963 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
2964 * regcprop.c (maybe_mode_change): Likewise.
2965 * reload1.c (alter_reg): Likewise.
2967 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2969 * inchash.h (inchash::hash::add_wide_int): New function.
2970 * lto-streamer-out.c (hash_tree): Use it.
2972 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2974 * inchash.h (inchash::hash::add_wide_int): Rename to...
2975 (inchash::hash::add_hwi): ...this.
2976 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
2977 (polymorphic_call_target_hasher::hash): Likewise.
2978 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
2979 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
2980 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
2981 * lto-streamer-out.c (hash_tree): Likewise.
2982 * optc-save-gen.awk: Likewise.
2983 * tree.c (add_expr): Likewise.
2985 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2988 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
2989 for ordered inequality comparisons even with TARGET_IEEE_FP.
2991 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
2994 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
2995 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2996 Expand with cmp<dwi>_doubleword.
2998 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3000 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3001 List CET intrinsics.
3002 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3003 specific to -fcf-protection option.
3005 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3007 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3008 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3009 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3010 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3011 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3012 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3013 (extra_objs): Add cet.o for Linux/x86 targets.
3014 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3015 * config/i386/cet.c: New file.
3016 * config/i386/cetintrin.h: Likewise.
3017 * config/i386/t-cet: Likewise.
3018 * config/i386/cpuid.h (bit_SHSTK): New.
3019 (bit_IBT): Likewise.
3020 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3021 pass IBT and SHSTK bits.
3022 * config/i386/i386-builtin-types.def
3023 (VOID_FTYPE_UNSIGNED_PVOID): New.
3024 (VOID_FTYPE_UINT64_PVOID): Likewise.
3025 * config/i386/i386-builtin.def: Add CET intrinsics.
3026 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3027 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3028 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3029 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3031 * config/i386/i386.c (rest_of_insert_endbranch): New.
3032 (pass_data_insert_endbranch): Likewise.
3033 (pass_insert_endbranch): Likewise.
3034 (make_pass_insert_endbranch): Likewise.
3035 (ix86_notrack_prefixed_insn_p): Likewise.
3036 (ix86_target_string): Add -mibt, -mshstk flags.
3037 (ix86_option_override_internal): Add flag_cf_protection
3039 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3040 (ix86_print_operand): Add 'notrack' prefix output.
3041 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3042 (ix86_expand_builtin): Expand CET intrinsics.
3043 (x86_output_mi_thunk): Add 'endbranch' instruction.
3044 * config/i386/i386.h (TARGET_IBT): New.
3045 (TARGET_IBT_P): Likewise.
3046 (TARGET_SHSTK): Likewise.
3047 (TARGET_SHSTK_P): Likewise.
3048 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3049 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3050 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3051 (builtin_setjmp_setup): New pattern.
3052 (builtin_longjmp): Likewise.
3053 (rdssp<mode>): Likewise.
3054 (incssp<mode>): Likewise.
3055 (saveprevssp): Likewise.
3056 (rstorssp): Likewise.
3057 (wrss<mode>): Likewise.
3058 (wruss<mode>): Likewise.
3059 (setssbsy): Likewise.
3060 (clrssbsy): Likewise.
3061 (nop_endbr): Likewise.
3062 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3064 * config/i386/immintrin.h: Include <cetintrin.h>.
3065 * config/i386/linux-common.h
3066 (file_end_indicate_exec_stack_and_cet): New prototype.
3067 (TARGET_ASM_FILE_END): New.
3069 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3071 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3072 latencies instead of having separate table; make difference between
3073 integer and float costs.
3074 * i386.h (processor_costs): Remove scalar_stmt_cost,
3075 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3076 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3078 * x86-tune-costs.h: Remove entries which has been removed in
3079 procesor_costs from all tables; make cond_taken_branch_cost
3080 and cond_not_taken_branch_cost COST_N_INSNS based.
3082 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3084 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3086 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3088 * config/i386/i386.md (isa): Remove fma_avx512f.
3089 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3090 <avx512>_fmadd_<mode>_mask3<round_name>,
3091 <avx512>_fmsub_<mode>_mask<round_name>,
3092 <avx512>_fmsub_<mode>_mask3<round_name>,
3093 <avx512>_fnmadd_<mode>_mask<round_name>,
3094 <avx512>_fnmadd_<mode>_mask3<round_name>,
3095 <avx512>_fnmsub_<mode>_mask<round_name>,
3096 <avx512>_fnmsub_<mode>_mask3<round_name>,
3097 <avx512>_fmaddsub_<mode>_mask<round_name>,
3098 <avx512>_fmaddsub_<mode>_mask3<round_name>,
3099 <avx512>_fmsubadd_<mode>_mask<round_name>,
3100 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
3101 (*vec_widen_umult_even_v16si<mask_name>,
3102 *vec_widen_smult_even_v16si<mask_name>): Likewise.
3103 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
3105 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3107 * extend.texi: Add 'nocf_check' documentation.
3108 * gimple.texi: Add second parameter to
3109 gimple_build_call_from_tree.
3110 * invoke.texi: Add -fcf-protection documentation.
3111 * rtl.texi: Add REG_CALL_NOTRACK documenation.
3113 2017-10-20 Richard Biener <rguenther@suse.de>
3115 PR tree-optimization/82473
3116 * tree-vect-loop.c (vectorizable_reduction): Properly get at
3117 the largest input type.
3119 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3121 * c-attribs.c (handle_nocf_check_attribute): New function.
3122 (c_common_attribute_table): Add 'nocf_check' handling.
3123 * gimple-parser.c: Add second argument NULL to
3124 gimple_build_call_from_tree.
3125 * attrib.c (comp_type_attributes): Check nocf_check attribute.
3126 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
3128 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
3129 * common.opt: Add fcf-protection flag.
3130 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
3131 * flag-types.h: Add enum cf_protection_level.
3132 * gimple.c (gimple_build_call_from_tree): Add second parameter.
3133 Add 'nocf_check' attribute propagation to gimple call.
3134 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
3135 (gimple_build_call_from_tree): Update prototype.
3136 (gimple_call_nocf_check_p): New function.
3137 (gimple_call_set_nocf_check): Likewise.
3138 * gimplify.c: Add second argument to gimple_build_call_from_tree.
3139 * ipa-icf.c: Add nocf_check attribute in statement hash.
3140 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
3141 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
3142 * toplev.c (process_options): Add flag_cf_protection handling.
3144 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3146 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
3148 2017-10-20 Richard Biener <rguenther@suse.de>
3150 PR tree-optimization/82603
3151 * tree-if-conv.c (predicate_mem_writes): Make sure to only
3152 remove false predicated stores.
3154 2017-10-20 Richard Biener <rguenther@suse.de>
3156 * graphite-isl-ast-to-gimple.c
3157 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
3158 Remove return value and simplify, dump copied stmt after lhs
3160 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
3161 Reduce dump verbosity.
3162 (gsi_insert_earliest): Likewise.
3163 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3164 * graphite.c (print_global_statistics): Adjust dumping.
3165 (print_graphite_scop_statistics): Likewise.
3166 (print_graphite_statistics): Do not dump loops here.
3167 (graphite_transform_loops): But here.
3169 2017-10-20 Nicolas Roche <roche@adacore.com>
3171 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3172 * configure: Regenerate.
3174 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3177 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3178 functions when optimizing replace GIMPLE_RETURN stmts with
3179 calls to __builtin_unreachable ().
3182 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3183 for -fsanitize=thread link of executables.
3184 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3185 link of executables.
3188 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3190 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3191 define_insns for logical vector shifts to use VI248_AVX512BW
3192 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3193 condition, useless isa and prefix attributes. Change the first
3194 2 of these define_insns to ...
3195 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3196 define_insn for avx512vl.
3197 (<shift_insn><mode>3): ... and this, new define_insn without
3198 masking for non-avx512vl.
3201 * config/i386/sse.md (*andnot<mode>3,
3202 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3203 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3204 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3205 not applied use empty suffix even for TARGET_AVX512VL.
3206 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3207 is applied, supply evex,evex or evex,evex,evex instead of just
3210 2017-10-20 Julia Koval <julia.koval@intel.com>
3212 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3213 (OPTION_MASK_ISA_GFNI_UNSET): New.
3214 (ix86_handle_option): Handle OPT_mgfni.
3215 * config/i386/cpuid.h (bit_GFNI): New.
3216 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3217 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3218 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3219 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3220 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3221 * config/i386/i386.opt: Add mgfni.
3223 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3225 * config/msp430/msp430.c (msp430_option_override): Disable
3226 -fdelete-null-pointer-checks.
3227 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3229 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3231 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3232 of x87 and SSE instructions.
3234 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3236 * asan.c (create_cond_insert_point): Do not update edge count.
3237 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3238 (afdo_propagate_circuit): Likewise.
3239 (afdo_calculate_branch_prob): Likewise.
3240 (afdo_annotate_cfg): Likewise.
3241 * basic-block.h (struct edge_def): Remove count.
3242 (edge_def::count): New accessor.
3243 * bb-reorder.c (rotate_loop): Update.
3244 (find_traces_1_round): Update.
3245 (connect_traces): Update.
3246 (sanitize_hot_paths): Update.
3247 * cfg.c (unchecked_make_edge): Update.
3248 (make_single_succ_edge): Update.
3249 (check_bb_profile): Update.
3250 (dump_edge_info): Update.
3251 (update_bb_profile_for_threading): Update.
3252 (scale_bbs_frequencies_int): Update.
3253 (scale_bbs_frequencies_gcov_type): Update.
3254 (scale_bbs_frequencies_profile_count): Update.
3255 (scale_bbs_frequencies): Update.
3256 * cfganal.c (connect_infinite_loops_to_exit): Update.
3257 * cfgbuild.c (compute_outgoing_frequencies): Update.
3258 (find_many_sub_basic_blocks): Update.
3259 * cfgcleanup.c (try_forward_edges): Update.
3260 (try_crossjump_to_edge): Update
3261 * cfgexpand.c (expand_gimple_cond): Update
3262 (expand_gimple_tailcall): Update
3263 (construct_exit_block): Update
3264 * cfghooks.c (verify_flow_info): Update
3265 (redirect_edge_succ_nodup): Update
3266 (split_edge): Update
3267 (make_forwarder_block): Update
3268 (duplicate_block): Update
3269 (account_profile_record): Update
3270 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3271 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3272 * cfgloopmanip.c (scale_loop_profile): Update.
3274 (lv_adjust_loop_entry_edge): Update.
3275 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3276 (force_nonfallthru_and_redirect): Update.
3277 (purge_dead_edges): Update.
3278 (rtl_flow_call_edges_add): Update.
3279 * cgraphunit.c (init_lowered_empty_function): Update.
3280 (cgraph_node::expand_thunk): Update.
3281 * gimple-pretty-print.c (dump_probability): Update.
3282 (dump_edge_probability): Update.
3283 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3284 * haifa-sched.c (sched_create_recovery_edges): Update.
3285 * hsa-gen.c (convert_switch_statements): Update.
3286 * ifcvt.c (dead_or_predicable): Update.
3287 * ipa-inline-transform.c (inline_transform): Update.
3288 * ipa-split.c (split_function): Update.
3289 * ipa-utils.c (ipa_merge_profiles): Update.
3290 * loop-doloop.c (add_test): Update.
3291 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3292 * lto-streamer-in.c (input_cfg): Update.
3293 (input_function): Update.
3294 * lto-streamer-out.c (output_cfg): Update.
3295 * modulo-sched.c (sms_schedule): Update.
3296 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3297 * predict.c (maybe_hot_edge_p): Update.
3298 (unlikely_executed_edge_p): Update.
3299 (probably_never_executed_edge_p): Update.
3300 (dump_prediction): Update.
3301 (drop_profile): Update.
3302 (propagate_unlikely_bbs_forward): Update.
3303 (determine_unlikely_bbs): Update.
3304 (force_edge_cold): Update.
3305 * profile.c (compute_branch_probabilities): Update.
3306 * reg-stack.c (better_edge): Update.
3307 * shrink-wrap.c (handle_simple_exit): Update.
3308 * tracer.c (better_p): Update.
3309 * trans-mem.c (expand_transaction): Update.
3310 (split_bb_make_tm_edge): Update.
3311 * tree-call-cdce.c: Update.
3312 * tree-cfg.c (gimple_find_sub_bbs): Update.
3313 (gimple_split_edge): Update.
3314 (gimple_duplicate_sese_region): Update.
3315 (gimple_duplicate_sese_tail): Update.
3316 (gimple_flow_call_edges_add): Update.
3317 (insert_cond_bb): Update.
3318 (execute_fixup_cfg): Update.
3319 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3320 * tree-complex.c (expand_complex_div_wide): Update.
3321 * tree-eh.c (lower_resx): Update.
3322 (unsplit_eh): Update.
3323 (cleanup_empty_eh_move_lp): Update.
3324 * tree-inline.c (copy_edges_for_bb): Update.
3325 (freqs_to_counts): Update.
3326 (copy_cfg_body): Update.
3327 * tree-ssa-dce.c (remove_dead_stmt): Update.
3328 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3329 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3330 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3331 (unloop_loops): Update.
3332 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3333 * tree-ssa-loop-split.c (connect_loops): Update.
3334 (split_loop): Update.
3335 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3336 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3337 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3338 * tree-ssa-reassoc.c (branch_fixup): Update.
3339 * tree-ssa-tail-merge.c (replace_block_by): Update.
3340 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3341 (compute_path_counts): Update.
3342 (update_profile): Update.
3343 (recompute_probabilities): Update.
3344 (update_joiner_offpath_counts): Update.
3345 (estimated_freqs_path): Update.
3346 (freqs_to_counts_path): Update.
3347 (clear_counts_path): Update.
3348 (ssa_fix_duplicate_block_edges): Update.
3349 (duplicate_thread_path): Update.
3350 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3351 (case_bit_test_cmp): Update.
3352 (collect_switch_conv_info): Update.
3353 (gen_inbound_check): Update.
3354 (do_jump_if_equal): Update.
3355 (emit_cmp_and_jump_insns): Update.
3356 * tree-tailcall.c (decrease_profile): Update.
3357 (eliminate_tail_call): Update.
3358 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3359 (vect_do_peeling): Update.
3360 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3361 * ubsan.c (ubsan_expand_null_ifn): Update.
3362 (ubsan_expand_ptr_ifn): Update.
3363 * value-prof.c (gimple_divmod_fixed_value): Update.
3364 (gimple_mod_pow2): Update.
3365 (gimple_mod_subtract): Update.
3366 (gimple_ic): Update.
3367 (gimple_stringop_fixed_value): Update.
3369 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3372 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3374 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3376 PR rtl-optimization/82395
3377 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3378 based on non_spilled_static_chain_regno_p.
3380 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3382 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3383 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3384 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3386 2017-10-19 Martin Sebor <msebor@redhat.com>
3388 PR tree-optimization/82596
3389 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3391 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3393 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3394 (get_mem_refs_of_builtin_call): Likewise.
3395 * builtins.c (expand_builtin_apply): Adjust call to
3396 allocate_dynamic_stack_space.
3397 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3398 the third argument to allocate_dynamic_stack_space, otherwise -1.
3399 (expand_builtin): Deal with all alloca variants.
3400 (is_inexpensive_builtin): Likewise.
3401 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3402 * calls.c (special_function_p): Deal with all alloca variants.
3403 (initialize_argument_information): Adjust call to
3404 allocate_dynamic_stack_space.
3405 (expand_call): Likewise.
3406 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3407 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3408 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3409 use it for the stack usage computation.
3410 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3411 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3412 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3413 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3414 (in_loop_p): Remove first argument and useless check.
3415 (pass_walloca::execute): Remove useless test and adjust call to above.
3416 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3417 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3418 (gimplify_call_expr): Deal with all alloca variants.
3419 * hsa-gen.c (gen_hsa_alloca): Likewise.
3420 (gen_hsa_insns_for_call): Likewise.
3421 * ipa-pure-const.c (special_builtin_state): Likewise.
3422 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3423 * tree-object-size.c (alloc_object_size): Likewise.
3424 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3425 (call_may_clobber_ref_p_1): Likewise.
3426 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3427 (ccp_fold_stmt): Likewise.
3428 (optimize_stack_restore): Likewise.
3429 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3430 (mark_all_reaching_defs_necessary_1): Likewise.
3431 (propagate_necessity): Likewise.
3432 (eliminate_unnecessary_stmts): Likewise.
3433 * tree.c (build_common_builtin_nodes): Build
3434 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3435 (build_alloca_call_expr): New function.
3436 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3437 (CASE_BUILT_IN_ALLOCA): Likewise.
3438 (build_alloca_call_expr): Declare.
3439 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3441 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3444 * dwarf2out.c (new_die_raw): New static inline function.
3445 (new_die): Use it to create the DIE.
3446 (add_AT_external_die_ref): Likewise.
3447 (clone_die): Likewise.
3448 (clone_as_declaration): Likewise.
3449 (dwarf2out_vms_debug_main_pointer): Likewise.
3450 (base_type_die): Likewise. Remove early return for corner cases.
3451 Do not call add_pubtype on the DIE here.
3452 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3453 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3454 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3455 native order exists for base types, attach the DIE manually and call
3456 add_pubtype on it. Do not equate a reverse order DIE to the type.
3458 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3460 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3461 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3463 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3465 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3467 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3468 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3469 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3470 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3471 * opts.c (sanitizer_opts): Add builtin.
3472 * ubsan.c (instrument_builtin): New function.
3473 (pass_ubsan::execute): Call it.
3474 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3475 * doc/invoke.texi: Document -fsanitize=builtin.
3477 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3478 builtins, store max (log2 (align), 0) into uchar field instead of
3479 align into uptr field.
3480 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3481 store uchar 0 field instead of uptr 0 field.
3482 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3483 instead of passing one address of struct with 2 locations pass
3484 two addresses of structs with 1 location each.
3485 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3486 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3487 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3488 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3489 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3490 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3491 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3492 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3494 2017-10-19 Martin Liska <mliska@suse.cz>
3497 * file-find.c (remove_prefix): Remove.
3498 * file-find.h (remove_prefix): Likewise.
3499 * gcc-ar.c: Remove smartness of lookup.
3501 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3503 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3504 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3505 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3507 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3510 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3511 (setcc + and to xor + setcc): New peephole2.
3513 2017-10-19 Tom de Vries <tom@codesourcery.com>
3515 * doc/sourcebuild.texi (Test Directives, Variants of
3516 dg-require-support): Add dg-require-stack-size.
3518 2017-10-19 Martin Liska <mliska@suse.cz>
3521 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3522 that have a large alignment.
3523 (gimplify_target_expr): Likewise.
3525 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3527 PR rtl-optimization/82602
3528 * ira.c (rtx_moveable_p): Return false for volatile asm.
3530 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3533 * config/i386/i386-modes.def (CCGZ): New CC mode.
3534 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3535 * config/i386/predicates.md (ix86_comparison_operator):
3537 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3538 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3539 with double-word subtraction.
3540 (put_condition_code): Handle CCGZmode.
3542 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3544 * wide-int.cc (debug (const wide_int &)): New.
3545 (debug (const wide_int *)): New.
3546 (debug (const widest_int &)): New.
3547 (debug (const widest_int *)): New.
3549 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3552 * lra-constraints.c (curr_insn_transform): Use non-input operand
3553 instead of output one for matched reload.
3555 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3557 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3558 (tree-ssa-loop-ivopts.h): New header file.
3559 (struct builtin_info): New fields.
3560 (classify_builtin_1): Compute and record base and offset parts for
3561 memset builtin partition by calling strip_offset.
3562 (offset_cmp, fuse_memset_builtins): New functions.
3563 (finalize_partitions): Fuse adjacent memset partitions by calling
3565 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3566 Expose the interface.
3567 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3569 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3571 PR tree-optimization/82574
3572 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3573 that data reference must be executed exactly once per iteration
3574 against the outermost loop in nest.
3575 (classify_partition): Update call to above function.
3577 2017-10-18 Richard Biener <rguenther@suse.de>
3579 PR tree-optimization/82591
3580 * graphite.c (graphite_transform_loops): Move code gen message
3582 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3583 Here. Handle scop_to_isl_ast failing.
3584 (scop_to_isl_ast): Limit the number of ISL operations.
3586 2017-10-18 Richard Biener <rguenther@suse.de>
3588 * graphite-isl-ast-to-gimple.c
3589 (translate_isl_ast_to_gimple::set_rename): Simplify.
3590 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3591 (graphite_copy_stmts_from_block): ... here.
3592 (copy_bb_and_scalar_dependences): Simplify.
3593 (add_parameters_to_ivs_params): Canonicalize.
3594 (generate_entry_out_of_ssa_copies): Simplify.
3595 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3596 by passing in ISL dimension.
3597 (parameter_index_in_region_1): Rename to ...
3598 (parameter_index_in_region): ... this.
3599 (extract_affine): Adjust assert, pass down parameter index.
3600 (add_param_constraints): Use range-info when available.
3601 (build_scop_context): Adjust.
3602 * sese.c (new_sese_info): Adjust.
3603 (free_sese_info): Likewise.
3604 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
3605 Remove unused typedefs.
3606 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
3608 2017-10-18 Martin Liska <mliska@suse.cz>
3610 * combine.c (simplify_compare_const): Add gcc_fallthrough.
3612 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3614 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3615 (s390_sched_init): Do not reset s390_sched_state if we entered the
3616 current basic block via a fallthru edge and all others are unlikely.
3618 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3620 * config/s390/s390.c (NUM_SIDES): New variable.
3621 (LONGRUNNING_THRESHOLD): New variable.
3622 (LATENCY_FACTOR): New variable.
3623 (s390_sched_score): Decrease score for long-running instructions on
3625 (s390_sched_variable_issue): Perform bookkeeping for long-running
3628 2017-10-18 Richard Biener <rguenther@suse.de>
3630 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3631 Simplify with removal of the parameter rename map.
3632 (set_rename): Likewise.
3633 (should_copy_to_new_region): Likewise.
3634 (graphite_copy_stmts_from_block): Likewise.
3635 (copy_bb_and_scalar_dependences): Remove initialization of
3636 unused copied_bb_map.
3638 (copy_internal_parameters): Likewise.
3639 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3640 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3641 Use INTEGRAL_TYPE_P.
3642 (parameter_index_in_region_1): Rename to ...
3643 (assign_parameter_index_in_region): ... this. Assert we have
3644 a parameter we handle.
3645 (scan_tree_for_params): Adjust.
3646 * sese.h (parameter_rename_map_t): Remove.
3647 (struct sese_info_t): Remove unused parameter_rename_map and
3648 copied_bb_map members.
3649 * sese.c (new_sese_info): Adjust.
3650 (free_sese_info): Likewise.
3652 2017-10-18 Martin Liska <mliska@suse.cz>
3655 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3656 on an abnormal edge.
3658 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3660 * doc/invoke.texi (ffunction-sections and fdata-sections):
3663 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3665 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3666 the use statement can throw internally.
3668 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3670 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3671 any RTX present on the RHS of a SET.
3672 * compare-elim.c (try_eliminate_compare): Restore comment.
3674 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3676 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3677 may be also called on tcc_type nodes.
3678 * langhooks.c (lhd_tree_size): Likewise.
3680 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3682 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3683 format_warning_at_substring.
3684 (maybe_warn): Convert source_range * param to a location_t. Pass
3685 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3686 (format_directive): Remove code to extract source_ranges and
3687 source_range * in favor of just a location_t.
3688 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3690 * substring-locations.c (format_warning_va): Convert
3691 source_range * param to a location_t.
3692 (format_warning_at_substring): Likewise.
3693 * substring-locations.h (format_warning_va): Likewise.
3694 (format_warning_at_substring): Likewise.
3696 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3698 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3700 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3701 and scatter/gather ops.
3703 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3704 vec_gather_load and vec_scatter_store.
3705 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3706 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3707 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3708 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3709 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3710 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3712 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3714 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3715 Detect FTST instruction and handle its register pops. Only pop
3716 second operand if can_pop_second_op is true.
3717 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3718 set can_pop_second_op to false in the compare_for_stack_reg call.
3720 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3721 output_fp_compare for stack register operands.
3722 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3723 instructions here. Do not emit stack register pops here. Assert
3724 that FCOMPP pops next to top stack register. Rewrite function.
3726 2017-10-17 Nathan Sidwell <nathan@acm.org>
3729 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3730 use DECL_ASSEMBLER_NAME_RAW.
3733 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3736 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3737 Wilco Dijkstra <wilco.dijkstra@arm.com>
3739 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3740 converstion to Pmode from the buf_addr.
3742 2017-10-17 Richard Biener <rguenther@suse.de>
3744 * graphite-dependences.c (scop_get_reads_and_writes): Change
3745 output parameters to references.
3747 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3749 PR 71026/tree-optimization
3750 * fold-const.c (distribute_real_division): Removed.
3751 (fold_binary_loc): Remove calls to distribute_real_divison.
3753 2017-10-17 Richard Biener <rguenther@suse.de>
3755 * graphite-scop-detection.c
3756 (scop_detection::stmt_has_simple_data_refs_p): Always use
3757 the full nest as region.
3758 (try_generate_gimple_bb): Likewise.
3759 * sese.c (scalar_evolution_in_region): Simplify now that
3760 SCEV can handle instantiation in regions.
3761 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3762 in the non-loop part of a function if requested.
3764 2017-10-17 Richard Biener <rguenther@suse.de>
3766 PR tree-optimization/82563
3767 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3769 (graphite_regenerate_ast_isl): Call it.
3770 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3772 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3774 PR tree-optimization/82549
3775 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3776 Formatting fixes. Instead of calling make_bit_field_ref with negative
3779 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3781 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3782 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3783 _mm_maskz_reduce_ss): New.
3784 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3785 __builtin_ia32_reducess_mask): Ditto..
3786 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3787 * config/i386/sse.md (reduces<mode>): Renamed to ...
3788 (reduces<mode><mask_scalar_name>): ... this.
3789 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3791 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3792 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3794 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3796 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3797 * selftest-run-tests.c (selftest::run_tests): Call
3798 selftest::unique_ptr_tests_cc_tests.
3799 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3800 * unique-ptr-tests.cc: New file.
3802 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3805 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3807 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3808 (make_hard_regno_born, make_hard_regno_dead): Update
3809 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3811 2017-10-16 Jeff Law <law@redhat.com>
3813 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3815 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3817 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3819 2017-10-16 Olivier Hainque <hainque@adacore.com>
3821 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3822 with_cpu if we were configured for an e500v2 target cpu name.
3824 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3826 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3827 * doc/invoke.texi: Document +nodsp as a valid extension for
3830 2017-10-16 Martin Liska <mliska@suse.cz>
3832 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3833 (test_set_range): Likewise.
3834 (test_range_functions): Rename to ...
3835 (test_bit_in_range): ... this.
3836 (sbitmap_c_tests): Add new test.
3838 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3840 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3842 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3843 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3845 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3847 * config/aarch64/aarch64-builtins.c
3848 (aarch64_types_quadopu_lane_qualifiers): New.
3849 (TYPES_QUADOPU_LANE): New.
3850 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3851 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3852 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3853 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3854 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3855 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3856 (Vdottype, DOTPROD): New.
3857 (sur): Add SDOT and UDOT.
3859 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3861 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3862 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3863 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3865 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3866 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3867 Enable TARGET_DOTPROD.
3868 (cortex-a75.cortex-a55): Likewise.
3869 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3871 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3873 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3874 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3876 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3878 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3879 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3880 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3881 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3882 * config/arm/types.md (neon_dot, neon_dot_q): New.
3883 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3885 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3887 * config/arm/arm.h (TARGET_DOTPROD): New.
3888 * config/arm/arm.c (arm_arch_dotprod): New.
3889 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3890 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3891 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3892 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3893 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3894 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3895 * doc/invoke.texi (armv8.2-a): Document dotprod
3897 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3899 * i386.c (ix86_vec_cost): New function.
3900 (ix86_rtx_costs): Handle vector operations better.
3901 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3902 * x86-tune-costs.h: Add new costs to all tables.
3904 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3906 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3908 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3909 divsd, sqrtss and sqrtsd
3910 * x86-tune-costs.h: Add new entries to all costs.
3911 (znver1_cost): Fix to match real instruction latencies.
3913 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3914 Michael Collison <michael.collison@arm.com>
3916 * compare-elim.c: Include emit-rtl.h.
3917 (can_merge_compare_into_arith): New function.
3918 (try_validate_parallel): Likewise.
3919 (try_merge_compare): Likewise.
3920 (try_eliminate_compare): Call the above when no previous clobber
3922 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
3925 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3929 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
3930 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
3934 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
3935 to be any operand_equal_p operands. For & (B - 1) require
3936 B to be power of 2. Recognize
3937 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
3939 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3942 * optabs.c (expand_memory_blockage): Fix call of
3943 targetm.have_memory_blockage.
3945 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3948 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
3949 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
3950 objects to extra_objs instead of overwriting it.
3952 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3954 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
3955 Use any_fp_register_operand as operand[3] predicate. Simplify
3956 equality test for operands[2] and operands[4] memory location.
3957 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
3958 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
3959 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
3960 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
3961 any_fp_register_operand as operand[1] predicate. Simplify
3962 equality test for operands[0] and operands[3] memory location.
3963 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
3964 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
3965 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
3967 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
3969 * target-insns.def: Add memory_blockage.
3970 * optabs.c (expand_memory_blockage): New function.
3971 (expand_asm_memory_barrier): Rename ...
3972 (expand_asm_memory_blockage): ... to this.
3973 (expand_mem_thread_fence): Call expand_memory_blockage
3974 instead of expand_asm_memory_barrier.
3975 (expand_mem_singnal_fence): Ditto.
3976 (expand_atomic_load): Ditto.
3977 (expand_atomic_store): Ditto.
3978 * doc/md.texi (Standard Pattern Names For Generation):
3979 Document memory_blockage instruction pattern.
3981 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
3983 * config/rl78/rl78.c (rl78_emit_libcall): New function.
3984 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
3985 * config/rl78/rl78.md: New define_expand "adddi3".
3987 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
3989 * cfghooks.c (verify_flow_info): Disable check that all probabilities
3992 2017-10-13 Jeff Law <law@redhat.com>
3994 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
3996 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3999 * internal-fn.c (expand_mul_overflow): If both operands have
4000 the same highpart of -1 or 0 and the topmost bit of lowpart
4001 is different, overflow is if res <= 0 rather than res < 0.
4003 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4005 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4006 TARGET_P9_VECTOR code for unaligned_load case.
4008 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4010 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4012 2017-10-13 Nathan Sidwell <nathan@acm.org>
4014 * tree-core.h (tree_contains_struct): Make bool.
4015 * tree.c (tree_contains_struct): Likewise.
4016 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4017 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4018 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4019 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4020 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4022 2017-10-13 Richard Biener <rguenther@suse.de>
4024 * graphite-isl-ast-to-gimple.c
4025 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4026 parameters and dominance check.
4027 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4028 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4029 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4030 Do not update SSA form here or do intermediate IL verification.
4031 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4032 (graphite_initialize): Remove check on the number of loops in
4033 the function and inline into graphite_transform_loops.
4034 (graphite_finalize): Inline into graphite_transform_loops.
4035 (graphite_transform_loops): Perform SSA update and IL verification
4037 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4039 2017-10-13 Richard Biener <rguenther@suse.de>
4041 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4042 graphite_expression_type_precision): Avoid global constructor
4044 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4045 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4046 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4047 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4048 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4049 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4051 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4054 * config/i386/i386.h (ix86_red_zone_size): New.
4055 * config/i386/i386.md (push peephole2s): Replace
4056 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4058 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4059 Alan Hayward <alan.hayward@arm.com>
4060 David Sherwood <david.sherwood@arm.com>
4062 * combine.c (can_change_dest_mode): Reject changes in
4063 REGMODE_NATURAL_SIZE.
4065 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4066 Alan Hayward <alan.hayward@arm.com>
4067 David Sherwood <david.sherwood@arm.com>
4069 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4070 (expand_debug_source_expr): Likewise.
4071 * combine.c (combine_simplify_rtx): Likewise.
4072 * cse.c (fold_rtx): Likewise.
4073 * fwprop.c (canonicalize_address): Likewise.
4074 * targhooks.c (default_shift_truncation_mask): Likewise.
4076 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4077 Alan Hayward <alan.hayward@arm.com>
4078 David Sherwood <david.sherwood@arm.com>
4080 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4081 (widened_mode): Likewise.
4082 (expand_unop): Likewise.
4083 * ree.c (transform_ifelse): Likewise.
4084 (merge_def_and_ext): Likewise.
4085 (combine_reaching_defs): Likewise.
4086 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4088 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4089 Alan Hayward <alan.hayward@arm.com>
4090 David Sherwood <david.sherwood@arm.com>
4092 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4093 * combine.c (gen_lowpart_for_combine): Likewise.
4094 * dwarf2out.c (rtl_for_decl_location): Likewise.
4095 * final.c (alter_subreg): Likewise.
4096 * rtlhooks.c (gen_lowpart_general): Likewise.
4097 (gen_lowpart_if_possible): Likewise.
4099 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4100 Alan Hayward <alan.hayward@arm.com>
4101 David Sherwood <david.sherwood@arm.com>
4103 * calls.c (expand_call): Use subreg_lowpart_offset.
4104 * cse.c (cse_insn): Likewise.
4105 * regcprop.c (copy_value): Likewise.
4106 (copyprop_hardreg_forward_1): Likewise.
4108 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4111 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
4112 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
4113 =Q constraints to +Q and into insn condition add check
4114 that operands[0] and operands[1] are equal.
4115 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
4116 =Q constraints to +Q and into insn condition add check
4117 that operands[0] is equal to either operands[1] or operands[2].
4120 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
4121 instead of handling MINUS_EXPR twice (once for each argument),
4122 canonicalize operand order and handle just once, use rtype where
4123 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
4126 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
4127 any values of __C while still being pattern recognizable as a simple
4130 2017-10-13 Richard Biener <rguenther@suse.de>
4132 PR tree-optimization/82451
4134 2017-10-02 Richard Biener <rguenther@suse.de>
4136 PR tree-optimization/82355
4137 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4138 a mapping for the enclosing loop but avoid generating one for
4140 (copy_bb_and_scalar_dependences): Remove premature codegen
4141 error on PHIs in blocks duplicated into multiple places.
4142 * graphite-scop-detection.c
4143 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4144 in the region use it as loop and nest to analyze the DR in.
4145 (try_generate_gimple_bb): Likewise.
4146 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4147 (add_loop_constraints): For blocks in a loop not in the region
4148 create a dimension with a single iteration.
4149 * sese.h (gbb_loop_at_index): Remove assert.
4151 * cfgloop.c (loop_preheader_edge): For the loop tree root
4152 return the single successor of the entry block.
4153 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4154 Reset the SCEV hashtable and niters.
4155 * graphite-scop-detection.c
4156 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
4157 assert that we only have POLYNOMIAL_CHREC that vary in loops
4158 contained in the region.
4159 (scop_detection::graphite_can_represent_expr): Adjust.
4160 (scop_detection::stmt_has_simple_data_refs_p): For loops
4161 not in the region set loop to NULL. The nest is now the
4162 entry edge to the region.
4163 (try_generate_gimple_bb): Likewise.
4164 * sese.c (scalar_evolution_in_region): Adjust for
4165 instantiate_scev change.
4166 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4167 Make nest parameter the edge into the region.
4168 (create_data_ref): Likewise.
4169 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4170 entry edge into a region and adjust instantiate_scev calls.
4171 (create_data_ref): Likewise.
4172 (graphite_find_data_references_in_stmt): Likewise.
4173 (find_data_references_in_stmt): Pass the loop preheader edge
4174 from the nest argument.
4175 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4176 parameter the edge into the region.
4177 (instantiate_parameters): Use the loop preheader edge as entry.
4178 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4180 (get_instantiated_value_entry): Make instantiate_below parameter
4181 the edge into the region.
4182 (instantiate_scev_name): Likewise. Adjust dominance checks,
4183 when we cannot use loop-based instantiation instantiate by
4184 walking use-def chains.
4185 (instantiate_scev_poly): Adjust.
4186 (instantiate_scev_binary): Likewise.
4187 (instantiate_scev_convert): Likewise.
4188 (instantiate_scev_not): Likewise.
4189 (instantiate_array_ref): Remove.
4190 (instantiate_scev_3): Likewise.
4191 (instantiate_scev_2): Likewise.
4192 (instantiate_scev_1): Likewise.
4193 (instantiate_scev_r): Do not blindly handle N-operand trees.
4194 Do not instantiate array-refs. Handle all constants and invariants.
4195 (instantiate_scev): Make instantiate_below parameter
4196 the edge into the region.
4197 (resolve_mixers): Use the loop preheader edge for the region
4198 parameter to instantiate_scev_r.
4199 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4201 2017-10-13 Richard Biener <rguenther@suse.de>
4203 PR tree-optimization/82525
4204 * graphite-isl-ast-to-gimple.c
4205 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4207 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4208 Fail code generation when we cannot represent the isl integer.
4209 (binary_op_to_tree): Elide modulo operations that are no-ops
4210 in the type we code generate. Remove now superfluous code
4213 2017-10-13 Richard Biener <rguenther@suse.de>
4215 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4216 (scop_detection::harmful_loop_in_region): Remove premature
4217 IV type restriction.
4218 (scop_detection::graphite_can_represent_scev): We can handle
4219 pointer IVs just fine.
4221 2017-10-13 Alan Modra <amodra@gmail.com>
4223 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4224 "Clobbers and Scratch Registers". Add paragraph on
4225 alternative to clobbers for scratch registers and OpenBLAS
4228 2017-10-13 Alan Modra <amodra@gmail.com>
4230 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4231 example of a memory input for a string of known length. Move
4232 commentary out of table. Add a number of new examples
4233 covering array memory inputs.
4235 2017-10-12 Martin Liska <mliska@suse.cz>
4237 PR tree-optimization/82493
4238 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4239 (test_range_functions): New function.
4240 (sbitmap_c_tests): Likewise.
4241 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4242 * selftest.h (sbitmap_c_tests): New function.
4244 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4246 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4248 * config/rs6000/amo.h: Fix spacing issue.
4250 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4253 * config/i386/i386.md (*ashl<mode>3_mask_1,
4254 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4255 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4258 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4260 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4261 (profile_probability): Set max_probability
4262 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4264 * profile-count.c (profile_probability::differs_from_p): Do not
4265 rely on max_probaiblity == 10000
4267 2017-10-12 Jeff Law <law@redhat.com>
4269 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4272 2017-10-12 Martin Sebor <msebor@redhat.com>
4276 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4277 (handle_alias_pairs): Call it.
4278 * common.opt (-Wattribute-alias): New option.
4279 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4280 * doc/invoke.texi (-Wattribute-alias): Document.
4282 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4285 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4287 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4289 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4290 (make_hard_regno_born, make_hard_regno_dead): Update
4291 bb_killed_pseudos and bb_gen_pseudos.
4293 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4295 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4297 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4299 * config/alpha/alpha.c (alpha_split_conditional_move):
4300 Use std::swap instead of manually swapping.
4301 (alpha_stdarg_optimize_hook): Ditto.
4302 (alpha_canonicalize_comparison): Ditto.
4304 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4306 * tree-loop-distribution.c (struct builtin_info): New struct.
4307 (struct partition): Refactor fields into struct builtin_info.
4308 (partition_free): Free struct builtin_info.
4309 (build_size_arg_loc, build_addr_arg_loc): Delete.
4310 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4311 information from struct builtin_info.
4312 (find_single_drs): New function refactored from classify_partition.
4313 Also moved builtin validity checks to this function.
4314 (compute_access_range, alloc_builtin): New functions.
4315 (classify_builtin_st, classify_builtin_ldst): New functions.
4316 (classify_partition): Refactor code into functions find_single_drs,
4317 classify_builtin_st and classify_builtin_ldst.
4318 (distribute_loop): Don't do runtime alias check when distributing
4320 (find_seed_stmts_for_distribution): New function.
4321 (pass_loop_distribution::execute): Refactor code finding seed
4322 stmts into above function. Support distribution for the innermost
4323 two-level loop nest. Adjust dump information.
4325 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4327 * tree-loop-distribution.c: Adjust the general comment.
4328 (NUM_PARTITION_THRESHOLD): New macro.
4329 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4330 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4331 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4332 in call to build_partition_graph.
4333 (finalize_partitions): New parameter. Make loop distribution more
4334 conservative by fusing more partitions.
4335 (distribute_loop): Don't do runtime alias check in case of loop nest
4337 (find_seed_stmts_for_distribution): New function.
4338 (prepare_perfect_loop_nest): New function.
4339 (pass_loop_distribution::execute): Refactor code finding seed stmts
4340 and loop nest into above functions. Support loop nest distribution.
4341 Adjust dump information accordingly.
4343 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4345 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4346 and set PTYPE_SEQUENTIAL for merged partition.
4348 2017-10-12 Richard Biener <rguenther@suse.de>
4350 PR tree-optimization/69728
4352 2017-09-19 Richard Biener <rguenther@suse.de>
4354 PR tree-optimization/69728
4355 * graphite-sese-to-poly.c (schedule_error): New global.
4356 (add_loop_schedule): Handle empty domain by failing the
4358 (build_original_schedule): Handle schedule_error.
4360 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4361 domain by returning an unchanged schedule.
4363 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4365 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4366 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4368 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4370 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4373 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4376 * expr.c (store_field): Don't optimize away bitsize == 0 store
4377 from CALL_EXPR with addressable return type.
4379 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4381 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4382 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4383 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4384 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4385 TARGET_ISEL instead of TARGET_ISEL<sel>.
4387 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4389 * config/rs6000/rs6000.c
4390 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4392 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4394 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4396 (reg_or_cint_operand): Fix comment.
4397 (reg_or_zero_operand): New predicate.
4398 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4399 * config/rs6000/rs6000.c (output_isel): Delete.
4400 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4401 instead of reg_or_cint_operand. Output instruction directly (not via
4403 (isel_unsigned_<mode>): Ditto.
4404 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4405 gpc_reg_operand. Add an instruction alternative for this. Output
4406 instruction directly.
4407 (*isel_reversed_unsigned_<mode>): Ditto.
4409 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4411 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4412 (TARGET_CANONICALIZE_COMPARISON): Define.
4414 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4417 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4418 Check whether the dest is REG before adding REG_EQUIV note.
4420 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4423 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4425 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4426 (make_hard_regno_born, make_hard_regno_dead): Update
4427 bb_killed_pseudos and bb_gen_pseudos.
4429 2017-10-11 Nathan Sidwell <nathan@acm.org>
4431 * incpath.h (enum incpath_kind): Name enum, prefix values.
4432 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4433 * incpath.c (heads, tails): Use INC_MAX.
4434 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4435 (merge_include_chains, split_quote_chain,
4436 register_include_chains): Update incpath_kind names.
4437 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4438 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4440 (add_framework_path, darwin_register_objc_includes): Likewise.
4441 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4443 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4445 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4446 Do not use float_operator operator predicate.
4447 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4448 * config/i386/predicates.md (float_operator): Remove predicate.
4450 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4452 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4453 (*jccxf_i387): Ditto.
4454 (*jcc<mode>_i387): Ditto.
4455 (*jccu<mode>_i387): Ditto.
4456 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4457 (*jcc_*_i387 splitters): Remove.
4458 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4459 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4460 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4463 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4465 * profile-count.h (slow_safe_scale_64bit): New function.
4466 (safe_scale_64bit): New inline.
4467 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4468 * profile-count.c: Include wide-int.h
4469 (slow_safe_scale_64bit): New.
4471 2017-10-11 Nathan Sidwell <nathan@acm.org>
4473 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4474 HAS_DECL_ASSEMBLER_NAME_P.
4475 * gimple-expr.c (gimple_decl_printable_name: Check
4476 HAS_DECL_ASSEMBLER_NAME_P too.
4477 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4478 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4479 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4480 * passes.c (rest_of_decl_compilation): Check
4481 HAS_DECL_ASSEMBLER_NAME_P too.
4482 * recog.c (verify_changes): Likewise.
4483 * tree-pretty-print.c (dump_decl_name): Likewise.
4484 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4486 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4487 (SET_DECL_ASSEMBLER_NAME): Use it.
4488 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4489 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4490 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4492 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4494 * config.gcc (i386, x86_64): Add extra objects.
4495 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4496 (ix86_min_insn_size): Declare.
4497 (ix86_issue_rate): Declare.
4498 (ix86_adjust_cost): Declare.
4499 (ia32_multipass_dfa_lookahead): Declare.
4500 (ix86_macro_fusion_p): Declare.
4501 (ix86_macro_fusion_pair_p): Declare.
4502 (ix86_bd_has_dispatch): Declare.
4503 (ix86_bd_do_dispatch): Declare.
4504 (ix86_core2i7_init_hooks): Declare.
4505 (ix86_atom_sched_reorder): Declare.
4506 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4507 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4508 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4509 (rip_relative_addr_p): Rename to ...
4510 (ix86_rip_relative_addr_p): ... this one; export.
4511 (memory_address_length): Update.
4512 (ix86_issue_rate): Move to x86-tune-sched.c.
4513 (ix86_flags_dependent): Move to x86-tune-sched.c.
4514 (ix86_agi_dependent): Move to x86-tune-sched.c.
4515 (exact_dependency_1): Move to x86-tune-sched.c.
4516 (exact_store_load_dependency): Move to x86-tune-sched.c.
4517 (ix86_adjust_cost): Move to x86-tune-sched.c.
4518 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4519 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4520 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4521 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4522 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4523 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4524 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4525 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4526 (min_insn_size): Rename to ...
4527 (ix86_min_insn_size): ... this one; export.
4528 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4529 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4530 (core2i7_first_cycle_multipass_backtrack): Move to
4531 x86-tune-sched-core.c.
4532 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4533 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4534 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4535 (ix86_avoid_jump_mispredicts): Update.
4536 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4537 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4538 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4539 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4540 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4541 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4542 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4543 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4544 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4545 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4546 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4547 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4548 (BIG): Move to ix86-tune-sched-bd.c.
4549 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4550 (enum insn_path): Move to ix86-tune-sched-bd.c.
4551 (get_mem_group): Move to ix86-tune-sched-bd.c.
4552 (is_cmp): Move to ix86-tune-sched-bd.c.
4553 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4554 (is_branch): Move to ix86-tune-sched-bd.c.
4555 (is_prefetch): Move to ix86-tune-sched-bd.c.
4556 (init_window): Move to ix86-tune-sched-bd.c.
4557 (allocate_window): Move to ix86-tune-sched-bd.c.
4558 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4559 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4560 (process_end_window): Move to ix86-tune-sched-bd.c.
4561 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4562 (find_constant): Move to ix86-tune-sched-bd.c.
4563 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4564 (has_immediate): Move to ix86-tune-sched-bd.c.
4565 (get_insn_path): Move to ix86-tune-sched-bd.c.
4566 (get_insn_group): Move to ix86-tune-sched-bd.c.
4567 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4568 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4569 (add_insn_window): Move to ix86-tune-sched-bd.c.
4570 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4571 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4572 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4573 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4574 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4575 (do_dispatch): Move to ix86-tune-sched-bd.c.
4576 (has_dispatch): Move to ix86-tune-sched-bd.c.
4577 * config/i386/t-i386: Add new object files.
4578 * config/i386/x86-tune-costs.h: New file.
4579 * config/i386/x86-tune-sched-atom.c: New file.
4580 * config/i386/x86-tune-sched-bd.c: New file.
4581 * config/i386/x86-tune-sched-core.c: New file.
4582 * config/i386/x86-tune-sched.c: New file.
4584 2017-10-11 Liu Hao <lh_mouse@126.com>
4586 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4587 the generic version below instead.
4588 (should_colorize): Recognize Windows consoles as terminals
4590 * pretty-print.c [__MINGW32__] (write_all): New function.
4591 [__MINGW32__] (find_esc_head): Likewise.
4592 [__MINGW32__] (find_esc_terminator): Likewise.
4593 [__MINGW32__] (eat_esc_sequence): Likewise.
4594 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4596 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
4599 2017-10-11 Richard Biener <rguenther@suse.de>
4601 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
4602 Properly call analyze_scalar_evolution with the loop of the stmt.
4604 2017-10-11 Richard Biener <rguenther@suse.de>
4606 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
4607 * tree-core.h (tree_base): Add chrec_var union member.
4608 * tree.h (CHREC_VAR): Remove.
4609 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
4610 * tree-chrec.h (build_polynomial_chrec): Adjust.
4611 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
4612 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
4614 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4616 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4617 * match.pd: ... here.
4618 ((T) X == (T) Y): Relax condition.
4620 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4622 PR tree-optimization/82472
4623 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4625 (break_alias_scc_partitions): Update postorder number.
4627 2017-10-11 Martin Liska <mliska@suse.cz>
4630 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4632 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4634 2017-10-11 Martin Liska <mliska@suse.cz>
4636 * print-rtl.c (print_insn): Move declaration of idbuf
4637 to same scope as name.
4639 2017-10-11 Martin Liska <mliska@suse.cz>
4644 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4647 2017-10-11 Martin Liska <mliska@suse.cz>
4650 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4653 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4655 * config.gcc (default_gnu_indirect_function): Default to yes for
4656 arm*-*-linux* with glibc.
4658 2017-10-11 Richard Biener <rguenther@suse.de>
4660 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4661 default-defs and types we do not want to analyze.
4662 (interpret_loop_phi): Replace unreachable code with an assert.
4663 (compute_scalar_evolution_in_loop): Remove and inline ...
4664 (analyze_scalar_evolution_1): ... here, replacing condition with
4665 what makes the intent clearer. Remove handling of cases
4666 get_scalar_evolution now handles.
4668 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4670 PR rtl-optimization/81434
4671 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4672 comment for main loop. In sched_group_found if, also add checks for
4673 pass and min_cost_group.
4675 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4677 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4678 (rs6000_insn_cost): New function.
4679 * config/rs6000/rs6000.md (cost): New attribute.
4681 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4682 H.J. Lu <hongjiu.lu@intel.com>
4686 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4687 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4688 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4689 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4690 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4691 OPTION_MASK_ISA_MMX and some other bit set require both
4692 mmx and the ISAs without the mmx bit.
4693 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4694 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4695 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4696 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4697 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4698 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4699 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4700 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4701 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4702 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4703 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4704 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4705 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4706 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4707 Add OPTION_MASK_ISA_MMX.
4709 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4711 * config.gcc (armv7*-*-freebsd*): New target.
4712 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4714 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4716 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4717 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4720 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4722 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4723 a template parameter.
4724 (WIDE_INT_REF_FOR): Update accordingly.
4725 * tree.h (wi::int_traits <const_tree>): Delete.
4726 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4727 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4728 (wi::tree_to_wide_ref): New typedef.
4729 (wi::to_wide): New function.
4730 * calls.c (get_size_range): Use wi::to_wide when operating on
4732 * cgraph.c (cgraph_node::create_thunk): Likewise.
4733 * config/i386/i386.c (ix86_data_alignment): Likewise.
4734 (ix86_local_alignment): Likewise.
4735 * dbxout.c (stabstr_O): Likewise.
4736 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4737 * expr.c (const_vector_from_tree): Likewise.
4738 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4739 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4740 (fold_negate_expr_1, int_const_binop_1, const_binop)
4741 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4742 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4743 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4744 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4745 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4746 (fold_not_const, round_up_loc): Likewise.
4747 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4748 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4749 (alloca_call_type): Likewise.
4750 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4751 * godump.c (go_output_typedef): Likewise.
4752 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4753 * internal-fn.c (get_min_precision): Likewise.
4754 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4755 * ipa-polymorphic-call.c
4756 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4757 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4758 (ipa_modify_call_arguments): Likewise.
4759 * match.pd: Likewise.
4760 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4761 * print-tree.c (print_node_brief, print_node): Likewise.
4762 * stmt.c (expand_case): Likewise.
4763 * stor-layout.c (layout_type): Likewise.
4764 * tree-affine.c (tree_to_aff_combination): Likewise.
4765 * tree-cfg.c (group_case_labels_stmt): Likewise.
4766 * tree-data-ref.c (dr_analyze_indices): Likewise.
4767 (prune_runtime_alias_test_list): Likewise.
4768 * tree-dump.c (dequeue_and_dump): Likewise.
4769 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4770 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4771 * tree-pretty-print.c (dump_generic_node): Likewise.
4772 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4773 (simple_iv_with_niters): Likewise.
4774 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4775 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4776 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4777 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4778 (refine_value_range_using_guard, number_of_iterations_ne_max)
4779 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4780 (get_cst_init_from_scev, record_nonwrapping_iv)
4781 (scev_var_range_cant_overflow): Likewise.
4782 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4783 * tree-ssa-pre.c (compute_avail): Likewise.
4784 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4785 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4786 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4787 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4788 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4789 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4790 (dump_case_nodes, try_switch_expansion): Likewise.
4791 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4792 (vect_do_peeling): Likewise.
4793 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4794 * tree-vect-stmts.c (vectorizable_load): Likewise.
4795 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4796 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4797 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4798 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4799 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4800 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4801 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4802 (evrp_dom_walker::before_dom_children): Likewise.
4803 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4804 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4805 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4806 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4807 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4808 (get_range_pos_neg): Likewise.
4809 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4810 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4811 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4812 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4813 * config/avr/avr.c (avr_fold_builtin): Likewise.
4814 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4815 * config/msp430/msp430.c (msp430_attr): Likewise.
4816 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4817 * config/powerpcspe/powerpcspe-c.c
4818 (altivec_resolve_overloaded_builtin): Likewise.
4819 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4820 (rs6000_expand_ternop_builtin): Likewise.
4821 * config/rs6000/rs6000-c.c
4822 (altivec_resolve_overloaded_builtin): Likewise.
4823 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4824 (rs6000_expand_ternop_builtin): Likewise.
4825 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4827 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4829 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4830 when copying loop nest with only one inner loop.
4832 2017-10-10 Richard Biener <rguenther@suse.de>
4834 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4835 blocks if SCEV is active.
4836 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4838 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4839 (scev_initialize): Assert we are not yet initialized.
4841 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4843 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4844 inner loop's exit stmt by making it always exit the loop, otherwise
4845 we would generate an infinite empty loop.
4847 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4849 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4850 renaming variables in new preheader if it's deleted.
4852 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4854 * tree-loop-distribution.c (struct partition): Remove unused field
4855 loops of the structure.
4856 (partition_alloc, partition_free): Ditto.
4857 (build_rdg_partition_for_vertex): Ditto.
4859 2017-10-09 Jeff Law <law@redhat.com>
4861 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4862 return type to match prototype and documentation.
4864 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4866 * config/rs6000/rs6000.c (processor_costs): Move to ...
4867 * config/rs6000/rs6000.h: ... here.
4868 (rs6000_cost): Declare.
4870 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4872 * except.c (setjmp_fn): New global variable.
4873 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4874 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4875 if DONT_USE_BUILTIN_SETJMP is defined.
4877 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4879 * target.def (insn_cost): New hook.
4880 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4881 * doc/tm.texi: Regenerate.
4882 * rtlanal.c (insn_cost): Use the new hook.
4884 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4886 * combine.c (combine_validate_cost): Compute the new insn_cost,
4887 not just pattern_cost.
4888 (try_combine): Adjust comment.
4890 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4892 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4894 * combine.c (uid_insn_cost): Adjust comment.
4895 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4897 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4898 * dse.c (find_shift_sequence): Ditto.
4899 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4900 (bb_valid_for_noce_process_p): Use pattern_cost.
4901 * rtl.h (insn_rtx_cost): Delete.
4902 (pattern_cost): New prototype.
4903 (insn_cost): New prototype.
4904 * rtlanal.c (insn_rtx_cost): Rename to...
4905 (pattern_cost): ... this.
4908 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4910 * config/i386/i386.md (*jcc_2): Remove insn pattern.
4911 (*jcc<mode>_0_r_i387): Ditto.
4912 (*jccxf_r_i387): Ditto.
4913 (*jcc<mode>_r_i387): Ditto.
4914 (*jccu<mode>_r_i387): Ditto.
4915 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
4916 (*jcc): Rename from *jcc_1.
4918 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4920 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
4921 deferred rescans after the lvx/stvx recombination pre-pass.
4923 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
4925 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
4926 memory operation instruction support.
4927 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
4928 (rs6000-ibm-aix[789]*): Likewise.
4929 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
4930 Document new functions.
4932 2017-10-09 Richard Biener <rguenther@suse.de>
4934 PR tree-optimization/82397
4935 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
4936 equality only for semantically equal trees.
4938 2017-10-09 Richard Biener <rguenther@suse.de>
4940 PR tree-optimization/82449
4941 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
4942 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
4943 allow constant addresses.
4944 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
4947 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4949 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
4952 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4955 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
4958 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4961 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
4963 2017-10-09 Jakub Jelinek <jakub@redhat.com>
4966 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
4967 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
4969 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
4971 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
4972 (WI_BINARY_PREDICATE_RESULT): Likewise.
4973 (wi::binary_traits::operator_result): New type.
4974 (wi::binary_traits::predicate_result): Likewise.
4975 (generic_wide_int::operator~, unary generic_wide_int::operator-)
4976 (generic_wide_int::operator==, generic_wide_int::operator!=)
4977 (generic_wide_int::operator&, generic_wide_int::and_not)
4978 (generic_wide_int::operator|, generic_wide_int::or_not)
4979 (generic_wide_int::operator^, generic_wide_int::operator+
4980 (binary generic_wide_int::operator-, generic_wide_int::operator*):
4982 (operator~, unary operator-, operator==, operator!=, operator&)
4983 (operator|, operator^, operator+, binary operator-, operator*): New
4985 * expr.c (get_inner_reference): Use wi::bit_and_not.
4986 * fold-const.c (fold_binary_loc): Likewise.
4987 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
4988 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
4989 (bit_value_binop): Likewise.
4990 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
4991 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
4992 (extract_range_from_binary_expr_1): Likewise.
4993 (masked_increment): Likewise.
4994 (simplify_bit_ops_using_ranges): Likewise.
4996 2017-10-09 Martin Jambor <mjambor@suse.cz>
4999 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5000 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5001 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5003 (hsa_fixup_mov_insn_type): New function.
5004 (hsa_op_with_type::get_in_type): Use it.
5005 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5006 immediates in an assert.
5007 (hsa_op_with_type::extend_int_to_32bit): New method.
5008 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5009 types. Convert to dest type if necessary.
5010 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5011 (reg_for_gimple_ssa): Pass false as min32int to
5012 hsa_type_for_scalar_tree_type.
5013 (gen_hsa_addr): Fixup type when creating addresable temporary.
5014 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5015 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5016 necessary. Call hsa_fixup_mov_insn_type.
5017 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5018 extend operands and convert to dest type if necessary.
5019 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5020 to dest type if necessary.
5021 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5022 if conversion nt necessary and size matches.
5023 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5024 to dest type if necessary.
5025 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5026 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5028 (gen_hsa_clrsb): Likewise.
5029 (gen_hsa_ffs): Likewise.
5030 (gen_hsa_divmod): Extend operands and convert to dest type if
5032 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5034 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5036 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5037 Remove empty default arguments. Use a brace block as output
5039 (conditional return): Ditto.
5041 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5042 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5044 (group_ending_nop): Ditto.
5045 (doloop_end): Ditto.
5046 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5047 (splitters for those): Ditto.
5049 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5051 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5052 a conditional jump (and the compare for it) so that pc_rtx is the
5054 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5055 for the deleted and renamed ctr<mode>_internal[234] patterns.
5056 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5057 Delete second conditional return pattern.
5058 (ctr<mode>_internal2): Delete this second bdnz pattern.
5059 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5060 (ctr<mode>_internal4): Delete this second bdz pattern.
5062 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5064 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5065 (always_initialized_rtx_for_ssa_name_p): New predicate.
5066 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5067 (finish_out_of_ssa): Free new field of SA.
5068 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5069 * tree-ssa-coalesce.c: Include tree-ssa.h.
5070 (get_parm_default_def_partitions): Remove extern keyword.
5071 (get_undefined_value_partitions): New function.
5072 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5073 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5075 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5077 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5079 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5081 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5083 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5084 for targets that preffer 128bit.
5086 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5088 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5090 2017-10-08 Olivier Hainque <hainque@adacore.com>
5092 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5093 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5095 (thumb_set_return_address): Likewise.
5097 2017-10-08 Olivier Hainque <hainque@adacore.com>
5099 * common/config/arm/arm-common.c (arm_except_unwind_info):
5100 Handle DWARF2_UNWIND_INFO.
5102 2017-10-07 Michael Collison <michael.collison@arm.com>
5104 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
5107 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
5109 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
5110 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
5111 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
5112 defined, force the creation of a new block for a dispatch label.
5114 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
5116 * invoke.texi (Wsuggest-attribute=cold): Document.
5117 * common.opt (Wsuggest-attribute=cold): New
5118 * ipa-pure-const.c (warn_function_cold): New function.
5119 * predict.c (compute_function_frequency): Use it.
5120 * predict.h (warn_function_cold): Declare.
5122 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
5124 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
5127 2017-10-06 Martin Liska <mliska@suse.cz>
5129 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
5130 keyword for member functions.
5131 (struct sanopt_tree_couple): New struct.
5132 (struct sanopt_tree_couple_hash): New function.
5133 (struct sanopt_ctx): Add new hash_map.
5134 (has_dominating_ubsan_ptr_check): New function.
5135 (record_ubsan_ptr_check_stmt): Likewise.
5136 (maybe_optimize_ubsan_ptr_ifn): Likewise.
5137 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
5138 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
5140 2017-10-06 Sudakshina Das <sudi.das@arm.com>
5143 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
5144 aarch64_simd_valid_immediate on CONST_VECTORs.
5145 (aarch64_reg_or_bic_imm): Likewise.
5147 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
5149 PR rtl-optimization/82396
5150 * haifa-sched.c (ready_sort_real): Disable qsort checking.
5152 2017-10-06 Sebastian Pop <sebpop@gmail.com>
5154 * graphite-dependences.c (scop_get_reads): Move code to...
5155 (scop_get_must_writes): Move code to...
5156 (scop_get_may_writes): Move code to...
5157 (scop_get_reads_and_writes): ... here.
5158 (scop_get_dependences): Call scop_get_reads_and_writes.
5160 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5162 PR tree-optimization/82434
5163 * fold-const.h (can_native_encode_type_p,
5164 can_native_encode_string_p): Remove.
5165 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5166 don't encode anything, just return what would be otherwise returned.
5167 (native_encode_fixed, native_encode_complex, native_encode_vector):
5169 (native_encode_string): Likewise. Inline by hand
5170 can_native_encode_string_p.
5171 (can_native_encode_type_p): Remove.
5172 (can_native_encode_string_p): Remove.
5173 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5174 STRING_CSTs using can_native_encode_string_p, test all
5175 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5176 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5177 argument from native_encode_expr.
5178 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5179 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5182 2017-10-06 Richard Biener <rguenther@suse.de>
5184 PR tree-optimization/82397
5185 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5186 operand_equal_p but rely on data_ref_compare_tree for detecting
5188 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5189 to match up with dr_group_sort_cmp.
5191 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5194 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5196 * config/s390/s390-builtin-types.def: Regenerate.
5198 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5201 * config/s390/s390-builtin-types.def: Regenerate.
5202 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5203 Change flag from B_VXE to B_VX.
5204 (s390_vec_min_dbl): Remove B_VXE flag.
5206 2017-10-06 Richard Biener <rguenther@suse.de>
5208 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5209 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5210 translate_isl_ast_to_gimple::is_valid_rename,
5211 translate_isl_ast_to_gimple::get_rename,
5212 translate_isl_ast_to_gimple::get_def_bb_for_const,
5213 translate_isl_ast_to_gimple::get_new_name,
5214 translate_isl_ast_to_gimple::collect_all_ssa_names,
5215 translate_isl_ast_to_gimple::copy_loop_phi_args,
5216 translate_isl_ast_to_gimple::collect_all_ssa_names,
5217 translate_isl_ast_to_gimple::copy_loop_phi_args,
5218 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5219 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5220 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5221 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5222 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5223 translate_isl_ast_to_gimple::copy_cond_phi_args,
5224 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5225 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5226 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5227 translate_isl_ast_to_gimple::rename_uses,
5228 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5229 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5230 (set_rename_for_each_def): Likewise.
5231 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5232 here. Handle rewriting SCEV analyzable uses here.
5233 (copy_bb_and_scalar_dependences): Generate code for PHI
5235 (graphite_regenerate_ast_isl): Adjust.
5236 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5237 (add_write, add_read): New functions.
5238 (build_cross_bb_scalars_def): Use it and simplify.
5239 (build_cross_bb_scalars_use): Likewise.
5240 (graphite_find_cross_bb_scalar_vars): Inline into...
5241 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5242 simulating out-of-SSA. Compute liveout and add dependencies.
5243 (build_scops): Force an empty entry block.
5244 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5246 (sese_build_liveouts): Declare.
5247 (sese_trivially_empty_bb_p): Likewise.
5248 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5249 compute liveout and debug_liveout.
5250 (sese_bad_liveouts_use): Remove.
5251 (sese_reset_debug_liveouts_bb): Likewise.
5252 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5253 (sese_build_liveouts): Build liveout and debug_liveout and store
5255 (new_sese_info): Adjust.
5256 (free_sese_info): Likewise.
5257 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5258 do not build liveout here.
5259 (move_sese_in_condition): Adjust region entry.
5260 (scev_analyzable_p): Match up with chrec_apply requirements.
5261 (sese_trivially_empty_bb_p): New.
5262 * tree-into-ssa.c (get_reaching_def): Properly support generating
5263 default-defs for incremental rewrite of anonymous names.
5265 2017-10-06 Richard Biener <rguenther@suse.de>
5267 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5268 precision do not perform modulo reduction.
5270 2017-10-06 Richard Biener <rguenther@suse.de>
5272 PR tree-optimization/82436
5273 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5274 conservatively choose the vectorization factor when checking
5275 whether we can perform the required load permutation.
5276 (vect_transform_slp_perm_load): Assert when we may not fail.
5278 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5280 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5281 message for incompatible -msdata=* and -mcall-* options.
5283 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5285 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5286 rate for post-reload scheduling.
5288 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5290 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5292 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5294 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5295 to improve monte carlo in scimark.
5297 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5299 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5300 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5301 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5302 pentium4_cost, nocona_cost): Set reassociation width to 1.
5303 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5304 width to 2 for fp operations and 1 otherwise.
5305 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5307 (atom_cost): Set reassociation width to 2.
5308 (slm_cost, generic_cost): Set fp reassociation width
5309 to 2 and 1 otherwise.
5310 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5311 (core_cost): Set fp reassociation width to 4 and vector to 2.
5312 (ix86_reassociation_width): Rewrite using cost table; special case
5313 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5314 and TARGET_AVX128_OPTIMAL.
5315 * config/i386/i386.h (processor_costs): Add
5316 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5317 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5318 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5319 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5320 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5321 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5323 2017-10-05 Nathan Sidwell <nathan@acm.org>
5325 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5327 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5329 * config/arm/arm.c (arm_test_fpu_data): New.
5330 (arm_run_selftests): Call arm_test_fpu_data.
5332 2017-10-04 Nathan Sidwell <nathan@acm.org>
5334 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5336 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5338 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5340 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5341 check type for aarch64_simd_valid_immediate.
5342 (aarch64_output_simd_mov_immediate): Update prototype.
5343 (aarch64_simd_valid_immediate): Update prototype.
5344 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5345 support for ORR-immediate.
5346 (and<mode>3): modified pattern to add support for BIC-immediate.
5347 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5348 now checks for valid immediate for BIC and ORR based on new enum
5350 (aarch64_output_simd_mov_immediate): Function now used to output
5351 BIC/ORR imm as well based on new enum argument.
5352 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5354 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5355 (aarch64_reg_or_bic_imm): Likewise.
5357 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5359 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5360 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5362 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5366 PR rtl-optimization/82396
5367 * haifa-sched.c (autopref_multipass_init): Simplify
5369 (autopref_rank_data): Simplify sort order.
5370 * sched-int.h (autopref_multipass_data_): Remove
5371 multi_mem_insn_p, min_offset and max_offset.
5373 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5375 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5377 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5379 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5382 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5384 * doc/sourcebuild.texi: Document vect_long_mult.
5386 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5388 PR tree-optimization/82413
5389 * fold-const.c (build_range_check): Use widest_int when comparing
5390 the maximum ETYPE value with HIGH.
5392 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5394 PR rtl-optimization/82396
5395 * haifa-sched.c (autopref_multipass_init): Simplify
5397 (autopref_rank_data): Simplify sort order.
5398 * sched-int.h (autopref_multipass_data_): Remove
5399 multi_mem_insn_p, min_offset and max_offset.
5401 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5403 PR tree-optimization/82381
5404 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5405 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5408 PR tree-optimization/82374
5409 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5410 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5411 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5412 current_function_decl to the new decl.
5414 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5416 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5417 helper macro for IEEE float128 hardware built-in functions.
5418 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5420 (TRUNCF128_ODD): Likewise.
5421 (ADDF128_ODD): Likewise.
5422 (SUBF128_ODD): Likewise.
5423 (MULF128_ODD): Likewise.
5424 (DIVF128_ODD): Likewise.
5425 (FMAF128_ODD): Likewise.
5426 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5427 UNSPEC_TRUNC_ROUND_TO_ODD.
5428 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5429 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5430 floating point round to odd instructions.
5431 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5432 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5433 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5434 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5435 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5436 (trunc<mode>sf2_hw): Change the truncate with round to odd
5437 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5438 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5439 to odd hardware instructions.
5440 (sub<mode>3_odd): Likewise.
5441 (mul<mode>3_odd): Likewise.
5442 (div<mode>3_odd): Likewise.
5443 (sqrt<mode>2_odd): Likewise.
5444 (fma<mode>4_odd): Likewise.
5445 (fms<mode>4_odd): Likewise.
5446 (nfma<mode>4_odd): Likewise.
5447 (nfms<mode>4_odd): Likewise.
5448 (trunc<mode>df2_odd): Change the truncate with round to odd
5449 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5451 * doc/extend.texi (PowerPC built-in functions): Update documentation
5452 for existing IEEE float128-bit built-in functions. Add built-in
5453 functions that generate the IEEE 128-bit floating point round to
5456 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5458 PR rtl-optimization/77729
5459 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5460 to (X&(C1&~C2))|C2 transformations.
5462 2017-10-03 Martin Jambor <mjambor@suse.cz>
5464 PR tree-optimization/82363
5465 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5466 mismatch, mark lacc written regardless of racc.
5468 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5470 PR tree-optimization/82381
5471 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5472 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5473 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5474 fallthrough into reassoc_stmt_dominates_stmt_p.
5477 * combine.c (combine_instructions): Don't combine in unreachable
5480 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5483 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5484 function to not use the have_cpu variable. Do not set cpu_index,
5485 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5487 (rs6000_valid_attribute_p): Remove duplicate initializations of
5488 old_optimize and func_optimize.
5489 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5490 (rs6000_activate_target_options): Make global.
5491 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5494 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5496 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5497 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5498 Set *poffset to 0 and *psize and *pmax_size to -1 if
5499 *poffset + *psize overflows in HOST_WIDE_INT.
5501 PR tree-optimization/82387
5502 PR tree-optimization/82388
5503 PR tree-optimization/82389
5504 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5505 instead of live_bytes non-NULL.
5507 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5510 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5513 2017-10-02 Richard Biener <rguenther@suse.de>
5515 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5516 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5517 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5519 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5521 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5522 requested precision matches the type's.
5523 * calls.c (alloc_max_size): Calculate the new candidate size as
5524 a widest_int and use wi::to_widest when comparing it with the
5525 current candidate size.
5526 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5527 zero rather than integer_zero_node.
5528 * match.pd: Check for a no-op conversion before using wi::add
5529 rather than after. Use tree_to_uhwi when summing small shift
5530 counts into an unsigned int.
5532 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5533 Alan Hayward <alan.hayward@arm.com>
5534 David Sherwood <david.sherwood@arm.com>
5537 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5538 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5539 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5540 POINTER_AND_FP_REGS.
5542 2017-10-02 Richard Biener <rguenther@suse.de>
5544 PR tree-optimization/82355
5545 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5546 a mapping for the enclosing loop but avoid generating one for
5548 (copy_bb_and_scalar_dependences): Remove premature codegen
5549 error on PHIs in blocks duplicated into multiple places.
5550 * graphite-scop-detection.c
5551 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5552 in the region use it as loop and nest to analyze the DR in.
5553 (try_generate_gimple_bb): Likewise.
5554 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5555 (add_loop_constraints): For blocks in a loop not in the region
5556 create a dimension with a single iteration.
5557 * sese.h (gbb_loop_at_index): Remove assert.
5559 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5561 * omp-expand.c (adjust_context_scope): New function.
5562 (expand_parallel_call): Call adjust_context_scope.
5564 2017-10-01 Jeff Law <law@redhat.com>
5566 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5567 dom_opt_dom_walker class with direct access to private members.
5568 Add comments. Call test_for_singularity.
5569 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5570 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5571 m_dummy_cond anymore.
5572 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5574 (pass_dominator:execute): Build the dummy_cond here and pass it
5575 to the dom_opt_dom_walker ctor.
5576 (test_for_singularity): New function.
5578 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5579 Maya Rashish <coypu@sdf.org>
5581 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5582 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5583 (alpha*-*-netbsd*) Use nbsd_tm_file.
5584 (arm*-*-netbsdelf*) Likewise.
5585 (i[34567]86-*-netbsdelf*) Likewise.
5586 (x86_64-*-netbsd*) Likewise.
5587 (mips*-*-netbsd*) Likewise.
5588 (powerpc-*-netbsd*) Likewise.
5589 (sh*-*-netbsd*) Likewise.
5590 (sparc-*-netbsdelf*) Likewise.
5591 (sparc64-*-netbsd*) Likewise.
5592 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5594 (vax-*-netbsdelf*) Likewise.
5595 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5596 (UINT_FAST8_TYPE) Likewise.
5597 (INT_FAST16_TYPE) Check CHAR_FAST16.
5598 (UINT_FAST16_TYPE) Likewise.
5600 2017-09-30 Jakub Jelinek <jakub@redhat.com>
5603 * config/i386/i386.md
5604 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
5605 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
5606 *divmodsi4_zext_2): New define_insn_and_split.
5607 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
5608 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
5609 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
5610 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
5611 New define_insn_and_split.
5612 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
5613 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
5614 operands[1] having DImode when mode is SImode.
5616 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5617 always SImode for DIV and MOD in REG_EQUAL notes.
5619 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5622 * match.pd: Fix handling of NaNs in pattern.
5624 2017-09-29 Jeff Law <law@redhat.com>
5626 * sbitmap.c (bitmap_bit_in_range_p): New function.
5627 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5628 * tree-ssa-dse.c (live_bytes_read): New function.
5629 (dse_classify_store): Ignore reads of dead bytes.
5631 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5632 typos and whitespace errors.
5633 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5634 * config/s390/s390.c (s390_emit_prologue): Likewise.
5636 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5639 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5640 with a symbol for LRA.
5642 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5644 PR rtl-optimization/82338
5645 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5647 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5649 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5650 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5651 (qsort_chk): Declare.
5652 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5653 (qsort_chk): New function.
5655 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5657 PR tree-optimization/82337
5658 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5659 phi definition if the PHI result appears in an abnormal PHI.
5660 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5661 basis appears in an abnormal PHI.
5663 2017-09-29 Richard Biener <rguenther@suse.de>
5665 * graphite-isl-ast-to-gimple.c
5666 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5667 (binary_op_to_tree): Use it.
5668 (get_rename_from_scev): Likewise.
5669 (copy_loop_phi_nodes): Likewise.
5670 (copy_bb_and_scalar_dependences): Likewise.
5671 (translate_pending_phi_nodes): Likewise.
5673 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5676 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5677 for movabsq $(i32 << shift), r64.
5679 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5681 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5682 index when encoding %esp as %rsp to avoid 0x67 prefix.
5684 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5686 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5687 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5689 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5691 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5692 Extensions with more than 16 double VFP registers.
5693 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5694 to_clear_mask and all code related to it. Replace the remaining
5695 entry by a sbitmap and adapt code accordingly.
5697 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5699 * brig-builtins.def: Change pure attributes to const.
5701 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5703 * config.gcc (default_gnu_indirect_function): Default to yes for
5704 sparc*-*-linux* with glibc.
5706 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5708 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5709 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5710 when creating .init_array and .fini_array sections with priority
5713 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5716 * config/aarch64/aarch64.c
5717 (aarch64_builtin_support_vector_misalignment): Always return false
5718 when misalignment is unknown.
5720 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5722 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5723 this function to return false if the definition used by the swap
5724 instruction is artificial, or if the memory address from which the
5725 constant value is loaded is not represented by a base address held
5726 in a register or if the base address register is a frame or stack
5727 pointer. Additionally, return false if the base address of the
5728 loaded constant is a SYMBOL_REF but is not considered to be a
5730 (replace_swapped_load_constant): New function.
5731 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5732 loaded constant vector with a load of a swapped constant vector.
5734 2017-09-27 Carl Love <cel@us.ibm.com>
5736 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5737 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5738 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5741 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5743 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5744 first, always call autopref_rank_data otherwise.
5746 2017-09-27 Richard Biener <rguenther@suse.de>
5748 * graphite-scop-detection.c (find_scop_parameters): Move
5749 loop bound handling ...
5750 (gather_bbs::before_dom_children): ... here, avoiding the need
5751 to build scop_info->loop_nest.
5752 (record_loop_in_sese): Remove.
5753 * sese.h (sese_info_t::loop_nest): Remove.
5754 * sese.c (new_sese_info): Do not allocate loop_nest.
5755 (free_sese_info): Do not free loop_nest.
5757 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5760 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5761 lhs from calls if the lhs has addressable type.
5763 2017-09-27 Richard Biener <rguenther@suse.de>
5765 * graphite.h (scop::max_alias_set): New member.
5766 * graphite-scop-detection.c: Remove references to non-existing
5767 --param in comments.
5768 (build_alias_sets): Record the maximum alias set used for drs.
5769 (build_scops): Support zero as unlimited for
5770 --param graphite-max-arrays-per-scop.
5771 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5773 (build_poly_sr_1): ... here. Compute alias set based on the
5774 maximum alias set used for drs rather than
5775 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5777 2017-09-27 Richard Biener <rguenther@suse.de>
5779 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5780 --param loop-block-tile-size=0 to disable tiling.
5782 2017-09-27 Richard Biener <rguenther@suse.de>
5784 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5785 (graphite-max-nb-scop-params): Document special value zero.
5786 * domwalk.h (dom_walker::STOP): New symbolical constant.
5787 (dom_walker::dom_walker): Add optional parameter for bb to
5789 (dom_walker::~dom_walker): Declare.
5790 (dom_walker::before_dom_children): Document STOP return value.
5791 (dom_walker::m_user_bb_to_rpo): New member.
5792 (dom_walker::m_bb_to_rpo): Likewise.
5793 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5794 mapping here if not provided by the user.
5795 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5796 provided by the user.
5797 (dom_walker::STOP): Define.
5798 (dom_walker::walk): Do not compute bb to RPO mapping here.
5799 Support STOP return value from before_dom_children to stop
5801 * graphite-optimize-isl.c (optimize_isl): If the schedule
5802 is the same still generate code if -fgraphite-identity
5803 or -floop-parallelize-all are given.
5804 * graphite-scop-detection.c: Include cfganal.h.
5805 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5807 (gather_bbs::before_dom_children): Return STOP for BBs
5809 (build_scops): Compute bb to RPO mapping and pass it to
5810 the domwalk. Treat --param graphite-max-nb-scop-params=0
5811 as not limiting the number of params.
5812 * graphite.c (graphite_initialize): Remove limit on the
5813 number of basic-blocks in a function.
5814 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5815 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5816 default value of 10.
5818 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5820 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5821 Adjust code to eliminate needing to do the shift right 32-bits
5822 operation after XSCVDPSPN.
5824 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5826 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5827 ((X / Y) != 0 -> X >= Y): Likewise.
5829 2017-09-26 Carl Love <cel@us.ibm.com>
5831 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5832 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5833 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5834 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5835 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5836 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5837 definitions and overloading.
5838 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5839 statement for P9V_BUILTIN_XST_LEN_R.
5840 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5841 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5842 define_expand and define_insn for the instructions and builtins.
5843 * doc/extend.texi: Update the built-in documentation file for the new
5845 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5846 define_insn for the instructions
5848 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5851 * gcc/config/netbsd-protos.h: New file.
5852 * gcc/config/netbsd.c: New file.
5853 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5854 * gcc/config/t-netbsd: New file.
5855 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5856 (tmake_file) Add t-netbsd.
5857 (extra_objs) Add netbsd.o.
5859 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5863 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5865 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5867 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5868 sign extension from a vector register to a GPR by doing a 32-bit
5869 direct move and then an EXTSW.
5870 (extendsi<mode>2 splitter): Likewise.
5871 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5872 right or vector extract after doing XSCVDPSPN. Use
5873 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5875 (movdi_from_sf_zero_ext): Likewise.
5876 (reload_gpr_from_vsxsf): Likewise.
5877 (p8_mfvsrd_4_disf): Delete, no longer used.
5878 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5879 then needing to move the SFmode to a GPR to use the XSCVDPSP
5880 instruction instead of FRSP and XSCVDPSPN.
5881 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5882 it is adjacent to the other XSCVSPDP insns.
5883 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5884 SFmode to be in traditional Altivec registers.
5885 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5886 (vsx_xscvspdpn): Likewise.
5887 (vsx_xscvspdpn_scalar): Likewise.
5889 2017-09-26 Martin Jambor <mjambor@suse.cz>
5891 * tree-sra.c (compare_access_positions): Put integral types first,
5892 stabilize sorting of integral types, remove conditions putting
5893 non-full-precision integers last.
5894 (sort_and_splice_var_accesses): Disable scalarization if a
5895 non-integert would be represented by a non-full-precision integer.
5897 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5899 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5900 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5901 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5902 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5903 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5904 conditionals inside the function instead of around it. Call
5905 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5906 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5908 2017-09-26 Richard Biener <rguenther@suse.de>
5910 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5912 (scop_detection::build_scop_breadth): ... this. Removed.
5913 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
5914 (scop_detection::harmful_stmt_in_bb): Likewise.
5915 (scop_detection::graphite_can_represent_stmt): Likewise.
5916 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
5917 (scop_detection::can_represent_loop): Remove recursion, fold in ...
5918 (scop_detection::can_represent_loop_1): ... this. Removed.
5919 (scop_detection::harmful_loop_in_region): Simplify after inlining
5920 the above and remove more quadraticness.
5921 (build_scops): Adjust.
5922 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
5925 2017-09-26 Jakub Jelinek <jakub@redhat.com>
5928 * config/i386/i386.c (ix86_print_operand_address_as): Only test
5929 REGNO (base) == SP_REG if base is a REG.
5932 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
5933 if it is different SSA_NAME.
5934 (optimize_range_tests_cmp_bitwise): New function.
5935 (optimize_range_tests): Call it.
5937 2017-09-26 Richard Biener <rguenther@suse.de>
5939 PR tree-optimization/82321
5940 * graphite.c (canonicalize_loop_closed_ssa): Properly check
5941 for the def being inside the loop.
5943 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5945 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
5947 * config/s390/s390-builtins.def: Fix constraint on op4.
5949 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5951 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
5952 independent expanders.
5953 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
5954 ("vec_ordered", "vec_unordered"): New expanders.
5956 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5958 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
5961 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5963 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
5964 vec_unpacks_lo_v16qi.
5965 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
5967 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5969 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
5970 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
5971 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
5973 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5975 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
5977 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
5979 ("*vec_slb<mode>"): New insn pattern.
5980 ("vec_shr_<mode>"): New expander.
5981 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
5982 and force the shift count operand to V16QImode.
5983 ("vec_srb<mode>"): Set shift count mode to V16QI.
5985 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5987 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
5988 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
5989 ("vec_widen_smult_hi_<mode>"): New expander definitions.
5991 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
5994 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
5996 2017-09-26 Richard Biener <rguenther@suse.de>
5998 PR tree-optimization/82320
5999 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6002 2017-09-25 Jeff Law <law@redhat.com>
6004 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6005 prototype for new argument.
6006 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6007 mostly extracted from rs6000_emit_allocate_stack.
6008 (rs6000_emit_probe_stack_range_stack_clash): New function.
6009 (rs6000_emit_allocate_stack): Call
6010 rs6000_emit_probe_stack_range_stack_clash as needed.
6011 (rs6000_emit_probe_stack_range): Add additional argument
6012 to call to gen_probe_stack_range{si,di}.
6013 (output_probe_stack_range): New.
6014 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6015 (output_probe_stack_range_stack_clash): New.
6016 (rs6000_emit_prologue): Emit notes into dump file as requested.
6017 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6018 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6019 Add additional operand and pass it to output_probe_stack_range.
6021 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6023 PR tree-optimization/82163
6024 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6025 (checking_verify_loop_closed_ssa): New parameter.
6026 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6027 (check_loop_closed_ssa_stmt): Delete.
6028 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6029 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6030 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6033 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6035 * brig-builtins.def: Treat HSAIL barrier builtins as
6036 setjmp/longjump style functions.
6038 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6040 * target.def (constant_alignment): New hook.
6041 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6042 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6043 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6044 * doc/tm.texi: Regenerate.
6045 * targhooks.h (default_constant_alignment): Declare.
6046 (constant_alignment_word_strings): Likewise.
6047 * targhooks.c (default_constant_alignment): New function.
6048 (constant_alignment_word_strings): Likewise.
6049 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6050 instead of CONSTANT_ALIGNMENT.
6051 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6052 (force_const_mem): Likewise.
6053 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6054 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6055 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6056 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6057 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6059 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6060 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6061 constant_alignment_word_strings.
6062 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6063 (CONSTANT_ALIGNMENT): Likewise.
6064 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6065 (arm_constant_alignment): New function.
6066 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6067 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6068 constant_alignment_word_strings.
6069 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6070 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6071 constant_alignment_word_strings.
6072 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6073 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6074 (cris_constant_alignment): New function.
6075 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6076 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6077 (epiphany_constant_alignment): New function.
6078 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6079 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6080 constant_alignment_word_strings.
6081 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6082 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6083 constant_alignment_word_strings.
6084 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6085 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6086 constant_alignment_word_strings.
6087 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6088 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6089 * config/i386/i386.c (ix86_constant_alignment): Make static.
6090 Use the same interface as the target hook.
6091 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6092 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6093 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6094 constant_alignment_word_strings.
6095 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6096 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6097 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6098 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
6099 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6100 constant_alignment_word_strings.
6101 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
6102 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6103 constant_alignment_word_strings.
6104 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
6105 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6106 constant_alignment_word_strings.
6107 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
6108 * config/microblaze/microblaze.c (microblaze_constant_alignment):
6110 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6111 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
6112 * config/mips/mips.c (mips_constant_alignment): New function.
6113 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6114 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
6115 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
6116 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6117 (mmix_constant_alignment): Make static. Use the same interface
6119 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
6120 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6121 constant_alignment_word_strings.
6122 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
6123 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6124 constant_alignment_word_strings.
6125 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
6126 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6127 constant_alignment_word_strings.
6128 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
6129 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6130 (rs6000_constant_alignment): New function.
6131 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
6132 * config/riscv/riscv.c (riscv_constant_alignment): New function.
6133 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6134 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
6135 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6136 (rs6000_constant_alignment): New function.
6137 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
6138 * config/s390/s390.c (s390_constant_alignment): New function.
6139 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6140 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
6141 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6142 constant_alignment_word_strings.
6143 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
6144 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6145 (sparc_constant_alignment): New function.
6146 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
6147 * config/spu/spu.c (spu_constant_alignment): New function.
6148 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6149 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
6150 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6151 constant_alignment_word_strings.
6152 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
6153 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6154 constant_alignment_word_strings.
6155 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
6156 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6157 constant_alignment_word_strings.
6158 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
6159 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6160 (visium_constant_alignment): New function.
6161 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6162 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6163 (xtensa_constant_alignment): New function.
6164 * system.h (CONSTANT_ALIGNMENT): Poison.
6166 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6168 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6169 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6170 (rs6000_builtin_valid_without_lhs): New helper function.
6171 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6172 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6174 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6176 * target.h (vec_perm_indices): Use unsigned short rather than
6178 (auto_vec_perm_indices): Likewise.
6179 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6180 Use unsigned int rather than unsigned char.
6181 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6183 2017-09-25 Richard Biener <rguenther@suse.de>
6185 * cfgloop.h (sort_sibling_loops): Declare.
6186 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6187 (sort_sibling_loops): New function sorting the sibling loop list
6189 * graphite.c (graphite_transform_loops): Sort sibling loops.
6191 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6193 * target.def (vec_perm_const_ok): Change sel parameter to
6195 * optabs-query.c (can_vec_perm_p): Update accordingly.
6196 * doc/tm.texi: Regenerate.
6197 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6198 auto_vec_perm_indices and remove separate nelt field.
6199 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6200 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6201 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6202 (aarch64_expand_vec_perm_const): Update accordingly.
6203 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6204 to vec_perm_indices.
6205 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6206 auto_vec_perm_indices and remove separate nelt field.
6207 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6208 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6209 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6211 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6212 to vec_perm_indices.
6213 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6214 sel to vec_perm_indices.
6215 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6216 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6217 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6219 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6222 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6225 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6226 on the FUNCTION_DECL function context if it has a DIE that is a
6229 2017-09-25 Richard Biener <rguenther@suse.de>
6231 PR tree-optimization/82285
6232 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6235 2017-09-25 Tom de Vries <tom@codesourcery.com>
6239 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6242 2017-09-25 Richard Biener <rguenther@suse.de>
6244 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6245 ISL errors other than isl_error_quota happen. Dump if the
6246 schedule is the same.
6247 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6248 errors instead of aborting inside ISL.
6250 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6253 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6254 of libgcc_eh for m64.
6255 * config/i386/darwin64.h: Likewise.
6257 2017-09-25 Richard Biener <rguenther@suse.de>
6260 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6261 attribute for incomplete types nor twice for complete ones.
6263 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6266 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6267 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6270 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6273 * config/i386/i386.opt (mprefer-avx256): Use
6274 ix86_target_flags variable.
6275 * config/i386/i386.c (ix86_target_string): Move
6276 -mprefer-avx256 to flag2_opts.
6278 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6281 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6282 and x != -1 | y != -1 into (x & y) != -1.
6284 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6286 * config.gcc: Add new case statement to set
6287 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6288 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6289 s390x-*-linux* case statements. Added aarch64 to the list of
6290 supported architectures.
6292 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6294 PR tree-optimization/82289
6295 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6296 STMT_VINFO_RELEVANT_P.
6298 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6299 Alan Hayward <alan.hayward@arm.com>
6300 David Sherwood <david.sherwood@arm.com>
6302 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6303 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6304 (extract_range_from_binary_expr_1): Don't call
6305 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6307 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6308 Alan Hayward <alan.hayward@arm.com>
6309 David Sherwood <david.sherwood@arm.com>
6311 * target.def (preferred_vector_alignment): New hook.
6312 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6314 * doc/tm.texi: Regenerate.
6315 * targhooks.h (default_preferred_vector_alignment): Declare.
6316 * targhooks.c (default_preferred_vector_alignment): New function.
6317 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6319 (DR_TARGET_ALIGNMENT): New macro.
6320 (aligned_access_p): Update commentary.
6321 (vect_known_alignment_in_bytes): New function.
6322 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6324 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6325 Calculate the misalignment based on the target alignment rather than
6327 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6328 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6329 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6330 the target alignment, rather than masking the element misalignment
6331 with the number of elements in a vector. Also use the target
6332 alignment when calculating the maximum number of peels.
6333 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6334 instead of TYPE_ALIGN_UNIT.
6335 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6336 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6337 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6338 (vect_create_data_ref_ptr): Likewise.
6339 (vect_setup_realignment): Realign by ANDing with
6340 -DR_TARGET_MISALIGNMENT.
6341 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6342 the number of peels based on DR_TARGET_ALIGNMENT.
6343 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6344 with the guaranteed alignment boundary when deciding whether
6346 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6347 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6348 (ensure_base_align): Remove stmt_info parameter. Get the
6349 target base alignment from DR_TARGET_ALIGNMENT.
6350 (vectorizable_store): Update call accordingly. Interpret
6351 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6353 (vectorizable_load): Likewise.
6355 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6356 Alan Hayward <alan.hayward@arm.com>
6357 David Sherwood <david.sherwood@arm.com>
6359 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6360 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6361 (vect_enhance_data_refs_alignment): Likewise.
6363 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6365 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6366 error. Only quit immediately if parsing is complete.
6367 (BEGIN): Initialize fatal_err and parse_done.
6368 (begin fpu, end fpu): Check number of arguments.
6369 (begin arch, end arch): Likewise.
6370 (begin cpu, end cpu): Likewise.
6371 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6372 (optalias): Likewise.
6374 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6376 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6377 * config/arm/arm-isa.h: Delete. Move definitions to ...
6378 * arm-cpus.in: ... here. Use new feature and fgroup values.
6379 * config/arm/arm.c (arm_option_override): Use lower case for feature
6381 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6382 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6383 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6385 (print_isa_bits_for): New function.
6386 (gen_isa): New function.
6387 (gen_comm_data): Use print_isa_bits_for.
6388 (define feature): New keyword.
6389 (define fgroup): New keyword.
6390 * config/arm/t-arm (TM_H): Remove.
6391 (GTM_H): Add arm-isa.h.
6392 (arm-isa.h): Add rule to generate file.
6393 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6394 case for feature bit names.
6396 2017-09-22 Richard Biener <rguenther@suse.de>
6398 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6400 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6401 print of no dependency loops ...
6402 * graphite.c (graphite_transform_loops): ... here.
6403 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6405 (same_close_phi_node, remove_duplicate_close_phi,
6406 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6407 (canonicalize_loop_closed_ssa): ... here and simplify.
6408 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6409 (optimize_isl): Use dump_printf_loc to tell when we stopped
6410 optimizing because of an ISL timeout.
6412 2017-09-22 Richard Biener <rguenther@suse.de>
6414 PR tree-optimization/82291
6415 * tree-if-conv.c (predicate_mem_writes): Make sure to
6416 remove writes in blocks predicated with false.
6418 2017-09-22 Richard Biener <rguenther@suse.de>
6420 * sese.c: Include cfganal.h.
6421 (if_region_set_false_region): Remove.
6422 (create_if_region_on_edge): Likewise.
6423 (move_sese_in_condition): Re-implement without destroying
6426 2017-09-22 Richard Biener <rguenther@suse.de>
6428 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6429 Verify both BBs contain loop PHI nodes before dispatching to
6431 (graphite_regenerate_ast_isl): Do not recompute dominators,
6432 do not verify three times. Restructure for clarity.
6433 * graphite-scop-detection.c (same_close_phi_node,
6434 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6435 defined_in_loop_p, canonicalize_loop_closed_ssa,
6436 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6437 checking and SSA rewrite, move to ...
6438 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6439 (graphite_initialize): Do not pass in ctx, do not reset the
6440 SCEV cache, compute only dominators.
6441 (graphite_transform_loops): Allocate ISL ctx after
6442 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6443 Maintain post-dominators only around build_scops.
6444 * sese.c (if_region_set_false_region): Make static. Free
6445 and recompute dominators.
6446 (move_sese_in_condition): Assert we don't get called with
6447 post-dominators computed.
6448 * sese.h (if_region_set_false_region): Remove.
6450 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6452 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6453 mode attribute for TARGET_AVX512VL.
6455 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6457 * config/i386/i386.opt (mprefer-avx256): New option.
6458 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6460 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6461 for TARGET_PREFER_AVX256.
6462 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6464 2017-09-21 Jeff Law <law@redhat.com>
6466 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6467 Fix dump output if the only stack space is for pushed registers.
6469 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6471 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6474 2017-09-21 Martin Sebor <msebor@redhat.com>
6477 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6478 code (in C++) or code that triggers warnings.
6480 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6482 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6484 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6486 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6487 * sched-rgn.c: Ditto.
6488 * sel-sched-ir.c: Ditto.
6490 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6492 * toplev.h (set_random_seed): Adjust return type.
6493 * toplev.c (init_local_tick): Move eager initialization of random_seed
6494 to get_random_seed. Adjust comment.
6495 (init_random_seed): Inline to get_random_seed, delete.
6496 (get_random_seed): Initialize random_seed lazily.
6497 (set_random_seed): Do not return previous value.
6498 (print_switch_value): Do not call get_random_seed.
6500 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6502 * cgraph.c (delete_function_version): New, broken out from...
6503 (cgraph_node::delete_function_version): ...here. Rename to
6504 cgraph_node::delete_function_version_by_decl. Update all uses.
6505 (cgraph_node::remove): Call delete_function_version.
6507 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6510 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6511 VAR_DECLs to which addressable non-volatile parameters are mapped
6512 and for id->retvar after the return value assignment. Clear
6513 id->retval and id->retbnd after inlining.
6515 2017-09-21 Richard Biener <rguenther@suse.de>
6517 PR tree-optimization/82276
6518 PR tree-optimization/82244
6519 * tree-vrp.c (build_assert_expr_for): Set
6520 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6522 (remove_range_assertions): Revert earlier change.
6524 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6527 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6529 2017-09-21 Richard Biener <rguenther@suse.de>
6531 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6532 Restore valid IL after code generation errors.
6533 * graphite.c (graphite_transform_loops): Diagnose code
6534 generation issues as MSG_MISSED_OPTIMIZATION and continue
6535 with processing SCOPs.
6537 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6538 Alan Hayward <alan.hayward@arm.com>
6539 David Sherwood <david.sherwood@arm.com>
6541 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6542 rather than choosing between plus_constant and gen_rtx_<CODE>.
6543 * expr.c (emit_push_insn): Likewise.
6544 (expand_expr_real_2): Likewise.
6546 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6547 Alan Hayward <alan.hayward@arm.com>
6548 David Sherwood <david.sherwood@arm.com>
6550 * loop-unroll.c (split_iv): Call copy_rtx on the step.
6552 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6553 Alan Hayward <alan.hayward@arm.com>
6554 David Sherwood <david.sherwood@arm.com>
6556 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6557 calling tree_to_uhwi.
6559 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6560 Alan Hayward <alan.hayward@arm.com>
6561 David Sherwood <david.sherwood@arm.com>
6563 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6564 INTEGER_CST rather than a negative test for ADDR_EXPR.
6566 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6567 Alan Hayward <alan.hayward@arm.com>
6568 David Sherwood <david.sherwood@arm.com>
6570 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6571 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6573 2017-09-21 Richard Biener <rguenther@suse.de>
6575 PR tree-optimization/71351
6576 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6577 graphite_create_new_loop_guard): Remove, fold remaining parts
6579 (translate_isl_ast_node_for): ... here and simplify.
6581 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6584 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6585 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6586 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6587 alternative always use QI mode, for -Os imov (=R,R) alternative
6588 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6591 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6592 Jeff Law <law@redhat.com>
6594 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6595 (allocate_stack_space): New function, partially extracted from
6597 (s390_emit_prologue): Track offset to most recent stack probe.
6598 Code to allocate space moved into allocate_stack_space.
6599 Dump actions when no stack is allocated.
6600 (s390_prologue_plus_offset): New function.
6601 (s390_emit_stack_probe): Likewise.
6603 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
6605 * common.opt (Wa, Wl, Wp, g, gz=): Add
6607 (gno-column-info): Remove.
6608 (gcolumn-info): Drop RejectNegative.
6610 (gno-record-gcc-switches): Remove.
6611 (grecord-gcc-switches): Drop RejectNegative.
6612 (gno-split-dwarf): Remove.
6613 (gsplit-dwarf): Drop RejectNegative.
6614 (gno-strict-dwarf): Remove.
6615 (gstrict-dwarf): Drop RejectNegative.
6616 * config/darwin.opt (gfull, gused): Add RejectNegative.
6617 * dwarf2out.c (gen_producer_string): Drop
6618 gno-record-gcc-switches handler.
6619 * optc-gen.awk: Add g to prefixes with negative forms.
6620 * opts-common.c (remapping_prefix_p): New.
6621 (find_opt): Check it.
6622 (generate_canonical_option): Test g prefix.
6623 (option_map): Add -gno- mapping.
6624 (add_misspelling_candidates): Check remapping_prefix_p.
6626 2017-09-20 Jeff Law <law@redhat.com>
6628 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6629 thinko in stack clash protection support.
6631 * explow.c (compute_stack_clash_protection_loop_data): Use
6632 CONST_INT_P instead of explicit test. Verify object is a
6633 CONST_INT_P before looking at INTVAL.
6634 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6635 instead of explicit test.
6637 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6640 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6641 address instead of to r1 and r11.
6643 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6645 * config.gcc: Support "knm".
6646 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6647 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6649 * config/i386/i386.c (m_KNM): Define.
6650 (processor_target_table): Add "knm".
6652 (ix86_option_override_internal): Add "knm".
6653 (ix86_issue_rate): Add PROCESSOR_KNM.
6654 (ix86_adjust_cost): Ditto.
6655 (ia32_multipass_dfa_lookahead): Ditto.
6656 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6657 (fold_builtin_cpu): Add M_INTEL_KNM.
6658 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6659 (processor_type): Add PROCESSOR_KNM.
6660 * config/i386/x86-tune.def: Add m_KNM.
6661 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6663 2017-09-20 Richard Biener <rguenther@suse.de>
6665 PR tree-optimization/80213
6666 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6667 are allowed in empty BBs as well.
6668 (canonicalize_loop_closed_ssa): Also look for other complex
6670 (scop_detection::get_sese): Include the loop-closed PHI block
6672 (scop_detection::merge_sese): Remove code adding extra blocks.
6673 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6674 (build_scops): Assert the final returned scop is invalid.
6676 2017-09-20 Richard Biener <rguenther@suse.de>
6678 PR tree-optimization/82264
6679 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6681 (vn_phi_lookup): Likewise.
6682 (vn_phi_insert): Likewise.
6684 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6686 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6687 that fits into uhwi or shwi, add DW_AT_const_value regardless
6688 of early_dwarf without going through RTL, using add_AT_unsigned
6691 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6692 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6693 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6694 (DEBUG_MACRO_SECTION): Likewise.
6695 (DEBUG_LTO_MACRO_SECTION): Likewise.
6696 (DEBUG_STR_DWO_SECTION): Likewise.
6697 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6698 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6699 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6700 (DEBUG_LTO_LINE_STR_SECTION): Define.
6701 (init_sections_and_labels): Initialize debug_line_str_section
6702 variable. Initialize debug_loc_section for -gdwarf-5 to
6703 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6705 2017-09-20 Richard Biener <rguenther@suse.de>
6707 * graphite-sese-to-poly.c (extract_affine): Properly handle
6708 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6710 2017-09-20 Richard Biener <rguenther@suse.de>
6712 PR tree-optimization/81373
6713 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6714 Force SESE live-out defs to be handled even if they are
6717 2017-09-19 Jeff Law <law@redhat.com>
6719 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6720 nothing for stack adjustments with REG_STACK_CHECK.
6721 * sched-deps.c (parse_add_or_inc): Reject insns with
6722 REG_STACK_CHECK from dependency breaking.
6723 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6724 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6725 * reg-notes.def (STACK_CHECK): New note.
6727 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6728 (ix86_expand_prologue): Dump stack clash info as needed.
6729 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6731 * function.c (dump_stack_clash_frame_info): New function.
6732 * function.h (dump_stack_clash_frame_info): Prototype.
6733 (enum stack_clash_probes): New enum.
6735 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6736 flag_stack_clash_protection.
6737 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6738 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6739 (arm_frame_pointer_required): Likewise.
6740 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6741 (ia64_expand_prologue): Likewise.
6742 * config/mips/mips.c (mips_expand_prologue): Likewise.
6743 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6744 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6745 (sparc_flat_expand_prologue): Likewise.
6746 * config/spu/spu.c (spu_expand_prologue): Likewise.
6748 * explow.c: Include "params.h".
6749 (anti_adjust_stack_and_probe_stack_clash): New function.
6750 (get_stack_check_protect): Likewise.
6751 (compute_stack_clash_protection_loop_data): Likewise.
6752 (emit_stack_clash_protection_loop_start): Likewise.
6753 (emit_stack_clash_protection_loop_end): Likewise.
6754 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6755 Use anti_adjust_stack_and_probe_stack_clash.
6756 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6757 (emit_stack_clash_protection_loop_start): Likewise.
6758 (emit_stack_clash_protection_loop_end): Likewise.
6759 * rtl.h (get_stack_check_protect): Prototype.
6760 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6761 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6762 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6764 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6766 * doc/tm.texi: Rebuilt.
6767 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6768 get_stack_check_protect.
6769 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6770 * config/arm/arm.c (arm_expand_prologue): Likewise.
6771 (arm_frame_pointer_required): Likewise.
6772 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6773 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6774 * config/mips/mips.c (mips_expand_prologue): Likewise.
6775 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6776 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6777 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6778 (sparc_flat_expand_prologue): Likewise.
6780 * common.opt (-fstack-clash-protection): New option.
6781 * flag-types.h (enum stack_check_type): Note difference between
6782 -fstack-check= and -fstack-clash-protection.
6783 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6784 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6785 * toplev.c (process_options): Issue warnings/errors for cases
6786 not handled with -fstack-clash-protection.
6787 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6788 (-fstack-check): Note additional problem with -fstack-check=generic.
6789 Note that -fstack-check is primarily for Ada and refer users
6790 to -fstack-clash-protection for stack-clash-protection.
6791 Document new params for stack clash protection.
6793 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6795 * config/i386/i386.c (ix86_split_long_move): Do not handle
6796 address used for LEA in a special way.
6798 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6800 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6802 2017-09-19 Martin Sebor <msebor@redhat.com>
6805 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6806 of incompatible types.
6808 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6810 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6811 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6812 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6813 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6815 2017-09-19 Richard Biener <rguenther@suse.de>
6817 PR tree-optimization/82244
6818 * tree-vrp.c (remove_range_assertions): Do not propagate
6819 a constant to abnormals but replace the assert with a copy.
6821 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6823 PR rtl-optimization/57878
6824 PR rtl-optimization/68988
6825 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6826 avoidance test involving non_reload_pseudos. Move frequency test
6827 below the general fragmentation avoidance test.
6829 2017-09-19 Richard Biener <rguenther@suse.de>
6831 PR tree-optimization/69728
6832 * graphite-sese-to-poly.c (schedule_error): New global.
6833 (add_loop_schedule): Handle empty domain by failing the
6835 (build_original_schedule): Handle schedule_error.
6837 2017-09-19 Richard Biener <rguenther@suse.de>
6839 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6840 Do not iterate to sibling loops but only to siblings of inner
6843 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6846 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6847 that the registers are different.
6849 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6851 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6852 to processor_model and "amdfam17h" to arch_names_table.
6853 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6855 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6858 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6860 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6861 Alan Hayward <alan.hayward@arm.com>
6862 David Sherwood <david.sherwood@arm.com>
6864 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6866 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6868 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6869 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6870 vect_schedule_slp_instance.
6871 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6872 Update call to vect_slp_analyze_node_operations. Simplify return
6874 (vect_slp_analyze_bb_1): Update call accordingly.
6875 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6876 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6877 (vect_schedule_slp): Update call accordingly.
6879 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6880 Alan Hayward <alan.hayward@arm.com>
6881 David Sherwood <david.sherwood@arm.com>
6883 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6884 with types that aren't in fact scalar.
6886 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6888 * tree-vect-slp.c (vect_record_max_nunits): New function,
6890 (vect_build_slp_tree_1): ...here.
6891 (vect_build_slp_tree_2): Call it for phis too.
6893 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6895 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6896 to vect_get_vec_def_for_operand when getting the mask operand.
6898 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6899 Alan Hayward <alan.hayward@arm.com>
6900 David Sherwood <david.sherwood@arm.com>
6902 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
6905 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6906 Alan Hayward <alan.hayward@arm.com>
6907 David Sherwood <david.sherwood@arm.com>
6909 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
6910 calculation for vector booleans.
6912 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6913 Alan Hayward <alan.hayward@arm.com>
6914 David Sherwood <david.sherwood@arm.com>
6916 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
6918 (vect_transform_stmt): ...here.
6919 (vect_analyze_stmt): Use it instead of calling
6920 vectorizable_live_operation directly.
6922 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
6924 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
6925 non-SIMT targets in acc vector loops.
6927 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
6929 * configure.ac: Add arc and check if assembler supports gdwarf2.
6930 * configure: Regenerate.
6932 2017-09-18 Richard Biener <rguenther@suse.de>
6934 PR tree-optimization/82220
6935 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
6936 epilogue niters from the min_profitable_iters compute.
6938 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6941 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
6942 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
6943 (ix86_init_pic_reg): Revert 2017-09-01 changes.
6945 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
6948 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
6949 switching to a new text section.
6951 2017-09-18 Richard Biener <rguenther@suse.de>
6953 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
6955 (build_alias_set): Reject aliases with no access function.
6957 2017-09-18 Richard Biener <rguenther@suse.de>
6959 PR tree-optimization/79622
6960 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
6962 (build_cross_bb_scalars_use): Likewise.
6964 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
6966 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
6968 2017-09-18 Alan Modra <amodra@gmail.com>
6971 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
6972 stack_pointer_rtx for count 0. Update comments. Break up
6973 large rtl expression.
6975 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
6977 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
6978 Increase to 20 bytes.
6979 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
6980 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
6981 or avx version of the stub.
6983 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
6986 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
6987 compute the minimum stack alignment. Also update preferred stack
6988 boundary for leaf functions.
6990 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
6992 PR tree-optimization/82228
6993 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
6996 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
6998 * common/config/nds32/nds32-common.c
6999 (nds32_option_optimization_table): Refine formatting.
7000 (nds32_option_optimization_table): Use -fsched-pressure and
7001 -fomit-frame-pointer for specific optimization level.
7003 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7005 * config/nds32/nds32.c: Refine formatting and comments.
7006 * config/nds32/nds32.h: Likewise.
7007 * config/nds32/nds32.md: Likewise.
7008 * config/nds32/nds32-cost.c: Likewise.
7009 * config/nds32/nds32-isr.c: Likewise.
7010 * config/nds32/nds32-md-auxiliary.c: Likewise.
7011 * config/nds32/nds32-multiple.md: Likewise.
7012 * config/nds32/nds32-predicates.c: Likewise.
7014 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7015 Jakub Jelinek <jakub@redhat.com>
7017 Add support for -std=c++2a.
7018 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7020 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7022 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7025 * doc/extend.texi (Common Function Attributes): Add
7026 references to ARM, AArch64, and S/390 specific attributes.
7027 (Function Specific Option Pragmas): Add AArch64 and S/390
7028 to list of back ends that support the target pragma.
7030 2017-09-15 Nathan Sidwell <nathan@acm.org>
7032 * doc/standards.texi: Fix C++17 description. Update URLs for
7035 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7037 * common.opt (Wcast-align=strict): New warning option.
7038 * doc/invoke.texi: Document -Wcast-align=strict.
7040 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7042 * cgraph.h (cgraph_thunk_info): Add comments.
7043 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7044 assert for VIRTUAL_* arguments stricter.
7046 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7048 PR tree-optimization/71026
7049 * match.pd: Move RDIV patterns from fold-const.c
7050 * fold-const.c (distribute_real_division): Removed.
7051 (fold_binary_loc): Remove calls to distribute_real_divison.
7053 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7055 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7056 c++1z and gnu++1z as deprecated. Change other references to
7057 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7058 Change -Wc++1z-compat to -Wc++17-compat.
7059 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7060 * dwarf2out.c (highest_c_language): Handle C++17.
7061 (gen_compile_unit_die): Likewise.
7063 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7065 PR rtl-optimization/82192
7066 * combine.c (make_extraction): Don't look through non-paradoxical
7067 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7070 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7071 Alan Hayward <alan.hayward@arm.com>
7072 David Sherwood <david.sherwood@arm.com>
7074 * target.def (function_arg_offset): New hook.
7075 * targhooks.h (default_function_arg_offset): Declare.
7076 * targhooks.c (default_function_arg_offset): New function.
7077 * function.c (locate_and_pad_parm): Use
7078 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7079 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7080 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7081 * doc/tm.texi: Regenerate.
7082 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7083 * config/spu/spu.c (spu_function_arg_offset): New function.
7084 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7085 * system.h (FUNCTION_ARG_OFFSET): Poison.
7087 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7088 Alan Hayard <alan.hayward@arm.com>
7089 David Sherwood <david.sherwood@arm.com>
7091 * target.def (truly_noop_truncation): New hook.
7092 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7093 than TRULY_NOOP_TRUNCATION.
7094 * hooks.h (hook_bool_uint_uint_true): Declare.
7095 * hooks.c (hook_bool_uint_uint_true): New function.
7096 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7097 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7098 * doc/tm.texi: Regenerate.
7099 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
7100 rather than TRULY_NOOP_TRUNCATION in comments.
7101 (simplify_comparison): Likewise.
7102 (record_truncated_value): Likewise.
7103 * expmed.c (extract_bit_field_1): Likewise.
7104 (extract_split_bit_field): Likewise.
7105 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
7106 instead of TRULY_NOOP_TRUNCATION.
7107 * function.c (assign_parm_setup_block): Likewise.
7108 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
7109 * rtlhooks.c: Include target.h.
7110 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
7111 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
7112 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
7113 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
7114 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
7115 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
7116 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
7117 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
7118 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
7119 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
7120 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
7121 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
7122 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
7123 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
7124 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
7125 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
7126 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
7127 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
7128 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
7129 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
7130 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
7131 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
7132 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
7133 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
7134 * config/mips/mips.c (mips_truly_noop_truncation): New function.
7135 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7136 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7137 rather than TRULY_NOOP_TRUNCATION in comments.
7138 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
7139 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
7140 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
7141 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
7142 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
7143 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
7144 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
7145 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
7146 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
7147 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
7148 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
7149 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7150 rather than TRULY_NOOP_TRUNCATION in comments.
7151 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
7152 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
7153 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
7154 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
7155 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
7156 TRULY_NOOP_TRUNCATION condition.
7157 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
7158 (TRULY_NOOP_TRUNCATION): Delete.
7159 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
7160 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
7161 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7162 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7163 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7164 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7165 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7166 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7167 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7168 rather than TRULY_NOOP_TRUNCATION in comments.
7169 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7170 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7171 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7172 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7173 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7174 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7176 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7179 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7180 (*cmp_ior): Likewise.
7181 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7182 (*ior_scc_scc_cmp): Likewise.
7183 (*and_scc_scc): Likewise.
7184 (*and_scc_scc_cmp): Likewise.
7186 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7187 Alan Hayard <alan.hayward@arm.com>
7188 David Sherwood <david.sherwood@arm.com>
7190 * target.def (can_change_mode_class): New hook.
7191 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7192 (hard_regno_nregs): Likewise.
7193 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7194 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7195 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7196 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7197 (LOAD_EXTEND_OP): Update accordingly.
7198 * doc/tm.texi: Regenerate.
7199 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7200 CANNOT_CHANGE_MODE_CLASS.
7201 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7202 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7203 * combine.c (simplify_set): Update accordingly.
7204 * emit-rtl.c (validate_subreg): Likewise.
7205 * recog.c (general_operand): Likewise.
7206 * regcprop.c (mode_change_ok): Likewise.
7207 * reload1.c (choose_reload_regs): Likewise.
7208 (inherit_piecemeal_p): Likewise.
7209 * rtlanal.c (simplify_subreg_regno): Likewise.
7210 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7211 instead of CANNOT_CHANGE_MODE_CLASS.
7212 (reload_cse_simplify_operands): Likewise.
7213 * reload.c (push_reload): Use targetm.can_change_mode_class
7214 instead of CANNOT_CHANGE_MODE_CLASS.
7215 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7216 REG_CANNOT_CHANGE_MODE_P.
7217 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7218 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7219 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7220 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7221 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7222 (arm_can_change_mode_class): New function.
7223 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7224 than CANNOT_CHANGE_MODE_CLASS in comments.
7225 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7226 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7227 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7228 (ix86_can_change_mode_class): ...this new function, inverting the
7229 sense of the return value.
7230 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7231 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7232 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7233 (ia64_can_change_mode_class): New function.
7234 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7235 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7236 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7237 (m32c_can_change_mode_class): ...this new function, inverting the
7238 sense of the return value.
7239 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7240 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7241 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7242 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7243 (mips_can_change_mode_class): ...this new function, inverting the
7244 sense of the return value.
7245 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7246 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7247 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7248 (msp430_can_change_mode_class): New function.
7249 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7250 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7251 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7252 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7253 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7254 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7255 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7256 (pa_cannot_change_mode_class): Replace with...
7257 (pa_can_change_mode_class): ...this new function, inverting the
7258 sense of the return value.
7259 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7260 than CANNOT_CHANGE_MODE_CLASS in comments.
7261 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7262 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7263 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7264 (pdp11_cannot_change_mode_class): Replace with...
7265 (pdp11_can_change_mode_class): ...this new function, inverting the
7266 sense of the return value.
7267 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7268 * config/powerpcspe/powerpcspe-protos.h
7269 (rs6000_cannot_change_mode_class_ptr): Delete.
7270 * config/powerpcspe/powerpcspe.c
7271 (rs6000_cannot_change_mode_class_ptr): Delete.
7272 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7273 (rs6000_option_override_internal): Assign to
7274 targetm.can_change_mode_class instead of
7275 rs6000_cannot_change_mode_class_ptr.
7276 (rs6000_cannot_change_mode_class): Replace with...
7277 (rs6000_can_change_mode_class): ...this new function, inverting the
7278 sense of the return value.
7279 (rs6000_debug_cannot_change_mode_class): Replace with...
7280 (rs6000_debug_can_change_mode_class): ...this new function.
7281 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7282 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7283 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7284 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7285 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7287 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7288 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7289 (rs6000_option_override_internal): Assign to
7290 targetm.can_change_mode_class instead of
7291 rs6000_cannot_change_mode_class_ptr.
7292 (rs6000_cannot_change_mode_class): Replace with...
7293 (rs6000_can_change_mode_class): ...this new function, inverting the
7294 sense of the return value.
7295 (rs6000_debug_cannot_change_mode_class): Replace with...
7296 (rs6000_debug_can_change_mode_class): ...this new function.
7297 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7298 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7299 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7300 (s390_can_change_mode_class): ...this new function, inverting the
7301 sense of the return value.
7302 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7303 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7304 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7305 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7306 (sh_cannot_change_mode_class): Replace with...
7307 (sh_can_change_mode_class): ...this new function, inverting the
7308 sense of the return value.
7309 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7310 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7311 (sparc_can_change_mode_class): New function.
7312 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7313 * config/spu/spu.c (spu_can_change_mode_class): New function.
7314 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7315 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7316 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7317 (visium_can_change_mode_class): New function.
7318 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7320 2017-09-15 Richard Biener <rguenther@suse.de>
7322 PR tree-optimization/82217
7323 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7324 but not undefined case.
7326 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7329 * postreload.c (reload_cse_simplify_operands): Skip
7330 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7332 2017-09-15 Richard Biener <rguenther@suse.de>
7334 PR tree-optimization/68823
7335 * graphite-scop-detection.c (build_alias_set): If we have a
7336 possible dependence check whether we can handle them by just
7337 looking at the DRs DR_ACCESS_FNs.
7338 (build_scops): If build_alias_set fails, fail the SCOP.
7340 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7342 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7343 to support float128 built-in functions that require the ISA 3.0
7345 (BU_FLOAT128_3_HW): Likewise.
7346 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7348 (FMAF128): Likewise.
7350 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7351 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7352 floating point instructions.
7353 (rs6000_invalid_builtin): Likewise.
7354 (rs6000_builtin_mask_names): Likewise.
7355 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7356 (RS6000_BTM_FLOAT128_HW): Likewise.
7357 (RS6000_BTM_COMMON): Likewise.
7358 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7360 * doc/extend.texi (RS/6000 built-in functions): Document the
7361 IEEE 128-bit floating point square root and fused multiply-add
7364 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7366 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7367 reg (r2) isn't in the set of registers defined in the prologue.
7369 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7370 Alan Hayward <alan.hayward@arm.com>
7371 David Sherwood <david.sherwood@arm.com>
7373 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7374 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7375 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7376 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7377 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7379 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7380 max_vectorization_factor.
7381 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7383 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7384 Alan Hayward <alan.hayward@arm.com>
7385 David Sherwood <david.sherwood@arm.com>
7387 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7388 (vect_worthwhile_without_simd_p): Declare.
7389 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7390 (vectorizable_reduction): Use it.
7391 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7392 (vectorizable_operation): Likewise.
7394 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7395 Alan Hayward <alan.hayward@arm.com>
7396 David Sherwood <david.sherwood@arm.com>
7398 * tree-vectorizer.h (vect_get_num_copies): New function.
7399 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7400 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7401 (vectorizable_induction): Likewise.
7402 (vectorizable_live_operation): Likewise.
7403 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7404 (vectorizable_bswap): Likewise.
7405 (vectorizable_call): Likewise.
7406 (vectorizable_conversion): Likewise.
7407 (vectorizable_assignment): Likewise.
7408 (vectorizable_shift): Likewise.
7409 (vectorizable_operation): Likewise.
7410 (vectorizable_store): Likewise.
7411 (vectorizable_load): Likewise.
7412 (vectorizable_condition): Likewise.
7413 (vectorizable_comparison): Likewise.
7414 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7416 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7417 Alan Hayward <alan.hayward@arm.com>
7418 David Sherwood <david.sherwood@arm.com>
7420 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7421 of vect_init_vector.
7423 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7424 Alan Hayward <alan.hayward@arm.com>
7425 David Sherwood <david.sherwood@arm.com>
7427 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7428 an inline wrapper that provides a location.
7429 (gimple_build_vector): Likewise.
7430 * gimple-fold.c (gimple_build_vector_from_val): New function.
7431 (gimple_build_vector): Likewise.
7432 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7433 functions to build the initial value. Always return a gimple value.
7434 (get_initial_defs_for_reduction): Likewise. Only compute
7436 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7437 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7438 (vectorizable_induction): Use gimple_build_vector rather than
7441 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7442 Alan Hayward <alan.hayward@arm.com>
7443 David Sherwood <david.sherwood@arm.com>
7445 * target.h (vec_perm_indices): New typedef.
7446 (auto_vec_perm_indices): Likewise.
7447 * optabs-query.h: Include target.h
7448 (can_vec_perm_p): Take a vec_perm_indices *.
7449 * optabs-query.c (can_vec_perm_p): Likewise.
7450 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7451 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7452 * tree-vect-generic.c (lower_vec_perm): Likewise.
7453 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7454 (vect_grouped_load_supported): Likewise.
7455 (vect_shift_permute_load_chain): Likewise.
7456 (vect_permute_store_chain): Use auto_vec_perm_indices.
7457 (vect_permute_load_chain): Likewise.
7458 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7459 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7460 Update uses of can_vec_perm_p.
7461 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7462 mode with a number of elements. Take a vec_perm_indices *.
7463 (vect_create_epilog_for_reduction): Update accordingly.
7464 Use auto_vec_perm_indices.
7465 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7466 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7467 (vect_transform_slp_perm_load): Likewise.
7468 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7469 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7470 (vect_gen_perm_mask_checked): Likewise.
7471 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7472 (vect_gen_perm_mask_checked): Likewise.
7473 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7474 (vectorizable_store): Likewise.
7475 (vectorizable_load): Likewise.
7476 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7477 (vectorizable_bswap): Likewise.
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 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7484 * tree.c (build_vector): Likewise.
7485 (build_vector_from_ctor): Update accordingly.
7486 (build_vector_from_val): Likewise.
7487 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7488 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7489 * tree-vect-generic.c (add_rshift): Likewise.
7490 (expand_vector_divmod): Likewise.
7491 (optimize_vector_constructor): Likewise.
7492 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7493 (vect_transform_slp_perm_load): Likewise.
7494 (vect_schedule_slp_instance): Likewise.
7495 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7496 (vectorizable_call): Likewise.
7497 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7498 * expmed.c (make_tree): Likewise.
7499 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7500 a vector passed to build_vector.
7501 (fold_convert_const): Likewise.
7502 (exact_inverse): Likewise.
7503 (fold_ternary_loc): Likewise.
7504 (fold_relational_const): Likewise.
7505 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7506 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7507 (const_unop): Likewise. Store the reduction accumulator in a
7508 variable rather than an array.
7509 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7510 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7511 the new vector, rather than constructing it after the input arrays.
7512 (native_interpret_vector): Use auto_vec<tree> when building
7513 a vector passed to build_vector. Add elements in order.
7514 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7515 auto_vec<tree> when building a vector passed to build_vector.
7516 (vect_create_epilog_for_reduction): Likewise.
7517 (vectorizable_induction): Likewise.
7518 (get_initial_def_for_reduction): Likewise. Fix indentation of
7520 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7522 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7523 passed to build_vector.
7525 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7526 Alan Hayward <alan.hayward@arm.com>
7527 David Sherwood <david.sherwood@arm.com>
7529 * tree-core.h (tree_base::u): Add an "nelts" field.
7530 (tree_vector): Use VECTOR_CST_NELTS as the length.
7531 * tree.c (tree_size): Likewise.
7532 (make_vector): Initialize VECTOR_CST_NELTS.
7533 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7534 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7535 TYPE_VECTOR_SUBPARTS.
7536 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7537 as the number of units, setting it from VECTOR_CST_NELTS.
7538 (const_vector_from_tree): Likewise.
7539 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7540 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7541 (fold_negate_expr_1): Likewise.
7542 (fold_convert_const): Likewise.
7543 (const_binop): Likewise. Differentiate the number of output and
7545 (const_unop): Likewise.
7546 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7547 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7548 in cases that did the opposite.
7550 2017-09-14 Richard Biener <rguenther@suse.de>
7552 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7555 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7557 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7559 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7562 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7563 if and where to split a bb, except for splitting before debug insn
7564 sequences followed by non-label real insn. Delete debug insns
7565 in between basic blocks.
7567 * combine.c (make_compound_operation_int): Formatting fixes.
7569 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7570 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7571 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7572 * config/sol2.h (LINK_EH_SPEC): Likewise.
7573 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7574 * config/s390/linux.h (LINK_SPEC): Likewise.
7575 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7576 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7577 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7578 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7579 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7580 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7581 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7582 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7584 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7586 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7588 (ENDFILE_LINUX_SPEC): Likewise.
7589 (LINK_EH_SPEC): Likewise.
7590 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7591 (LINK_OS_LINUX_SPEC32): Likewise.
7592 (LINK_OS_LINUX_SPEC64): Likewise.
7593 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7594 (LINK_OS_LINUX_SPEC): Likewise.
7596 2017-09-13 Martin Liska <mliska@suse.cz>
7599 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
7600 CASE_HIGH is NULL_TREE.
7602 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7603 Alan Hayward <alan.hayward@arm.com>
7604 David Sherwood <david.sherwood@arm.com>
7606 * target.def (secondary_memory_needed): New hook.
7607 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
7608 instead of SECONDARY_MEMORY_NEEDED.
7609 (secondary_memory_needed_mode): Likewise.
7610 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
7611 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
7612 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
7613 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
7614 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7615 * doc/tm.texi: Regenerate.
7616 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7617 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7618 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7619 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7620 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7621 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7622 mode argument first and change the reg_class arguments to reg_class_t.
7623 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7624 Make static. Update the call to inline_secondary_memory_needed.
7625 (ix86_register_move_cost): Update the call to
7626 inline_secondary_memory_needed.
7627 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7628 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7630 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7631 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7633 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7634 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7635 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7636 and match hook interface. Add comment from mips.h.
7637 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7638 * config/mmix/mmix.md (truncdfsf2): Refer to
7639 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7641 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7642 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7643 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7644 (pa_secondary_memory_needed): New function.
7645 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7646 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7647 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7648 (pdp11_secondary_memory_needed): Make static and match hook interface.
7649 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7650 * config/powerpcspe/powerpcspe-protos.h
7651 (rs6000_secondary_memory_needed_ptr): Delete.
7652 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7654 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7655 (rs6000_option_override_internal): Assign to
7656 targetm.secondary_memory_needed rather than
7657 rs6000_secondary_memory_needed_ptr.
7658 (rs6000_secondary_memory_needed): Match hook interface.
7659 (rs6000_debug_secondary_memory_needed): Likewise.
7660 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7661 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7662 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7663 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7664 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7665 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7667 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7668 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7669 (rs6000_option_override_internal): Assign to
7670 targetm.secondary_memory_needed rather than
7671 rs6000_secondary_memory_needed_ptr.
7672 (rs6000_secondary_memory_needed): Match hook interface.
7673 (rs6000_debug_secondary_memory_needed): Likewise.
7674 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7675 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7676 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7677 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7678 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7679 (sparc_secondary_memory_needed): New function.
7680 * lra-constraints.c (check_and_process_move): Refer to
7681 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7683 (curr_insn_transform): Likewise.
7684 (process_alt_operands): Use targetm.secondary_memory_needed
7685 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7686 (check_secondary_memory_needed_p): Likewise.
7687 (choose_split_class): Likewise.
7688 * reload.c: Unconditionally include code that was previously
7689 conditional on SECONDARY_MEMORY_NEEDED.
7690 (push_secondary_reload): Use targetm.secondary_memory_needed
7691 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7692 (push_reload): Likewise.
7693 * reload1.c: Unconditionally include code that was previously
7694 conditional on SECONDARY_MEMORY_NEEDED.
7695 (choose_reload_regs): Use targetm.secondary_memory_needed
7696 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7697 (gen_reload): Likewise.
7698 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7700 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7701 Alan Hayward <alan.hayward@arm.com>
7702 David Sherwood <david.sherwood@arm.com>
7704 * target.def (secondary_memory_needed_mode): New hook:
7705 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7706 * targhooks.h (default_secondary_memory_needed_mode): New function.
7707 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7708 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7709 * doc/tm.texi: Regenerate.
7710 * lra-constraints.c (check_and_process_move): Use
7711 targetm.secondary_memory_needed_mode instead of
7712 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7713 (curr_insn_transform): Likewise.
7714 * reload.c (get_secondary_mem): Likewise.
7715 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7716 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7718 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7719 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7720 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7721 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7722 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7724 * config/powerpcspe/powerpcspe-protos.h
7725 (rs6000_secondary_memory_needed_mode): Delete.
7726 * config/powerpcspe/powerpcspe.c
7727 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7728 (rs6000_secondary_memory_needed_mode): Make static.
7729 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7730 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7732 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7734 (rs6000_secondary_memory_needed_mode): Make static.
7735 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7736 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7737 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7738 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7739 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7741 (sparc_secondary_memory_needed_mode): New function.
7742 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7744 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7746 * config/aarch64/constraints.md (Umq): New constraint.
7747 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7749 (mov<mode>): Update condition.
7751 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7753 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7754 when bitposition is the same.
7756 2017-09-13 Richard Biener <rguenther@suse.de>
7758 * dwarf2out.c (output_die_symbol): Remove.
7759 (output_die): Do not output a DIEs symbol.
7761 2017-09-13 Richard Biener <rguenther@suse.de>
7764 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7765 default-def to avoid breaking iterator update with the weird
7766 interaction with cgraph_update_edges_for_call_stmt_node.
7768 2017-09-13 Richard Biener <rguenther@suse.de>
7770 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7771 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7772 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7773 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7775 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7777 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7778 Disable pc relative literal load irrespective of
7779 TARGET_FIX_ERR_A53_84341 for default.
7781 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7783 * config/sparc/sparc.c (output_return): Output the source location of
7784 the insn in the delay slot, if any.
7785 (output_sibcall): Likewise.
7787 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7790 * common.opt (-static-pie): New alias.
7791 (shared): Negate static-pie.
7792 (-no-pie): Update help text.
7794 (static-pie): New option.
7795 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7796 -static-pie support.
7797 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7798 (LINK_EH_SPEC): Likewise.
7799 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7800 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7801 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7802 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7803 (init_gcc_specs): Likewise.
7804 (init_spec): Likewise.
7805 (display_help): Update help message for -pie.
7806 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7809 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7811 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7812 (movdi_aarch64): Likewise.
7813 (movti_aarch64): Likewise.
7815 2017-09-12 Simon Wright <simon@pushface.org>
7818 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7819 calculation of the minor version, always output as 0.
7821 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7824 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7825 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7826 on it early, rather than manual conversion late. For
7827 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7828 instead of performing manual conversion.
7830 2017-09-12 Carl Love <cel@us.ibm.com>
7832 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7833 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7835 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7836 VMULOSW): Add definitions.
7837 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7838 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7839 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7840 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7841 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7843 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7845 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7847 (movti_aarch64): Likewise.
7848 (movdf_aarch64): Likewise.
7849 (movtf_aarch64): Likewise.
7850 (load_pairdi): Likewise.
7851 (store_pairdi): Likewise.
7852 (load_pairdf): Likewise.
7853 (store_pairdf): Likewise.
7854 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7855 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7856 (ldr_got_small_<mode>): Likewise.
7857 (ldr_got_small_28k_<mode>): Likewise.
7858 (ldr_got_tiny): Likewise.
7859 * config/aarch64/iterators.md (ldst_sz): New.
7860 (ldpstp_sz): Likewise.
7861 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7863 (thunderx_load): Split load_8 to load_16.
7864 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7866 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7867 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7868 (xgene1_store_pair): Split store_8 to store_16.
7869 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7870 (falkor_st_0_st_sd): Split store_8 to store_16.
7872 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7874 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7875 and store1/2/3/4 to store_4/8/12/16.
7876 * config/aarch64/aarch64.md: Update for rename.
7877 * config/arm/arm.md: Likewise.: Likewise.
7878 * config/arm/arm.c: Likewise.
7879 * config/arm/thumb1.md: Likewise.
7880 * config/arm/thumb2.md: Likewise.
7881 * config/arm/vfp.md: Likewise.
7882 * config/arm/arm-generic.md: Likewise.
7883 * config/arm/arm1020e.md: Likewise.
7884 * config/arm/arm1026ejs.md: Likewise.
7885 * config/arm/arm1136jfs.md: Likewise.
7886 * config/arm/arm926ejs.md: Likewise.
7887 * config/arm/cortex-a15.md: Likewise.
7888 * config/arm/cortex-a17.md: Likewise.
7889 * config/arm/cortex-a5.md: Likewise.
7890 * config/arm/cortex-a53.md: Likewise.
7891 * config/arm/cortex-a57.md: Likewise.
7892 * config/arm/cortex-a7.md: Likewise.
7893 * config/arm/cortex-a8.md: Likewise.
7894 * config/arm/cortex-a9.md: Likewise.
7895 * config/arm/cortex-m4.md: Likewise.
7896 * config/arm/cortex-m7.md: Likewise.
7897 * config/arm/cortex-r4.md: Likewise.
7898 * config/arm/exynos-m1.md: Likewise.
7899 * config/arm/fa526.md: Likewise.
7900 * config/arm/fa606te.md: Likewise.
7901 * config/arm/fa626te.md: Likewise.
7902 * config/arm/fa726te.md: Likewise.
7903 * config/arm/fmp626.md: Likewise.
7904 * config/arm/iwmmxt.md: Likewise.
7905 * config/arm/ldmstm.md: Likewise.
7906 * config/arm/marvell-pj4.md: Likewise.
7907 * config/arm/xgene1.md: Likewise.
7908 * config/aarch64/thunderx.md: Likewise.
7909 * config/aarch64/thunderx2t99.md: Likewise.
7910 * config/aarch64/falkor.md: Likewise.
7912 2017-09-12 Martin Liska <mliska@suse.cz>
7914 * attribs.c (private_lookup_attribute): New function.
7915 * attribs.h (private_lookup_attribute): Declared here.
7916 (lookup_attribute): Called from this place.
7918 2017-09-12 Richard Biener <rguenther@suse.de>
7920 PR tree-optimization/82157
7921 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
7922 stmts with side-effects.
7924 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7925 Alan Hayward <alan.hayward@arm.com>
7926 David Sherwood <david.sherwood@arm.com>
7928 * target.def (hard_regno_nregs): New hook.
7929 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
7930 * targhooks.h (default_hard_regno_nregs): Declare.
7931 * targhooks.c (default_hard_regno_nregs): New function.
7932 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
7933 (TARGET_HARD_REGNO_NREGS): ...this hook.
7934 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
7935 (CLASS_MAX_NREGS): Likewise.
7936 * doc/tm.texi: Regenerate.
7937 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
7938 instead of HARD_REGNO_NREGS.
7939 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
7940 HARD_REGNO_NREGS in the comment.
7941 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
7942 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
7943 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
7944 Return an unsigned int.
7945 (TARGET_HARD_REGNO_NREGS): Redefine.
7946 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
7947 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
7948 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
7949 (arc_hard_regno_nregs): New function.
7950 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
7951 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
7952 (arm_hard_regno_nregs): New function.
7953 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
7954 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
7955 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
7956 (TARGET_HARD_REGNO_NREGS): Redefine.
7957 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
7958 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
7959 (HARD_REGNO_NREGS): Delete.
7960 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
7961 (cr16_hard_regno_nregs): New function.
7962 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
7963 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
7964 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
7965 (cris_hard_regno_nregs): New function.
7966 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
7967 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
7968 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
7969 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
7970 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
7971 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
7972 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
7973 (frv_hard_regno_nregs): Make static. Take and return an
7975 (frv_class_max_nregs): Remove outdated copy of documentation.
7976 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
7977 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
7978 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
7979 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
7980 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
7981 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
7982 (TARGET_HARD_REGNO_NREGS): Redefine.
7983 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
7984 (CLASS_MAX_NREGS): Update comment.
7985 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
7986 (ia64_hard_regno_nregs): New function.
7987 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
7988 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
7989 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
7990 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
7991 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
7993 (m32c_hard_regno_nregs): Likewise. Make static.
7994 (TARGET_HARD_REGNO_NREGS): Redefine.
7995 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
7996 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
7997 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
7998 (m68k_hard_regno_nregs): New function.
7999 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8000 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8001 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8002 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8003 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8004 Take and return an unsigned int.
8005 (TARGET_HARD_REGNO_NREGS): Redefine.
8006 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8007 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8008 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8009 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8010 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8011 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8012 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8013 (msp430_hard_regno_nregs): Make static. Take and return an
8015 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8016 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8017 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8018 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8019 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8020 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8021 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8022 (TARGET_HARD_REGNO_NREGS): Redefine.
8023 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8024 (PA_HARD_REGNO_NREGS): ...this.
8025 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8026 (PA_HARD_REGNO_NREGS): ...this.
8027 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8028 (pa_hard_regno_nregs): New function.
8029 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8030 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8031 (pdp11_hard_regno_nregs): New function.
8032 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8033 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8034 (rs6000_hard_regno_nregs_hook): New function.
8035 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8036 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8037 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8038 Take and return an unsigned int. Move earlier in file.
8039 (TARGET_HARD_REGNO_NREGS): Redefine.
8040 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8041 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8042 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8043 (rl78_hard_regno_nregs): Make static. Take and return an
8045 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8046 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8047 (rs6000_hard_regno_nregs_hook): New function.
8048 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8049 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8050 (TARGET_HARD_REGNO_NREGS): Redefine.
8051 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8052 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8053 instead of HARD_REGNO_NREGS.
8054 (s390_hard_regno_nregs): New function.
8055 (s390_hard_regno_mode_ok): Add comment from s390.h.
8056 (TARGET_HARD_REGNO_NREGS): Redefine.
8057 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8058 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8059 (sh_hard_regno_nregs): New function.
8060 (sh_pass_in_reg_p): Use it.
8061 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8062 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8063 (sparc_hard_regno_nregs): New function.
8064 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8065 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8066 (spu_function_arg_advance): Use it, supplying a valid register number.
8067 (TARGET_HARD_REGNO_NREGS): Redefine.
8068 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8069 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8070 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8071 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8072 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8073 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8074 (CLASS_MAX_NREGS): Remove copy of old documentation.
8075 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8076 (visium_hard_regno_nregs): New function.
8077 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8078 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8079 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8080 xtensa_hard_regno_nregs): New function.
8081 * system.h (HARD_REGNO_NREGS): Poison.
8083 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8085 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8086 hard_regno_nregs instead of HARD_REGNO_NREGS.
8087 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8088 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8089 (c6x_expand_epilogue): Likewise.
8090 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8091 (frv_read_iacc_argument): Likewise.
8092 * config/sh/sh.c: Include regs.h.
8093 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8094 (regs_used): Likewise.
8095 (output_stack_adjust): Likewise.
8096 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8097 * expmed.c: Include regs.h.
8098 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8099 * ree.c: Include regs.h.
8100 (combine_reaching_defs): Use hard_regno_nregs instead of
8102 (add_removable_extension): Likewise.
8104 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8106 * regs.h (hard_regno_nregs): Turn into a function.
8107 (end_hard_regno): Update accordingly.
8108 * caller-save.c (setup_save_areas): Likewise.
8109 (save_call_clobbered_regs): Likewise.
8110 (replace_reg_with_saved_mem): Likewise.
8111 (insert_restore): Likewise.
8112 (insert_save): Likewise.
8113 * combine.c (can_change_dest_mode): Likewise.
8114 (move_deaths): Likewise.
8115 (distribute_notes): Likewise.
8116 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
8117 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
8118 (rs6000_split_multireg_move): Likewise.
8119 (rs6000_register_move_cost): Likewise.
8120 (rs6000_memory_move_cost): Likewise.
8121 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
8122 (rs6000_split_multireg_move): Likewise.
8123 (rs6000_register_move_cost): Likewise.
8124 (rs6000_memory_move_cost): Likewise.
8125 * cselib.c (cselib_reset_table): Likewise.
8126 (cselib_lookup_1): Likewise.
8127 * emit-rtl.c (set_mode_and_regno): Likewise.
8128 * function.c (aggregate_value_p): Likewise.
8129 * ira-color.c (setup_profitable_hard_regs): Likewise.
8130 (check_hard_reg_p): Likewise.
8131 (calculate_saved_nregs): Likewise.
8132 (assign_hard_reg): Likewise.
8133 (improve_allocation): Likewise.
8134 (calculate_spill_cost): Likewise.
8135 * ira-emit.c (modify_move_list): Likewise.
8136 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
8137 (ira_hard_reg_in_set_p): Likewise.
8138 * ira.c (setup_reg_mode_hard_regset): Likewise.
8139 (clarify_prohibited_class_mode_regs): Likewise.
8140 (check_allocation): Likewise.
8141 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8142 (lra_setup_reg_renumber): Likewise.
8143 (setup_try_hard_regno_pseudos): Likewise.
8144 (spill_for): Likewise.
8145 (assign_hard_regno): Likewise.
8146 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
8147 * lra-constraints.c (in_class_p): Likewise.
8148 (lra_constraint_offset): Likewise.
8149 (simplify_operand_subreg): Likewise.
8150 (lra_constraints): Likewise.
8151 (split_reg): Likewise.
8152 (split_if_necessary): Likewise.
8153 (invariant_p): Likewise.
8154 (inherit_in_ebb): Likewise.
8155 * lra-lives.c (process_bb_lives): Likewise.
8156 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
8157 (get_hard_regs): Likewise.
8158 (do_remat): Likewise.
8159 * lra-spills.c (assign_spill_hard_regs): Likewise.
8160 * mode-switching.c (create_pre_exit): Likewise.
8161 * postreload.c (reload_combine_recognize_pattern): Likewise.
8162 * recog.c (peep2_find_free_register): Likewise.
8163 * regcprop.c (kill_value_regno): Likewise.
8164 (set_value_regno): Likewise.
8165 (copy_value): Likewise.
8166 (maybe_mode_change): Likewise.
8167 (find_oldest_value_reg): Likewise.
8168 (copyprop_hardreg_forward_1): Likewise.
8169 * regrename.c (check_new_reg_p): Likewise.
8170 (regrename_do_replace): Likewise.
8171 * reload.c (push_reload): Likewise.
8172 (combine_reloads): Likewise.
8173 (find_dummy_reload): Likewise.
8174 (operands_match_p): Likewise.
8175 (find_reloads): Likewise.
8176 (find_equiv_reg): Likewise.
8177 (reload_adjust_reg_for_mode): Likewise.
8178 * reload1.c (count_pseudo): Likewise.
8179 (count_spilled_pseudo): Likewise.
8180 (find_reg): Likewise.
8181 (clear_reload_reg_in_use): Likewise.
8182 (free_for_value_p): Likewise.
8183 (allocate_reload_reg): Likewise.
8184 (choose_reload_regs): Likewise.
8185 (reload_adjust_reg_for_temp): Likewise.
8186 (emit_reload_insns): Likewise.
8187 (delete_output_reload): Likewise.
8188 * rtlanal.c (subreg_get_info): Likewise.
8189 * sched-deps.c (sched_analyze_reg): Likewise.
8190 * sel-sched.c (init_regs_for_mode): Likewise.
8191 (mark_unavailable_hard_regs): Likewise.
8192 (choose_best_reg_1): Likewise.
8193 (verify_target_availability): Likewise.
8194 * valtrack.c (dead_debug_insert_temp): Likewise.
8195 * var-tracking.c (track_loc_p): Likewise.
8196 (emit_note_insn_var_location): Likewise.
8197 * varasm.c (make_decl_rtl): Likewise.
8198 * reginfo.c (choose_hard_reg_mode): Likewise.
8199 (init_reg_modes_target): Refer directly to
8200 this_target_regs->x_hard_regno_nregs.
8202 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8204 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8205 instead of hard_regno_nregs.
8207 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8209 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8210 end_hard_regno instead of hard_regno_nregs.
8211 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8212 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8213 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8214 * ira-color.c (improve_allocation): Likewise.
8215 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8216 * lra-lives.c (mark_regno_live): Likewise.
8217 (mark_regno_dead): Likewise.
8218 * lra-remat.c (operand_to_remat): Likewise.
8219 * lra.c (collect_non_operand_hard_regs): Likewise.
8220 * postreload.c (reload_combine_note_store): Likewise.
8221 (move2add_valid_value_p): Likewise.
8222 * reload.c (regno_clobbered_p): Likewise.
8224 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8226 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8228 * config/v850/v850.c (v850_reorg): Likewise.
8229 * reload.c (refers_to_regno_for_reload_p): Likewise.
8230 (find_equiv_reg): Likewise.
8231 * reload1.c (reload_reg_reaches_end_p): Likewise.
8233 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8235 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8237 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8238 * config/arm/arm.c (output_move_neon): Likewise.
8239 (arm_attr_length_move_neon): Likewise.
8240 (neon_split_vcombine): Likewise.
8241 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8242 (c6x_mark_reg_written): Likewise.
8243 (c6x_dwarf_register_span): Likewise.
8244 * config/i386/i386.c (ix86_save_reg): Likewise.
8245 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8246 (rws_access_reg): Likewise.
8247 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8248 * mode-switching.c (create_pre_exit): Likewise.
8249 * ree.c (combine_reaching_defs): Likewise.
8250 (add_removable_extension): Likewise.
8251 * regcprop.c (find_oldest_value_reg): Likewise.
8252 (copyprop_hardreg_forward_1): Likewise.
8253 * reload.c (reload_inner_reg_of_subreg): Likewise.
8254 (push_reload): Likewise.
8255 (combine_reloads): Likewise.
8256 (find_dummy_reload): Likewise.
8257 (reload_adjust_reg_for_mode): Likewise.
8258 * reload1.c (find_reload_regs): Likewise.
8259 (forget_old_reloads_1): Likewise.
8260 (reload_reg_free_for_value_p): Likewise.
8261 (reload_adjust_reg_for_temp): Likewise.
8262 (emit_reload_insns): Likewise.
8263 (delete_output_reload): Likewise.
8264 * sel-sched.c (choose_best_reg_1): Likewise.
8265 (choose_best_pseudo_reg): Likewise.
8267 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8268 Alan Hayward <alan.hayward@arm.com>
8269 David Sherwood <david.sherwood@arm.com>
8271 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8272 * target.def (slow_unaligned_access): New hook.
8273 * targhooks.h (default_slow_unaligned_access): Declare.
8274 * targhooks.c (default_slow_unaligned_access): New function.
8275 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8276 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8277 * doc/tm.texi: Regenerate.
8278 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8279 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8280 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8282 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8283 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8285 (rs6000_slow_unaligned_access): New function.
8286 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8287 (expand_block_compare): Likewise.
8288 (expand_strn_compare): Likewise.
8289 (rs6000_rtx_costs): Likewise.
8290 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8291 (riscv_slow_unaligned_access): Likewise.
8292 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8293 (riscv_slow_unaligned_access_p): ...this and make static.
8294 (riscv_option_override): Update accordingly.
8295 (riscv_slow_unaligned_access): New function.
8296 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8297 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8298 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8299 (rs6000_slow_unaligned_access): New function.
8300 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8301 (rs6000_rtx_costs): Likewise.
8302 * config/rs6000/rs6000-string.c (expand_block_compare)
8303 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8304 of SLOW_UNALIGNED_ACCESS.
8305 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8306 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8307 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8308 of SLOW_UNALIGNED_ACCESS.
8309 * expmed.c (simple_mem_bitfield_p): Likewise.
8310 * expr.c (alignment_for_piecewise_move): Likewise.
8311 (emit_group_load_1): Likewise.
8312 (emit_group_store): Likewise.
8313 (copy_blkmode_from_reg): Likewise.
8314 (emit_push_insn): Likewise.
8315 (expand_assignment): Likewise.
8316 (store_field): Likewise.
8317 (expand_expr_real_1): Likewise.
8318 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8319 * lra-constraints.c (simplify_operand_subreg): Likewise.
8320 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8321 * gimple-ssa-store-merging.c: Likewise in block comment at start
8323 * tree-ssa-strlen.c: Include target.h.
8324 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8325 of SLOW_UNALIGNED_ACCESS.
8326 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8328 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8330 PR rtl-optimization/82185
8331 * expmed.c (emit_store_flag_int): Only test tem if it has been
8334 2017-09-12 Richard Biener <rguenther@suse.de>
8337 * match.pd ((FTYPE) N CMP CST): Fix typo.
8339 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8341 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8343 (mips_near_type_p): Add 'short_call' attribute as a synonym
8345 * doc/extend.texi (short_call): Document new function attribute.
8347 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8350 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8351 assertion check that in the condition.
8352 (get_atomic_generic_size): Likewise. Before testing if parameter
8353 has pointer type, if it has array type, call for C++
8354 default_conversion to perform array-to-pointer conversion.
8356 2017-09-12 Richard Biener <rguenther@suse.de>
8358 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8359 for operations we cannot scalarize.
8361 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8363 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8364 vectors heap vectors. Clean up comments.
8365 Make visited_bbs a reference.
8366 (profitable_jump_thread_path): Make GC
8367 vectors heap vectors. Clean up comments.
8369 (convert_and_register_jump_thread_path): Make GC vectors heap
8371 (check_subpath_and_update_thread_path): Same. Clean up comments.
8372 Make visited_bbs a reference.
8373 (handle_phi): Abstract common code to to
8374 register_jump_thread_path_if_profitable.
8375 Rename VAR_BB to DEF_BB.
8377 Make GC vectors heap vectors.
8378 Make visited_bbs a reference.
8379 (handle_assignment): Same.
8380 (register_jump_thread_path_if_profitable): New.
8381 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8383 Make GC vectors heap vectors. Clean up comments.
8384 Make visited_bbs a reference.
8385 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8386 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8389 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8392 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8393 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8395 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8397 Revert r251800 and r251799.
8399 2017-09-11 Martin Jambor <mjambor@suse.cz>
8402 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8403 arguments in advance.
8404 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8405 use it to find predecessor edges.
8406 (naive_outof_ssa): Collect vector of predecessors.
8408 2017-09-08 Jason Merrill <jason@redhat.com>
8410 PR c++/70029 - ICE with ref-qualifier and -flto
8411 * langhooks.h (struct lang_hooks_for_types): Add
8412 copy_lang_qualifiers.
8413 * attribs.c (build_type_attribute_qual_variant): Use it.
8414 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8416 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8417 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8419 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8422 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8423 (*mulsi3_sp64): New instruction.
8424 (mulsi3): New expander.
8426 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8428 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8430 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8432 * sancov.c: Include memmodel.h.
8434 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8437 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8440 2017-09-07 Carl Love <cel@us.ibm.com>
8442 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8443 the sldi instruction.
8445 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8447 * sancov.c: Include tm_p.h.
8449 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8452 * output.h (switch_to_other_text_partition): New declaration.
8453 * varasm.c (switch_to_other_text_partition): New function.
8454 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8455 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8456 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8457 to the other text partition before emitting LCL label and switch back
8458 after emitting the word after it.
8460 2017-09-07 Richard Biener <rguenther@suse.de>
8462 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8463 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8464 critical edges here, after loop init.
8465 (pass_data_pre): Remove PROP_no_crit_edges flags.
8466 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8467 for valueization of call args to avoid leaking VN_TOP.
8468 (visit_use): Assert we do not visit default defs.
8469 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8470 Use error_mark_node to more easily detect leaking VN_TOP.
8471 All default-defs are varying, not VN_TOP. Mark them visited.
8472 (run_scc_vn): Make code match comment.
8474 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8476 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8477 OPTION_MASK_FLOAT128_KEYWORD.
8478 (POWERPC_MASKS): Likewise.
8479 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8480 support for the -mfloat128-type option, and make -mfloat128
8481 default on PowerPC Linux systems. Define or undefine
8482 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8483 Define __float128 to be __ieee128 if IEEE 128-bit support is
8484 enabled, or undefine it.
8485 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8486 Delete defining __FLOAT128_TYPE__.
8487 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8488 -mfloat128-type option and make -mfloat128 default on PowerPC
8490 (TARGET_FLOAT128_TYPE): Likewise.
8491 (-mfloat128-type): Likewise.
8492 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8493 Delete the -mfloat128-type option and make -mfloat128 default on
8494 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8495 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8496 128-bit floating point is enabled. Change tests from using
8497 -mfloat128-type to -mfloat128.
8498 (rs6000_mangle_type): Use the correct mangling for the __float128
8499 type even if normal long double is restricted to 64-bits.
8500 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8502 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8503 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8504 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8506 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8507 documentation for -mfloat128.
8509 2017-09-06 Olivier Hainque <hainque@adacore.com>
8511 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8513 2017-09-06 Wish Wu <wishwu007@gmail.com>
8514 Jakub Jelinek <jakub@redhat.com>
8516 * asan.c (initialize_sanitizer_builtins): Add
8517 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8518 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8519 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8520 BT_FN_VOID_UINT64_PTR variables.
8521 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8522 (BT_FN_VOID_UINT16_UINT16): Likewise.
8523 (BT_FN_VOID_UINT32_UINT32): Likewise.
8524 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8525 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8526 (BT_FN_VOID_UINT64_PTR): Likewise.
8527 * common.opt (flag_sanitize_coverage): New variable.
8528 (fsanitize-coverage=trace-pc): Remove.
8529 (fsanitize-coverage=): Add.
8530 * flag-types.h (enum sanitize_coverage_code): New enum.
8531 * fold-const.c (fold_range_test): Disable non-short-circuit
8532 optimization if flag_sanitize_coverage.
8533 (fold_truth_andor): Likewise.
8534 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8535 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8536 (coverage_sanitizer_opts): New array.
8537 (get_closest_sanitizer_option): Add OPTS argument, handle also
8538 OPT_fsanitize_coverage_.
8539 (parse_sanitizer_options): Adjusted to also handle
8540 OPT_fsanitize_coverage_.
8541 (common_handle_option): Add OPT_fsanitize_coverage_.
8542 * sancov.c (instrument_comparison, instrument_switch): New function.
8543 (sancov_pass): Add trace-cmp support.
8544 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8545 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8546 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8547 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8548 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8549 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8550 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8551 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8552 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8553 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8555 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8557 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8558 error. Only quit immediately if parsing is complete.
8559 (BEGIN): Initialize fatal_err and parse_done.
8560 (begin fpu, end fpu): Check number of arguments.
8561 (begin arch, end arch): Likewise.
8562 (begin cpu, end cpu): Likewise.
8563 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8564 (optalias): Likewise.
8566 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8568 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8569 * config/arm/arm-isa.h: Delete. Move definitions to ...
8570 * arm-cpus.in: ... here. Use new feature and fgroup values.
8571 * config/arm/arm.c (arm_option_override): Use lower case for feature
8573 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8574 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8575 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8577 (print_isa_bits_for): New function.
8578 (gen_isa): New function.
8579 (gen_comm_data): Use print_isa_bits_for.
8580 (define feature): New keyword.
8581 (define fgroup): New keyword.
8582 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8583 (arm-isa.h): Add rule to generate file.
8584 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8585 case for feature bit names.
8587 2017-09-06 Richard Biener <rguenther@suse.de>
8589 * tree-ssa-pre.c (NECESSARY): Remove.
8590 (create_expression_by_pieces): Do not touch pass-local flags.
8591 (insert_into_preds_of_block): Likewise.
8592 (do_pre_regular_insertion): Likewise.
8593 (eliminate_insert): Likewise.
8594 (eliminate_dom_walker::before_dom_children): Likewise.
8595 (fini_eliminate): Do not look at inserted_exprs.
8596 (mark_operand_necessary): Remove.
8597 (remove_dead_inserted_code): Replace with simple work-list
8598 algorithm based on inserted_exprs and SSA uses.
8599 (pass_pre::execute): Re-order fini_eliminate and
8600 remove_dead_inserted_code.
8602 2017-09-06 Olivier Hainque <hainque@adacore.com>
8604 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
8605 for VxWorks 7. Adjust surrounding comments.
8607 2017-09-06 Richard Biener <rguenther@suse.de>
8609 * gimple-ssa-strength-reduction.c
8610 (find_candidates_dom_walker::before_dom_children): Also allow
8613 2017-09-06 Richard Biener <rguenther@suse.de>
8615 PR tree-optimization/82108
8616 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8617 for gap in the non-permutation SLP case.
8619 2017-09-06 Martin Jambor <mjambor@suse.cz>
8621 PR tree-optimization/82078
8622 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8623 add_access_to_work_queue...
8624 (build_accesses_from_assign): ...here.
8625 (propagate_all_subaccesses): Make sure racc is the group
8626 representative, if there is one.
8628 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8631 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8634 2017-09-06 Richard Biener <rguenther@suse.de>
8636 * gimple-ssa-strength-reduction.c
8637 (find_candidates_dom_walker::before_doom_children): Use a
8638 type and not a mode check.
8640 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8643 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8645 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8647 2017-09-05 Jeff Law <law@redhat.com>
8649 PR tree-optimization/64910
8650 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8651 cases where we have 3 or more operands.
8653 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8656 * omp-low.c (lower_omp_for): Recompute tree invariant if
8657 gimple_omp_for_initial/final is ADDR_EXPR.
8660 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8661 into gimple val before gimplification fo the COND_EXPR.
8663 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8665 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8666 REGION_COPY argument.
8667 (thread_through_all_blocks): Remove unused argument to
8668 duplicate_thread_path.
8670 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8671 Alan Hayward <alan.hayward@arm.com>
8672 David Sherwood <david.sherwood@arm.com>
8674 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8675 Take a scalar_mode rather than a machine_mode.
8676 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8677 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8678 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8679 (aarch64_gen_adjusted_ldpstp): Likewise.
8680 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8682 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8683 Alan Hayward <alan.hayward@arm.com>
8684 David Sherwood <david.sherwood@arm.com>
8686 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8687 Take a scalar_int_mode instead of a machine_mode.
8688 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8689 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8690 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8691 (aarch64_simd_attr_length_rglist): Delete.
8692 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8693 a scalar_int_mode instead of a machine_mode.
8694 (aarch64_add_offset): Likewise.
8695 (aarch64_internal_mov_immediate): Likewise
8696 (aarch64_add_constant_internal): Likewise.
8697 (aarch64_add_constant): Likewise.
8698 (aarch64_movw_imm): Likewise.
8699 (aarch64_rtx_arith_op_extract_p): Likewise.
8700 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8701 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8702 Remove assert that the mode isn't a vector.
8703 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8704 (aarch64_expand_mov_immediate): Update calls after above changes.
8705 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8706 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8707 (aarch64_move_imm): Likewise.
8708 (aarch64_can_const_movi_rtx_p): Likewise.
8709 (aarch64_strip_extend): Likewise.
8710 (aarch64_extr_rtx_p): Likewise.
8711 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8712 a CONST_INT when the mode parameter is VOIDmode.
8713 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8715 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8717 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8718 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8719 (bitwise_type_for_mode): Update accordingly.
8721 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8723 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8724 * stor-layout.c (mode_for_size_tree): Likewise.
8725 (mode_for_array): Update accordingly.
8726 (layout_decl): Likewise.
8727 (compute_record_mode): Likewise. Only set the mode once.
8729 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8731 * target.def (get_mask_mode): Change return type to opt_mode.
8733 * doc/tm.texi: Regenerate.
8734 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8735 * targhooks.c (default_get_mask_mode): Likewise.
8736 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8737 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8738 targetm.get_mask_mode.
8739 * tree.c (build_truth_vector_type): Likewise.
8741 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8743 * machmode.h (mode_for_vector): Return an opt_mode.
8744 * stor-layout.c (mode_for_vector): Likewise.
8745 (mode_for_int_vector): Update accordingly.
8746 (layout_type): Likewise.
8747 * config/i386/i386.c (emit_memmov): Likewise.
8748 (ix86_expand_set_or_movmem): Likewise.
8749 (ix86_expand_vector_init): Likewise.
8750 (ix86_get_mask_mode): Likewise.
8751 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8753 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8754 * expmed.c (extract_bit_field_1): Likewise.
8755 * expr.c (expand_expr_real_2): Likewise.
8756 * optabs-query.c (can_vec_perm_p): Likewise.
8757 (can_vec_mask_load_store_p): Likewise.
8758 * optabs.c (expand_vec_perm): Likewise.
8759 * targhooks.c (default_get_mask_mode): Likewise.
8760 * tree-vect-stmts.c (vectorizable_store): Likewise.
8761 (vectorizable_load): Likewise.
8762 (get_vectype_for_scalar_type_and_size): Likewise.
8764 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8766 * machmode.h (mode_for_int_vector): New function.
8767 * stor-layout.c (mode_for_int_vector): Likewise.
8768 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8769 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8770 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8771 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8772 (s390_expand_vcond): Likewise.
8774 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8776 * machmode.h (opt_machine_mode): New type.
8777 (opt_mode<T>): Allow construction from anything that can be
8779 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8780 (mode_for_size): Return an opt_machine_mode.
8781 * stor-layout.c (mode_for_size): Likewise.
8782 (mode_for_size_tree): Update call accordingly.
8783 (bitwise_mode_for_mode): Likewise.
8784 (make_fract_type): Likewise.
8785 (make_accum_type): Likewise.
8786 * caller-save.c (replace_reg_with_saved_mem): Update call
8788 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8789 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8790 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8791 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8792 * expmed.c (extract_bit_field_1): Likewise.
8793 * reload.c (get_secondary_mem): Likewise.
8794 * varasm.c (assemble_integer): Likewise.
8795 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8798 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8800 * machmode.h (decimal_float_mode_for_size): New function.
8801 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8802 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8803 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8804 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8805 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8806 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8808 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8810 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8811 (get_builtin_sync_mode): Likewise.
8812 (expand_ifn_atomic_compare_exchange): Likewise.
8813 (expand_builtin_atomic_clear): Likewise.
8814 (expand_builtin_atomic_test_and_set): Likewise.
8815 (fold_builtin_atomic_always_lock_free): Likewise.
8816 * calls.c (compute_argument_addresses): Likewise.
8817 (emit_library_call_value_1): Likewise.
8818 (store_one_arg): Likewise.
8819 * combine.c (combine_instructions): Likewise.
8820 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8821 * config/arm/arm.c (arm_function_value): Likewise.
8822 (aapcs_allocate_return_reg): Likewise.
8823 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8824 * config/i386/i386.c (construct_container): Likewise.
8825 (ix86_gimplify_va_arg): Likewise.
8826 (ix86_expand_sse_cmp): Likewise.
8827 (emit_memmov): Likewise.
8828 (emit_memset): Likewise.
8829 (expand_small_movmem_or_setmem): Likewise.
8830 (ix86_expand_pextr): Likewise.
8831 (ix86_expand_pinsr): Likewise.
8832 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8833 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8835 * config/mips/mips.c (mips_function_value_1) Likewise.
8836 (mips_block_move_straight): Likewise.
8837 (mips_expand_ins_as_unaligned_store): Likewise.
8838 * config/powerpcspe/powerpcspe.c
8839 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8840 (rs6000_darwin64_record_arg_flush): Likewise.
8841 * config/rs6000/rs6000.c
8842 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8843 (rs6000_darwin64_record_arg_flush): Likewise.
8844 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8845 (sparc_function_value_1): Likewise.
8846 * config/spu/spu.c (adjust_operand): Likewise.
8847 (spu_emit_branch_or_set): Likewise.
8848 (arith_immediate_p): Likewise.
8849 * emit-rtl.c (gen_lowpart_common): Likewise.
8850 * expr.c (expand_expr_real_1): Likewise.
8851 * function.c (assign_parm_setup_block): Likewise.
8852 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8853 * reload1.c (alter_reg): Likewise.
8854 * stor-layout.c (mode_for_vector): Likewise.
8855 (layout_type): Likewise.
8857 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8859 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8860 (spu_convert_move): Likewise.
8861 * lower-subreg.c (resolve_simple_move): Likewise.
8863 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8866 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8867 define_insn to a define_expand.
8868 (altivec_vsum2sws_direct): New define_insn.
8869 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8871 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8873 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8876 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8881 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8882 realign stack if stack alignment needed is less than incoming
8885 2017-09-05 Marek Polacek <polacek@redhat.com>
8888 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8891 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8893 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8895 2017-09-05 Richard Biener <rguenther@suse.de>
8897 PR tree-optimization/82084
8898 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8900 2017-09-05 Richard Biener <rguenther@suse.de>
8902 PR tree-optimization/82102
8903 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8905 2017-09-05 Martin Liska <mliska@suse.cz>
8907 PR tree-optimization/82032
8908 * tree-cfg.c (generate_range_test): New function.
8909 * tree-cfg.h (generate_range_test): Declared here.
8910 * tree-cfgcleanup.c (convert_single_case_switch): New function.
8911 (cleanup_control_expr_graph): Use it.
8912 * tree-switch-conversion.c (try_switch_expansion): Remove
8914 (emit_case_nodes): Use generate_range_test.
8916 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8919 * config/i386/i386.md (*<btsc><mode>_mask): Add
8920 TARGET_USE_BT to insn constraint.
8921 (*btr<mode>_mask): Ditto.
8923 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
8925 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
8926 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
8928 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
8931 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
8932 TARGET_NEON and TARGET_IWMMXT.
8933 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
8934 TARGET_NEON and TARGET_IWMMXT.
8935 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
8937 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8939 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
8940 (ix86_rewrite_tls_address): Ditto.
8941 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
8942 (ix86_rewrite_tls_address_1): Ditto.
8943 (ix86_rewrite_tls_address): Ditto.
8944 * config/i386/predicates.md (tls_address_pattern): New predicate.
8945 * config/i386/i386.md (TLS address splitter): New splitter.
8947 2017-09-04 Richard Biener <rguenther@suse.de>
8949 PR tree-optimization/82084
8950 * fold-const.h (can_native_encode_string_p): Declare.
8951 * fold-const.c (can_native_encode_string_p): Factor out from ...
8952 (native_encode_string): ... here.
8953 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
8954 vectorizing stores from constants we later cannot handle.
8956 2017-09-04 Marek Polacek <polacek@redhat.com>
8959 * doc/invoke.texi: Update -Wtautological-compare documentation.
8961 2017-09-04 Jeff Law <law@redhat.com>
8963 PR tree-optimization/64910
8964 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
8965 swap the first and last operand if the last is a constant.
8967 2017-09-04 Marek Polacek <polacek@redhat.com>
8970 * convert.c (do_narrow): When sanitizing signed integer overflows,
8971 bail out for signed types.
8972 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
8974 2017-09-04 Richard Biener <rguenther@suse.de>
8976 PR tree-optimization/82060
8977 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
8978 Move devirtualization after stmt folding and before EH/AB/noreturn
8979 cleanup to get the stmt refs canonicalized. Use a bool instead
8980 of gimple_modified_p since that doesn't work for NOPs. Schedule
8981 NOPs generated by folding for removal.
8983 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8984 Alan Hayward <alan.hayward@arm.com>
8985 David Sherwood <david.sherwood@arm.com>
8987 * coretypes.h (pad_direction): New enum.
8988 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
8989 (FUNCTION_ARG_PADDING): Likewise.
8990 * target.def (function_arg_padding): New hook.
8991 * targhooks.h (default_function_arg_padding): Declare.
8992 * targhooks.c (default_function_arg_padding): New function.
8993 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
8994 (TARGET_FUNCTION_ARG_PADDING): ...this.
8995 * doc/tm.texi: Regenerate.
8996 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
8997 instead of direction.
8998 (compute_argument_addresses): Likewise.
8999 (load_register_parameters): Likewise.
9000 (emit_library_call_value_1): Likewise.
9001 (store_one_arg): Use targetm.calls.function_arg_padding instead
9002 of FUNCTION_ARG_PADDING.
9003 (must_pass_in_stack_var_size_or_pad): Likewise.
9004 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9005 (emit_group_store): Likewise.
9006 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9007 instead of FUNCTION_ARG_PADDING.
9008 (emit_push_insn): Likewise, and propagate enum change throughout
9010 * function.h (direction): Delete.
9011 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9013 * function.c (assign_parm_find_stack_rtl): Likewise.
9014 (assign_parm_setup_block_p): Likewise.
9015 (assign_parm_setup_block): Likewise.
9016 (gimplify_parameters): Likewise.
9017 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9018 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9020 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9021 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9022 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9023 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9024 (aarch64_function_arg_padding): ...this new function.
9025 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9026 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9027 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9028 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9029 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9030 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9031 (arm_pad_arg_upward): Replace with...
9032 (arm_function_arg_padding): ...this new function.
9033 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9035 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9036 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9037 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9038 (ia64_hpux_function_arg_padding): Replace with...
9039 (ia64_function_arg_padding): ...this new function. Use pad_direction
9040 instead of direction. Check for TARGET_HPUX.
9041 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9042 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9043 (iq2000_function_arg_padding): New function.
9044 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9045 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9046 (mips_function_arg_padding): ...this new function.
9047 (mips_pad_reg_upward): Update accordingly.
9048 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9049 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9050 targetm.calls.function_arg_padding.
9051 (FUNCTION_ARG_PADDING): Delete.
9052 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9053 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9054 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9055 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9056 (nios2_block_reg_padding): Return pad_direction instead of direction.
9057 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9058 instead of direction.
9059 (nios2_function_arg_padding): Likewise. Make static.
9060 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9061 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9062 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9063 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9064 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9065 (pa_function_arg_padding): Make static. Return pad_direction instead
9067 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9068 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9069 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9070 instead of direction. Use targetm.calls.function_arg_padding.
9071 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9072 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9073 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9074 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9075 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9077 (function_arg_padding): Rename to...
9078 (rs6000_function_arg_padding): ...this. Make static. Return
9079 pad_direction instead of direction.
9080 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9081 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9082 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9083 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9084 instead of direction. Use targetm.calls.function_arg_padding.
9085 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9086 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9087 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9088 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9089 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9090 (function_arg_padding): Rename to...
9091 (rs6000_function_arg_padding): ...this. Make static. Return
9092 pad_direction instead of direction.
9093 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9094 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9095 * config/s390/s390.c (s390_function_arg_padding): New function.
9096 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9097 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9098 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
9099 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9100 (function_arg_padding): Rename to...
9101 (sparc_function_arg_padding): ...this. Make static. Return
9102 pad_direction instead of direction.
9103 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
9104 * config/spu/spu.c (spu_function_arg_padding): New function.
9105 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9106 * system.h (FUNCTION_ARG_PADDING): Poison.
9108 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9109 Alan Hayward <alan.hayward@arm.com>
9110 David Sherwood <david.sherwood@arm.com>
9112 * target.def (modes_tieable_p): New hook.
9113 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
9114 (TARGET_MODES_TIEABLE_P): ...this.
9115 * doc/tm.texi.in: Regenerate.
9116 * hooks.h (hook_bool_mode_mode_true): Declare.
9117 * hooks.c (hook_bool_mode_mode_true): New function.
9118 * combine.c (subst): Use targetm.modes_tieable_p instead of
9120 * dse.c (find_shift_sequence): Likewise.
9121 * expmed.c (extract_low_bits): Likewise.
9122 * lower-subreg.c: Include target.h.
9123 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
9125 * rtlanal.c (rtx_cost): Likewise.
9126 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
9127 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
9128 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
9129 (TARGET_MODES_TIEABLE_P): Redefine.
9130 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
9131 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
9132 (TARGET_MODES_TIEABLE_P): Redefine.
9133 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
9134 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
9135 (arc_modes_tieable_p): New function.
9136 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
9137 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
9138 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
9139 (arm_modes_tieable_p): Make static.
9140 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
9141 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
9142 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
9143 (TARGET_MODES_TIEABLE_P): Redefine.
9144 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
9145 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
9146 (TARGET_MODES_TIEABLE_P): Redefine.
9147 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
9148 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
9149 (cr16_modes_tieable_p): New function.
9150 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
9151 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
9152 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
9153 (TRULY_NOOP_TRUNCATION): Update comment.
9154 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
9155 (TRULY_NOOP_TRUNCATION): Update comment.
9156 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
9157 (frv_modes_tieable_p): New function.
9158 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
9159 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
9160 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
9161 (TARGET_MODES_TIEABLE_P): Redefine.
9162 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9163 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9164 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9165 (TARGET_MODES_TIEABLE_P): Redefine.
9166 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9167 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9168 (ia64_modes_tieable_p): New function.
9169 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9170 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9171 (iq2000_modes_tieable_p): New function.
9172 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9173 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9174 (lm32_modes_tieable_p): New function.
9175 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9176 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9177 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9178 (TARGET_MODES_TIEABLE_P): Redefine.
9179 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9180 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9181 (m32r_modes_tieable_p): New function.
9182 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9183 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9184 (m68k_modes_tieable_p): New function.
9185 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9186 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9187 (mcore_modes_tieable_p): New function.
9188 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9189 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9191 (TARGET_MODES_TIEABLE_P): Redefine.
9192 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9193 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9194 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9195 (TARGET_MODES_TIEABLE_P): Redefine.
9196 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9197 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9198 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9199 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9200 (mn10300_modes_tieable_p): ...this and make static.
9201 (TARGET_MODES_TIEABLE_P): Redefine.
9202 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9203 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9204 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9205 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9206 (msp430_modes_tieable_p): Make static.
9207 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9208 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9209 (TARGET_MODES_TIEABLE_P): Redefine.
9210 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9211 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9212 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9213 (TARGET_MODES_TIEABLE_P): Redefine.
9214 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9215 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9216 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9217 (TARGET_MODES_TIEABLE_P): Redefine.
9218 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9219 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9220 (pdp11_modes_tieable_p): New function.
9221 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9222 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9223 (rs6000_modes_tieable_p): New function.
9224 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9225 * config/powerpcspe/powerpcspe.md: Update comment.
9226 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9227 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9228 (TARGET_MODES_TIEABLE_P): Redefine.
9229 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9230 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9231 (rl78_modes_tieable_p): New function.
9232 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9233 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9234 (rs6000_modes_tieable_p): New function.
9235 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9236 * config/rs6000/rs6000.md: Update comment.
9237 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9238 * config/rx/rx.c (rx_modes_tieable_p): New function.
9239 (TARGET_MODES_TIEABLE_P): Redefine.
9240 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9241 * config/s390/s390.c (s390_modes_tieable_p): New function.
9242 (TARGET_MODES_TIEABLE_P): Redefine.
9243 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9244 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9245 (sh_modes_tieable_p): New function.
9246 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9247 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9248 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9249 (sparc_modes_tieable_p): Make static.
9250 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9251 * config/spu/spu.c (spu_modes_tieable_p): New function.
9252 (TARGET_MODES_TIEABLE_P): Redefine.
9253 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9254 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9255 (TARGET_MODES_TIEABLE_P): Redefine.
9256 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9257 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9258 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9259 * config/v850/v850.c (v850_modes_tieable_p): New function.
9260 (TARGET_MODES_TIEABLE_P): Redefine.
9261 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9262 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9263 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9264 (visium_modes_tieable_p): New function.
9265 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9266 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9267 (xtensa_modes_tieable_p): New function.
9268 * system.h (MODES_TIEABLE_P): Poison.
9270 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9271 Alan Hayward <alan.hayward@arm.com>
9272 David Sherwood <david.sherwood@arm.com>
9274 * target.def (hard_regno_mode_ok): New hook.
9275 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9276 (TARGET_HARD_REGNO_MODE_OK): ...this.
9277 * doc/tm.texi.in: Regenerate.
9278 * hooks.h (hook_bool_uint_mode_true): Declare.
9279 * hooks.c (hook_bool_uint_mode_true): New function.
9280 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9282 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9283 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9284 instead of HARD_REGNO_MODE_OK.
9285 * caller-save.c: Include target.h.
9286 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9288 * combine.c (can_combine_p): Likewise.
9289 (combinable_i3pat): Likewise.
9290 (can_change_dest_mode): Likewise.
9291 * expr.c (init_expr_target): Likewise.
9292 (convert_move): Likewise.
9293 (convert_modes): Likewise.
9294 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9295 (setup_prohibited_mode_move_regs): Likewise.
9296 * ira.h (target_ira): Likewise.
9297 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9298 * lra-constraints.c (process_alt_operands): Likewise.
9299 (split_reg): Likewise.
9300 * recog.c (peep2_find_free_register): Likewise.
9301 * ree.c (combine_reaching_defs): Likewise.
9302 * regcprop.c (maybe_mode_change): Likewise.
9303 * reginfo.c (init_reg_sets_1): Likewise.
9304 (choose_hard_reg_mode): Likewise.
9305 (simplifiable_subregs): Likewise.
9306 * regrename.c (check_new_reg_p): Likewise.
9307 * reload.c (find_valid_class): Likewise.
9308 (find_valid_class_1): Likewise.
9309 (reload_inner_reg_of_subreg): Likewise.
9310 (push_reload): Likewise.
9311 (combine_reloads): Likewise.
9312 (find_dummy_reload): Likewise.
9313 (find_reloads): Likewise.
9314 * reload1.c (find_reg): Likewise.
9315 (set_reload_reg): Likewise.
9316 (allocate_reload_reg): Likewise.
9317 (choose_reload_regs): Likewise.
9318 (reload_adjust_reg_for_temp): Likewise.
9319 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9320 (simplify_subreg_regno): Likewise.
9321 * sel-sched.c (init_regs_for_mode): Likewise.
9322 * varasm.c (make_decl_rtl): Likewise.
9323 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9324 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9326 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9327 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9328 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9329 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9330 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9331 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9332 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9333 (arc_mode_class): Delete.
9334 (HARD_REGNO_MODE_OK): Delete.
9335 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9336 (arc_hard_regno_mode_ok): Rename old array to...
9337 (arc_hard_regno_mode_ok_modes): ...this.
9338 (arc_conditional_register_usage): Update accordingly.
9339 (arc_mode_class): Make static.
9340 (arc_hard_regno_mode_ok): New function.
9341 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9342 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9343 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9344 (arm_hard_regno_mode_ok): Make static.
9345 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9347 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9348 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9349 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9351 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9352 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9353 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9354 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9355 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9356 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9357 * config/bfin/predicates.md (valid_reg_operand): Use
9358 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9359 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9360 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9361 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9362 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9363 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9364 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9365 (cr16_hard_regno_mode_ok): Make static and return a bool.
9366 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9367 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9368 (cris_hard_regno_mode_ok): New function.
9369 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9370 (epiphany_mode_class): Delete.
9371 (HARD_REGNO_MODE_OK): Delete.
9372 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9373 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9374 (hard_regno_mode_ok): Rename to...
9375 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9376 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9377 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9379 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9380 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9381 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9382 (frv_hard_regno_mode_ok): Make static and return a bool.
9383 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9385 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9386 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9387 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9388 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9390 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9391 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9392 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9393 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9395 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9396 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9397 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9398 (ia64_hard_regno_mode_ok): New function.
9399 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9400 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9401 (iq2000_hard_regno_mode_ok): New function.
9402 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9403 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9404 (lm32_hard_regno_mode_ok): New function.
9405 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9406 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9407 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9408 instead of HARD_REGNO_MODE_OK.
9409 (m32c_hard_regno_ok): Rename to...
9410 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9411 (m32c_cannot_change_mode_class): Update accordingly.
9412 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9413 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9414 (m32r_mode_class): Delete.
9415 (HARD_REGNO_MODE_OK): Delete.
9416 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9417 (m32r_hard_regno_mode_ok): Rename to...
9418 (m32r_hard_regno_modes): ...this.
9419 (m32r_mode_class): Make static.
9420 (m32r_hard_regno_mode_ok): New function.
9421 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9422 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9423 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9424 (m68k_hard_regno_mode_ok): Make static.
9425 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9426 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9427 (mcore_hard_regno_mode_ok): New function.
9428 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9429 (HARD_REGNO_MODE_OK): Delete.
9430 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9432 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9433 (microblaze_hard_regno_mode_ok): New function.
9434 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9435 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9436 (mips_hard_regno_mode_ok): Delete.
9437 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9438 (mips_hard_regno_mode_ok_p): ...this and make static.
9439 (mips_hard_regno_mode_ok_p): Rename to...
9440 (mips_hard_regno_mode_ok_uncached): ...this.
9441 (mips_hard_regno_mode_ok): New function.
9442 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9443 of HARD_REGNO_MODE_OK.
9444 (mips_option_override): Update after above name changes.
9445 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9446 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9447 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9448 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9449 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9450 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9451 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9452 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9453 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9454 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9455 (msp430_hard_regno_mode_ok): Make static and return a bool.
9456 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9457 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9458 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9460 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9461 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9462 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9463 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9464 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9465 (PA_HARD_REGNO_MODE_OK): ...this
9466 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9467 (PA_HARD_REGNO_MODE_OK): ...this.
9468 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9469 (pa_hard_regno_mode_ok): New function.
9470 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9471 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9472 (pdp11_hard_regno_mode_ok): New function.
9473 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9474 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9476 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9478 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9479 (rs6000_hard_regno_mode_ok): Rename to...
9480 (rs6000_hard_regno_mode_ok_uncached): ...this.
9481 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9482 (rs6000_hard_regno_mode_ok): New function.
9483 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9484 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9485 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9486 (riscv_hard_regno_mode_ok): ...this and make static.
9487 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9488 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9489 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9490 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9491 (rl78_hard_regno_mode_ok): Make static and return bool.
9492 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9493 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9495 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9496 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9497 (rs6000_hard_regno_mode_ok): Rename to...
9498 (rs6000_hard_regno_mode_ok_uncached): ...this.
9499 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9500 (rs6000_hard_regno_mode_ok): New function.
9501 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9502 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9503 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9504 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9505 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9506 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9507 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9508 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9509 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9510 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9511 (sh_hard_regno_mode_ok): Make static.
9512 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9513 instead of HARD_REGNO_MODE_OK.
9514 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9515 (sparc_mode_class): Delete.
9516 (HARD_REGNO_MODE_OK): Delete.
9517 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9518 (hard_regno_mode_classes): Make static.
9519 (sparc_mode_class): Likewise.
9520 (sparc_hard_regno_mode_ok): New function.
9521 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9522 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9523 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9525 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9526 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9527 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9528 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9529 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9530 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9531 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9532 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9533 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9534 (visium_hard_regno_mode_ok): New function.
9535 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9536 instead of HARD_REGNO_MODE_OK.
9537 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9538 (HARD_REGNO_MODE_OK): Delete.
9539 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9540 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9541 (xtensa_option_override): Update accordingly.
9542 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9543 (xtensa_hard_regno_mode_ok): New function.
9544 * system.h (HARD_REGNO_MODE_OK): Poison.
9546 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9547 Alan Hayward <alan.hayward@arm.com>
9548 David Sherwood <david.sherwood@arm.com>
9550 * target.def (hard_regno_call_part_clobbered): New hook.
9551 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9552 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9553 * doc/tm.texi: Regenerate.
9554 * hooks.h (hook_bool_uint_mode_false): Declare.
9555 * hooks.c (hook_bool_uint_mode_false): New function.
9556 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9557 * cselib.c (cselib_process_insn): Use
9558 targetm.hard_regno_call_part_clobbered instead of
9559 HARD_REGNO_CALL_PART_CLOBBERED.
9560 * ira-conflicts.c (ira_build_conflicts): Likewise.
9561 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9562 * lra-constraints.c (need_for_call_save_p): Likewise.
9563 * lra-lives.c: Include target.h.
9564 (check_pseudos_live_through_calls): Use
9565 targetm.hard_regno_call_part_clobbered instead of
9566 HARD_REGNO_CALL_PART_CLOBBERED.
9567 * regcprop.c: Include target.h.
9568 (copyprop_hardreg_forward_1): Use
9569 targetm.hard_regno_call_part_clobbered instead of
9570 HARD_REGNO_CALL_PART_CLOBBERED.
9571 * reginfo.c (choose_hard_reg_mode): Likewise.
9572 * regrename.c (check_new_reg_p): Likewise.
9573 * reload.c (find_equiv_reg): Likewise.
9574 * reload1.c (emit_reload_insns): Likewise.
9575 * sched-deps.c (deps_analyze_insn): Likewise.
9576 * sel-sched.c (init_regs_for_mode): Likewise.
9577 (mark_unavailable_hard_regs): Likewise.
9578 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9579 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9580 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9582 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9583 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9584 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9586 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9588 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9589 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9590 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9592 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9593 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9594 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9596 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9597 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
9599 * config/powerpcspe/powerpcspe.c
9600 (rs6000_hard_regno_call_part_clobbered): New function.
9601 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9602 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9603 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
9605 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9606 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9607 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
9609 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9610 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9611 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
9613 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9614 Alan Hayward <alan.hayward@arm.com>
9615 David Sherwood <david.sherwood@arm.com>
9617 * rtl.h (subreg_memory_offset): Declare.
9618 * emit-rtl.c (subreg_memory_offset): New function.
9619 * expmed.c (store_bit_field_1): Use it.
9620 * expr.c (undefined_operand_subword_p): Likewise.
9621 * simplify-rtx.c (simplify_subreg): Likewise.
9623 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9625 PR rtl-optimization/57448
9628 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9629 using atomic_load pattern.
9630 (expand_atomic_store): Likewise.
9632 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9635 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9636 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9637 handling. Use replace_call_with_value with NULL instead of
9638 gsi_replace, unlink_stmt_vdef and release_defs.
9640 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9643 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9645 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9648 * rtl.h (emit_library_call_value_1): Declare.
9649 (emit_library_call): Replace declaration with a series of overloads.
9650 Remove the parameter count argument.
9651 (emit_library_call_value): Likewise.
9652 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9653 with an "rtx_mode_t *".
9654 (emit_library_call_value): Delete.
9655 (emit_library_call): Likewise.
9656 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9657 (asan_emit_allocas_unpoison): Likewise.
9658 * builtins.c (expand_builtin_powi): Likewise.
9659 (expand_asan_emit_allocas_unpoison): Likewise.
9660 * cfgexpand.c (expand_main_function): Likewise.
9661 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9662 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9663 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9664 * config/arm/arm.c (arm_trampoline_init): Likewise.
9665 (arm_call_tls_get_addr): Likewise.
9666 (arm_expand_divmod_libfunc): Likewise.
9667 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9668 (smulsi3_highpart): Likewise.
9669 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9670 (c6x_expand_compare): Likewise.
9671 (c6x_expand_movmem): Likewise.
9672 * config/frv/frv.c (frv_trampoline_init): Likewise.
9673 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9674 (ix86_expand_divmod_libfunc): Likewise.
9675 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9676 (ia64_expand_compare): Likewise.
9677 (ia64_profile_hook): Likewise.
9678 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9679 (nonlocal_goto): Likewise.
9680 (restore_stack_nonlocal): Likewise.
9681 * config/m32r/m32r.c (block_move_call): Likewise.
9682 (m32r_trampoline_init): Likewise.
9683 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9684 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9685 (m68k_call_m68k_read_tp): Likewise.
9686 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9687 (microblaze_expand_divide): Likewise.
9688 * config/mips/mips.h (mips_args): Likewise.
9689 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9690 (MIPS_ICACHE_SYNC): Likewise.
9691 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9692 (nios2_trampoline_init): Likewise.
9693 * config/pa/pa.c (hppa_tls_call): Likewise.
9694 (pa_trampoline_init): Likewise.
9695 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9696 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9697 (expand_strn_compare): Likewise.
9698 (rs6000_generate_compare): Likewise.
9699 (rs6000_expand_float128_convert): Likewise.
9700 (output_profile_hook): Likewise.
9701 (rs6000_trampoline_init): Likewise.
9702 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9703 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9704 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9705 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9706 (rs6000_generate_compare): Likewise.
9707 (rs6000_expand_float128_convert): Likewise.
9708 (output_profile_hook): Likewise.
9709 (rs6000_trampoline_init): Likewise.
9710 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9711 * config/sh/sh.c (sh_trampoline_init): Likewise.
9712 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9713 (sparc_emit_float_lib_cmp): Likewise.
9714 (sparc32_initialize_trampoline): Likewise.
9715 (sparc64_initialize_trampoline): Likewise.
9716 (sparc_profile_hook): Likewise.
9717 * config/spu/spu.c (ea_load_store): Likewise.
9718 * config/spu/spu.md (floatunssidf2): Likewise.
9719 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9720 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9721 * config/visium/visium.c (expand_block_move_4): Likewise.
9722 (expand_block_move_2): Likewise.
9723 (expand_block_move_1): Likewise.
9724 (expand_block_set_4): Likewise.
9725 (expand_block_set_2): Likewise.
9726 (expand_block_set_1): Likewise.
9727 (visium_trampoline_init): Likewise.
9728 (visium_profile_hook): Likewise.
9729 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9730 (xtensa_setup_frame_addresses): Likewise.
9731 (xtensa_trampoline_init): Likewise.
9732 * except.c (sjlj_emit_function_enter): Likewise.
9733 (sjlj_emit_function_exit): Likewise.
9734 * explow.c (allocate_dynamic_stack_space): Likewise.
9735 (probe_stack_range): Likewise.
9736 * expr.c (convert_mode_scalar): Likewise.
9737 * optabs.c (expand_binop): Likewise.
9738 (expand_twoval_binop_libfunc): Likewise.
9739 (expand_unop): Likewise.
9740 (prepare_cmp_insn): Likewise.
9741 (prepare_float_lib_cmp): Likewise.
9742 (expand_float): Likewise.
9743 (expand_fix): Likewise.
9744 (expand_fixed_convert): Likewise.
9745 (maybe_emit_sync_lock_test_and_set): Likewise.
9746 (expand_atomic_compare_and_swap): Likewise.
9747 (expand_mem_thread_fence): Likewise.
9748 (expand_atomic_fetch_op): Likewise.
9750 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9752 * doc/generic.texi (OpenACC): Adjust URL.
9753 * doc/invoke.texi (C Dialect Options): Ditto.
9755 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9757 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9758 predicate for operand 1. Add (m,<S>) constraint.
9759 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9760 Prevent memory operand 1 with register operand 2.
9762 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9764 PR rtl-optimization/82024
9765 * combine.c (try_combine): If the combination result is a PARALLEL,
9766 and we only need to retain the SET in there that would be placed
9767 at I2, check that we can place that at I3 instead, before doing so.
9769 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9772 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9774 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9775 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9778 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9779 Jeff Law <law@redhat.com>
9781 * varasm.c (bss_initializer_p): Do not put constants into .bss
9782 (categorize_decl_for_section): Handle bss_initializer_p returning
9783 false when DECL_INITIAL is NULL.
9785 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9788 * config/s390/s390.c (s390_can_inline_p): New function.
9790 2017-09-01 Jeff Law <law@redhat.com>
9792 PR tree-optimization/82052
9793 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9794 Always initialize the returned slot after a hash table miss
9795 when INSERT is true.
9797 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9799 * config/s390/s390.md (mem_signal_fence): Remove.
9800 * doc/md.texi (mem_signal_fence): Remove.
9801 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9803 * target-insns.def (mem_signal_fence): Remove.
9805 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9808 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9811 * asan.c (create_odr_indicator): Strip name encoding from assembler
9812 name before appending it after __odr_asan_.
9814 2017-09-01 Martin Liska <mliska@suse.cz>
9816 PR tree-optimization/82059
9817 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9818 frequency only when an edge is redirected.
9820 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9822 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9823 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9824 (arc_conditional_register_usage): Remove ARC600 lp_count
9826 (arc_file_start): Emit Tag_ARC_CPU_variation.
9827 (arc_can_use_doloop_p): New conditions to use ZOLs.
9828 (hwloop_fail): New function.
9829 (hwloop_optimize): Likewise.
9830 (hwloop_pattern_reg): Likewise.
9831 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9832 (arc_reorg_loops): New function, calls reorg_loops.
9833 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9834 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9836 (arc_loop_hazard): Remove function, functionality moved into
9838 (arc_hazard): Remove arc_loop_hazard call.
9839 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9840 into hwloop_optimize.
9841 (arc_label_align): Remove ZOL handling.
9842 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9843 * config/arc/arc.md (doloop_begin): Remove pattern.
9844 (doloop_begin_i): Likewise.
9845 (doloop_end_i): Likewise.
9846 (doloop_fallback): Likewise.
9847 (doloop_fallback_m): Likewise.
9848 (doloop_end): Reimplement expand.
9849 (arc_lp): New pattern for LP instruction.
9850 (loop_end): New pattern.
9851 (loop_fail): Likewise.
9852 (decrement_and_branch_until_zero): Likewise.
9853 * config/arc/arc.opt (mlpc-width): New option.
9854 * doc/invoke.texi (mlpc-width): Document option.
9856 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9858 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9859 (arc_ccfsm_advance): Fix checking for delay slots.
9860 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9862 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9864 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9866 (movhi_insn): Update store instruction constraint which are saving
9867 6-bit short immediates.
9868 (movsi_insn): Consider also short scaled load operations.
9869 (zero_extendhisi2_i): Use Usd constraint instead of T.
9870 (extendhisi2_i): Add q constraint.
9871 (arc_clzsi2): Add type and length attributes.
9872 (arc_ctzsi2): Likewise.
9873 * config/arc/constraints.md (Usc): Update constraint, the
9874 assembler can parse two relocations for a single instruction.
9876 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9878 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9879 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9881 2017-08-31 Olivier Hainque <hainque@adacore.com>
9883 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9884 match as powerpc-wrs-vxworks*.
9886 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9888 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9889 register constraint for by-element operand.
9890 (aarch64_mls_elt_merge<mode>): Likewise.
9892 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9894 * config/arc/arc.c (arc_can_follow_jump): Check for short
9897 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9899 * config.gcc: Use g.opt for arc.
9900 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9901 functionality moved to ...
9902 (legitimate_scaled_address_p): New function, ...here.
9903 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9904 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9905 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9907 (arc_override_options): Handle G option.
9908 (arc_output_pic_addr_const): Correct function definition.
9909 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9910 (arc_decl_anon_ns_mem_p): Delete.
9911 (arc_in_small_data_p): Overhaul this function to take into
9912 consideration the value given via G option.
9913 (arc_rewrite_small_data_1): Renamed and corrected old
9914 arc_rewrite_small_data function.
9915 (arc_rewrite_small_data): New function.
9916 (small_data_pattern): Don't use pic_offset_table_rtx.
9917 * config/arc/arc.h (CC1_SPEC): Recognize G option.
9918 * config/arc/simdext.md (movmisalignv2hi): Use
9919 prepare_move_operands function.
9921 (movmisalign*): Likewise.
9922 * doc/invoke.texi (ARC options): Document -G option.
9924 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9926 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
9928 * config/arc/arc.c (arc_print_operand): Output scalled address for
9929 sdata whenever is possible.
9930 (arc_in_small_data_p): Allow sdata for 64bit datum when double
9931 load/stores are available.
9932 (compact_sda_memory_operand): Check for the alignment required by
9933 code density instructions.
9934 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
9936 * config/arc/constraints.md (Usd): Update constraint.
9937 (Us0): New constraint.
9938 (Usc): Update constraint.
9940 2017-08-31 Richard Biener <rguenther@suse.de>
9943 * dwarf2out.c (dwarf2out_early_global_decl): Process each
9946 2017-08-31 Tamar Christina <tamar.christina@arm.com>
9948 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
9949 Resize type_signature.
9951 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9952 Alan Hayward <alan.hayward@arm.com>
9953 David Sherwood <david.sherwood@arm.com>
9955 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
9956 subregs whose inner modes can be stored in GPRs.
9957 (aarch64_classify_index): Likewise.
9959 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9960 Alan Hayward <alan.hayward@arm.com>
9961 David Sherwood <david.sherwood@arm.com>
9963 * config/aarch64/iterators.md (V_cmp_result): Rename to...
9964 (V_INT_EQUIV): ...this.
9965 (v_cmp_result): Rename to...
9966 (v_int_equiv): ...this.
9967 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
9968 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
9969 (copysign<mode>3): Likewise.
9970 (aarch64_simd_bsl<mode>_internal): Likewise.
9971 (aarch64_simd_bsl<mode>): Likewise.
9972 (vec_cmp<mode><mode>): Likewise.
9973 (vcond<mode><mode>): Likewise.
9974 (vcond<v_cmp_mixed><mode>): Likewise.
9975 (vcondu<mode><v_cmp_mixed>): Likewise.
9976 (aarch64_cm<optab><mode>): Likewise.
9977 (aarch64_cmtst<mode>): Likewise.
9978 (aarch64_fac<optab><mode>): Likewise.
9979 (vec_perm_const<mode>): Likewise.
9980 (vcond_mask_<mode><v_cmp_result>): Rename to...
9981 (vcond_mask_<mode><v_int_equiv>): ...this.
9982 (vec_cmp<mode><v_cmp_result>): Rename to...
9983 (vec_cmp<mode><v_int_equiv>): ...this.
9985 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
9986 Alan Hayward <alan.hayward@arm.com>
9987 David Sherwood <david.sherwood@arm.com>
9989 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
9991 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
9992 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
9993 (UNSPEC_LD4_DREG): New unspecs.
9994 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
9995 (aarch64_ld2<mode>_dreg_be): Replace with...
9996 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
9998 (aarch64_ld3<mode>_dreg_le)
9999 (aarch64_ld3<mode>_dreg_be): Replace with...
10000 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10002 (aarch64_ld4<mode>_dreg_le)
10003 (aarch64_ld4<mode>_dreg_be): Replace with...
10004 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10007 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10009 PR tree-optimization/81987
10010 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10011 insert an initializer in a location not dominated by the stride
10014 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10016 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10017 on the entire header of the finally block in the fallthru case.
10019 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10021 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10023 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10025 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10026 rs6000_emit_move_from_cr and call renamed function.
10027 (rs6000_emit_prologue): Call renamed functions.
10028 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10029 movesi_from_cr, remove volatile CRs.
10031 2017-08-30 Jon Beniston <jon@beniston.com>
10032 Richard Biener <rguenther@suse.de>
10034 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10035 of VECTOR_MODE_P check.
10036 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10037 element vector types.
10039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10041 * df.h (df_read_modify_subreg_p): Remove in favor of...
10042 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10043 const_rtx instead of an rtx.
10044 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10045 * df-problems.c (df_word_lr_mark_ref): Likewise.
10046 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10047 (mark_pseudo_reg_dead): Likewise.
10048 (mark_ref_dead): Likewise.
10049 * reginfo.c (init_subregs_of_mode): Likewise.
10050 * sched-deps.c (sched_analyze_1): Likewise.
10051 * df-scan.c (df_def_record_1): Likewise.
10052 (df_uses_record): Likewise.
10053 (df_read_modify_subreg_p): Remove in favor of...
10054 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10055 const_rtx instead of an rtx.
10057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10058 Alan Hayward <alan.hayward@arm.com>
10059 David Sherwood <david.sherwood@arm.com>
10061 * rtl.h (partial_subreg_p): New function.
10062 * caller-save.c (save_call_clobbered_regs): Use it.
10063 * calls.c (expand_call): Likewise.
10064 * combine.c (combinable_i3pat): Likewise.
10065 (simplify_set): Likewise.
10066 (make_extraction): Likewise.
10067 (make_compound_operation_int): Likewise.
10068 (gen_lowpart_or_truncate): Likewise.
10069 (force_to_mode): Likewise.
10070 (make_field_assignment): Likewise.
10071 (reg_truncated_to_mode): Likewise.
10072 (record_truncated_value): Likewise.
10073 (move_deaths): Likewise.
10074 * cse.c (record_jump_cond): Likewise.
10075 (cse_insn): Likewise.
10076 * cselib.c (cselib_lookup_1): Likewise.
10077 * expmed.c (extract_bit_field_using_extv): Likewise.
10078 * function.c (assign_parm_setup_reg): Likewise.
10079 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10080 * ira-build.c (create_insn_allocnos): Likewise.
10081 * lra-coalesce.c (merge_pseudos): Likewise.
10082 * lra-constraints.c (match_reload): Likewise.
10083 (simplify_operand_subreg): Likewise.
10084 (curr_insn_transform): Likewise.
10085 * lra-lives.c (process_bb_lives): Likewise.
10086 * lra.c (new_insn_reg): Likewise.
10087 (lra_substitute_pseudo): Likewise.
10088 * regcprop.c (mode_change_ok): Likewise.
10089 (maybe_mode_change): Likewise.
10090 (copyprop_hardreg_forward_1): Likewise.
10091 * reload.c (push_reload): Likewise.
10092 (find_reloads): Likewise.
10093 (find_reloads_subreg_address): Likewise.
10094 * reload1.c (alter_reg): Likewise.
10095 (eliminate_regs_1): Likewise.
10096 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10098 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
10100 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
10101 back to if statements, including unpack.
10103 2017-08-30 Martin Liska <mliska@suse.cz>
10105 PR inline-asm/82001
10106 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
10108 (func_checker::compare_asm_inputs_outputs): ... this function.
10109 (func_checker::compare_gimple_asm): Use the function to compare
10110 also ASM constrains.
10111 * ipa-icf-gimple.h: Rename the function.
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 * coretypes.h (complex_mode): New type.
10118 * gdbhooks.py (build_pretty_printer): Handle it.
10119 * machmode.h (complex_mode): New class.
10120 (complex_mode::includes_p): New function.
10121 (is_complex_int_mode): Likewise.
10122 (is_complex_float_mode): Likewise.
10123 * genmodes.c (get_mode_class): Handle complex mode classes.
10124 * function.c (expand_function_end): Use is_complex_int_mode.
10126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10127 Alan Hayward <alan.hayward@arm.com>
10128 David Sherwood <david.sherwood@arm.com>
10130 * coretypes.h (scalar_mode_pod): New typedef.
10131 * gdbhooks.py (build_pretty_printer): Handle it.
10132 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
10133 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
10134 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
10135 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
10136 as_a <scalar_mode>.
10138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10139 Alan Hayward <alan.hayward@arm.com>
10140 David Sherwood <david.sherwood@arm.com>
10142 * machmode.h (mode_for_vector): Take a scalar_mode instead
10144 * stor-layout.c (mode_for_vector): Likewise.
10145 * explow.c (promote_mode): Use as_a <scalar_mode>.
10146 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
10148 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10149 Alan Hayward <alan.hayward@arm.com>
10150 David Sherwood <david.sherwood@arm.com>
10152 * target.def (preferred_simd_mode): Take a scalar_mode
10153 instead of a machine_mode.
10154 * targhooks.h (default_preferred_simd_mode): Likewise.
10155 * targhooks.c (default_preferred_simd_mode): Likewise.
10156 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
10157 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
10158 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
10159 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
10160 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
10161 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10162 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10163 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10165 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10166 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10167 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10168 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10169 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10170 * doc/tm.texi: Regenerate.
10171 * optabs-query.c (can_vec_mask_load_store_p): Return false for
10174 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10175 Alan Hayward <alan.hayward@arm.com>
10176 David Sherwood <david.sherwood@arm.com>
10178 * target.def (scalar_mode_supported_p): Take a scalar_mode
10179 instead of a machine_mode.
10180 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10181 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10182 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10183 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10184 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10185 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10186 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10187 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10188 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10189 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10190 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10191 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10192 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10193 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10195 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10196 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10197 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10198 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10199 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10201 * doc/tm.texi: Regenerate.
10203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10204 Alan Hayward <alan.hayward@arm.com>
10205 David Sherwood <david.sherwood@arm.com>
10207 * coretypes.h (opt_scalar_mode): New typedef.
10208 * gdbhooks.py (build_pretty_printers): Handle it.
10209 * machmode.h (mode_iterator::get_2xwider): Add overload for
10211 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10213 * expr.c (convert_mode_scalar): Likewise.
10214 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10215 * optabs.c (expand_float): Likewise.
10216 (expand_fix): Likewise.
10217 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10220 Alan Hayward <alan.hayward@arm.com>
10221 David Sherwood <david.sherwood@arm.com>
10223 * optabs.c (expand_float): Explicitly check for scalars before
10224 using a branching expansion.
10225 (expand_fix): Likewise.
10227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10228 Alan Hayward <alan.hayward@arm.com>
10229 David Sherwood <david.sherwood@arm.com>
10231 * expr.c (convert_mode): Split scalar handling out into...
10232 (convert_mode_scalar): ...this new function. Treat the modes
10235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10236 Alan Hayward <alan.hayward@arm.com>
10237 David Sherwood <david.sherwood@arm.com>
10239 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10241 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10243 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10244 Alan Hayward <alan.hayward@arm.com>
10245 David Sherwood <david.sherwood@arm.com>
10247 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10248 rather than a machine_mode.
10249 (fixed_from_string): Likewise.
10250 (fixed_convert): Likewise.
10251 (fixed_convert_from_int): Likewise.
10252 (fixed_convert_from_real): Likewise.
10253 (real_convert_from_fixed): Likewise.
10254 * fixed-value.c (fixed_from_double_int): Likewise.
10255 (fixed_from_string): Likewise.
10256 (fixed_convert): Likewise.
10257 (fixed_convert_from_int): Likewise.
10258 (fixed_convert_from_real): Likewise.
10259 (real_convert_from_fixed): Likewise.
10260 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10263 Alan Hayward <alan.hayward@arm.com>
10264 David Sherwood <david.sherwood@arm.com>
10266 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10267 of separate mode class checks. Do not allow vector modes here.
10268 (immed_wide_int_const): Use as_a <scalar_mode>.
10269 * explow.c (trunc_int_for_mode): Likewise.
10270 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10271 (wi::shwi): Likewise.
10272 (wi::min_value): Likewise.
10273 (wi::max_value): Likewise.
10274 * dwarf2out.c (loc_descriptor): Likewise.
10275 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10276 for CONST_WIDE_INT.
10278 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10279 Alan Hayward <alan.hayward@arm.com>
10280 David Sherwood <david.sherwood@arm.com>
10282 * tree.h (SCALAR_TYPE_MODE): New macro.
10283 * expr.c (expand_expr_addr_expr_1): Use it.
10284 (expand_expr_real_2): Likewise.
10285 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10286 (fold_convert_const_fixed_from_int): Likewise.
10287 (fold_convert_const_fixed_from_real): Likewise.
10288 (native_encode_fixed): Likewise
10289 (native_encode_complex): Likewise
10290 (native_encode_vector): Likewise.
10291 (native_interpret_fixed): Likewise.
10292 (native_interpret_real): Likewise.
10293 (native_interpret_complex): Likewise.
10294 (native_interpret_vector): Likewise.
10295 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10296 (simd_clone_adjust_argument_types): Likewise.
10297 (simd_clone_init_simd_arrays): Likewise.
10298 (simd_clone_adjust): Likewise.
10299 * stor-layout.c (layout_type): Likewise.
10300 * tree.c (build_minus_one_cst): Likewise.
10301 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10302 * tree-inline.c (estimate_move_cost): Likewise.
10303 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10304 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10305 (vectorizable_reduction): Likewise.
10306 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10307 (vect_recog_mixed_size_cond_pattern): Likewise.
10308 (check_bool_pattern): Likewise.
10309 (adjust_bool_pattern): Likewise.
10310 (search_type_for_mask_1): Likewise.
10311 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10312 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10313 (vectorizable_load): Likewise.
10314 (vectorizable_store): Likewise.
10315 * ubsan.c (ubsan_encode_value): Likewise.
10316 * varasm.c (output_constant): Likewise.
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 * coretypes.h (scalar_mode): New class.
10323 * machmode.h (scalar_mode): Likewise.
10324 (scalar_mode::includes_p): New function.
10325 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10326 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10327 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10328 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10329 * expmed.c (store_bit_field_1): Likewise.
10330 (extract_bit_field_1): Likewise.
10331 * expr.c (write_complex_part): Likewise.
10332 (read_complex_part): Likewise.
10333 (emit_move_complex_push): Likewise.
10334 (expand_expr_real_2): Likewise.
10335 * function.c (assign_parm_setup_reg): Likewise.
10336 (assign_parms_unsplit_complex): Likewise.
10337 * optabs.c (expand_binop): Likewise.
10338 * rtlanal.c (subreg_get_info): Likewise.
10339 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10340 * varasm.c (output_constant_pool_2): Likewise.
10342 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10343 Alan Hayward <alan.hayward@arm.com>
10344 David Sherwood <david.sherwood@arm.com>
10346 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10348 (expmed_mult_highpart_optab): Likewise.
10349 (expmed_mult_highpart): Likewise.
10351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10352 Alan Hayward <alan.hayward@arm.com>
10353 David Sherwood <david.sherwood@arm.com>
10355 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10356 instead of a machine_mode.
10357 (builtin_memset_read_str): Likewise.
10358 * builtins.c (c_readstr): Likewise.
10359 (builtin_memcpy_read_str): Likewise.
10360 (builtin_strncpy_read_str): Likewise.
10361 (builtin_memset_read_str): Likewise.
10362 (builtin_memset_gen_str): Likewise.
10363 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10364 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10365 instead of a machine_mode.
10366 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10368 (make_extraction): Likewise.
10369 (try_widen_shift_mode): Take and return scalar_int_modes instead
10371 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10372 a scalar_int_mode instead of a machine_mode.
10373 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10374 (avr_addr_space_pointer_mode): Likewise.
10375 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10376 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10377 (msp430_unwind_word_mode): Likewise.
10378 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10379 (spu_addr_space_pointer_mode): Likewise.
10380 (spu_addr_space_address_mode): Likewise.
10381 (spu_libgcc_cmp_return_mode): Likewise.
10382 (spu_libgcc_shift_count_mode): Likewise.
10383 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10384 (rl78_addr_space_pointer_mode): Likewise.
10385 (fl78_unwind_word_mode): Likewise.
10386 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10388 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10389 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10390 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10391 (mips_valid_pointer_mode): Likewise.
10392 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10393 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10394 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10396 (ft32_addr_space_address_mode): Likewise.
10397 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10398 scalar_int_mode instead of a machine_mode.
10399 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10401 (m32c_addr_space_address_mode): Likewise.
10402 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10403 (rs6000_eh_return_filter_mode): Likewise.
10404 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10405 (rs6000_eh_return_filter_mode): Likewise.
10406 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10407 (s390_libgcc_shift_count_mode): Likewise.
10408 (s390_unwind_word_mode): Likewise.
10409 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10411 * target.def (mode_rep_extended): Likewise.
10412 (valid_pointer_mode): Likewise.
10413 (addr_space.valid_pointer_mode): Likewise.
10414 (eh_return_filter_mode): Return a scalar_int_mode rather than
10416 (libgcc_cmp_return_mode): Likewise.
10417 (libgcc_shift_count_mode): Likewise.
10418 (unwind_word_mode): Likewise.
10419 (addr_space.pointer_mode): Likewise.
10420 (addr_space.address_mode): Likewise.
10421 * doc/tm.texi: Regenerate.
10422 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10424 (do_jump): Use scalar_int_mode for local variables.
10425 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10426 rather than a machine_mode.
10427 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10428 (scompare_loc_descriptor_wide): Likewise.
10429 (scompare_loc_descriptor_narrow): Likewise.
10430 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10432 * except.c (sjlj_emit_dispatch_table): Likewise.
10433 (expand_builtin_eh_copy_values): Likewise.
10434 * explow.c (convert_memory_address_addr_space_1): Likewise.
10435 Take a scalar_int_mode rather than a machine_mode.
10436 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10437 than a machine_mode.
10438 (memory_address_addr_space): Use scalar_int_mode for local variables.
10439 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10440 rather than a machine_mode.
10441 * expmed.c (mask_rtx): Likewise.
10442 (init_expmed_one_conv): Likewise.
10443 (expand_mult_highpart_adjust): Likewise.
10444 (extract_high_half): Likewise.
10445 (expmed_mult_highpart_optab): Likewise.
10446 (expmed_mult_highpart): Likewise.
10447 (expand_smod_pow2): Likewise.
10448 (expand_sdiv_pow2): Likewise.
10449 (emit_store_flag_int): Likewise.
10450 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10452 (extract_low_bits): Likewise.
10453 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10455 * expr.c (pieces_addr::adjust): Likewise.
10456 (can_store_by_pieces): Likewise.
10457 (store_by_pieces): Likewise.
10458 (clear_by_pieces_1): Likewise.
10459 (expand_expr_addr_expr_1): Likewise.
10460 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10461 (expand_expr_real_1): Likewise.
10462 (try_casesi): Likewise.
10463 * final.c (shorten_branches): Likewise.
10464 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10465 type of "mode" to machine_mode.
10466 * internal-fn.c (expand_arith_overflow_result_store): Take a
10467 scalar_int_mode rather than a machine_mode.
10468 (expand_mul_overflow): Use scalar_int_mode for local variables.
10469 * loop-doloop.c (doloop_modify): Likewise.
10470 (doloop_optimize): Likewise.
10471 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10472 than a machine_mode.
10473 (expand_doubleword_shift_condmove): Likewise.
10474 (expand_doubleword_shift): Likewise.
10475 (expand_doubleword_clz): Likewise.
10476 (expand_doubleword_popcount): Likewise.
10477 (expand_doubleword_parity): Likewise.
10478 (expand_absneg_bit): Use scalar_int_mode for local variables.
10479 (prepare_float_lib_cmp): Likewise.
10480 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10481 rather than a machine_mode.
10482 (convert_memory_address_addr_space): Likewise.
10483 (get_mode_bounds): Likewise.
10484 (get_address_mode): Return a scalar_int_mode rather than a
10486 * rtlanal.c (get_address_mode): Likewise.
10487 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10488 than a machine_mode.
10489 * targhooks.c (default_mode_rep_extended): Likewise.
10490 (default_valid_pointer_mode): Likewise.
10491 (default_addr_space_valid_pointer_mode): Likewise.
10492 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10493 than a machine_mode.
10494 (default_libgcc_cmp_return_mode): Likewise.
10495 (default_libgcc_shift_count_mode): Likewise.
10496 (default_unwind_word_mode): Likewise.
10497 (default_addr_space_pointer_mode): Likewise.
10498 (default_addr_space_address_mode): Likewise.
10499 * targhooks.h (default_eh_return_filter_mode): Likewise.
10500 (default_libgcc_cmp_return_mode): Likewise.
10501 (default_libgcc_shift_count_mode): Likewise.
10502 (default_unwind_word_mode): Likewise.
10503 (default_addr_space_pointer_mode): Likewise.
10504 (default_addr_space_address_mode): Likewise.
10505 (default_mode_rep_extended): Take a scalar_int_mode rather than
10507 (default_valid_pointer_mode): Likewise.
10508 (default_addr_space_valid_pointer_mode): Likewise.
10509 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10511 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10512 rather than a machine_mode.
10513 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10514 for local variables.
10515 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10516 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10517 than a machine_mode.
10519 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10520 Alan Hayward <alan.hayward@arm.com>
10521 David Sherwood <david.sherwood@arm.com>
10523 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10524 the mode argument to scalar_int_mode.
10525 (do_jump_by_parts_zero_rtx): Likewise.
10526 (do_jump_by_parts_equality_rtx): Likewise.
10527 (do_jump_by_parts_greater): Take a mode argument.
10528 (do_jump_by_parts_equality): Likewise.
10529 (do_jump_1): Update calls accordingly.
10531 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10532 Alan Hayward <alan.hayward@arm.com>
10533 David Sherwood <david.sherwood@arm.com>
10535 * is-a.h (safe_dyn_cast): New function.
10536 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10537 (jump_table_for_label): Likewise.
10538 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10539 instead of an rtx_insn *.
10540 (shorten_branches): Use dyn_cast instead of LABEL_P and
10541 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10542 rtx_jump_table_data::get_data_mode.
10543 (final_scan_insn): Likewise.
10545 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10546 Alan Hayward <alan.hayward@arm.com>
10547 David Sherwood <david.sherwood@arm.com>
10549 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10550 trying to combine a full-register integer set with a subreg
10553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10554 Alan Hayward <alan.hayward@arm.com>
10555 David Sherwood <david.sherwood@arm.com>
10557 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10558 that is always either address_mode or pointer_mode.
10560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10561 Alan Hayward <alan.hayward@arm.com>
10562 David Sherwood <david.sherwood@arm.com>
10564 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10565 when the two are known to be equal.
10567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10568 Alan Hayward <alan.hayward@arm.com>
10569 David Sherwood <david.sherwood@arm.com>
10571 * simplify-rtx.c (simplify_const_unary_operation): Use
10572 is_a <scalar_int_mode> instead of checking for a nonzero
10573 precision. Forcibly convert op_mode to a scalar_int_mode
10574 in that case. More clearly differentiate the operand and
10575 result modes and use the former when deciding what the value
10576 of a count-bits operation should be. Use is_int_mode instead
10577 of checking for a MODE_INT. Remove redundant check for whether
10578 this mode has a zero precision.
10580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10581 Alan Hayward <alan.hayward@arm.com>
10582 David Sherwood <david.sherwood@arm.com>
10584 * optabs.c (widen_leading): Change the type of the mode argument
10585 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10586 (widen_bswap): Likewise.
10587 (expand_parity): Likewise.
10588 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10589 (expand_ffs): Likewise.
10590 (epand_unop): Check for scalar integer modes before calling the
10593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10594 Alan Hayward <alan.hayward@arm.com>
10595 David Sherwood <david.sherwood@arm.com>
10597 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
10599 (expand_expr_real_1): Update call accordingly.
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 * expmed.c (store_bit_field_using_insv): Add op0_mode and
10606 value_mode arguments. Use scalar_int_mode internally.
10607 (store_bit_field_1): Rename the new integer mode from imode
10608 to op0_mode and use it instead of GET_MODE (op0). Update calls
10609 to store_split_bit_field, store_bit_field_using_insv and
10610 store_fixed_bit_field.
10611 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
10612 Use scalar_int_mode internally. Use a bit count rather than a mode
10613 when calculating the largest bit size for get_best_mode.
10614 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10615 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10616 Remove assertion that OP0 has a scalar integer mode.
10617 (store_split_bit_field): Add op0_mode and value_mode arguments.
10618 Update calls to extract_fixed_bit_field.
10619 (extract_bit_field_using_extv): Add an op0_mode argument.
10620 Use scalar_int_mode internally.
10621 (extract_bit_field_1): Rename the new integer mode from imode to
10622 op0_mode and use it instead of GET_MODE (op0). Update calls to
10623 extract_split_bit_field, extract_bit_field_using_extv and
10624 extract_fixed_bit_field.
10625 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10626 to extract_split_bit_field and extract_fixed_bit_field_1.
10627 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10628 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10629 on the target mode.
10630 (extract_split_bit_field): Add an op0_mode argument. Update call
10631 to extract_fixed_bit_field.
10633 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10634 Alan Hayward <alan.hayward@arm.com>
10635 David Sherwood <david.sherwood@arm.com>
10637 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10638 * explow.c (hard_function_value): Likewise.
10639 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10640 convert_to_mode call outside the loop.
10641 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10642 for the mode iterator. Require the mode specified by max_pieces
10644 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10646 (copy_blkmode_to_reg): Likewise.
10647 (set_storage_via_setmem): Likewise.
10648 * optabs.c (prepare_cmp_insn): Likewise.
10649 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10650 * stor-layout.c (finish_bitfield_representative): Likewise.
10652 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10653 Alan Hayward <alan.hayward@arm.com>
10654 David Sherwood <david.sherwood@arm.com>
10656 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10657 * expr.c (convert_move): Use them.
10658 (convert_modes): Likewise.
10659 (store_expr_with_bounds): Likewise.
10661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10662 Alan Hayward <alan.hayward@arm.com>
10663 David Sherwood <david.sherwood@arm.com>
10665 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10666 parameter for the mode of "x". Remove the "known_x", "known_mode"
10667 and "known_ret" arguments. Change the type of the mode argument
10668 to scalar_int_mode.
10669 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10670 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10671 (reg_num_sign_bit_copies_for_combine): Likewise.
10672 * rtlanal.c (nonzero_bits1): Likewise.
10673 (num_sign_bit_copies1): Likewise.
10674 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10675 (reg_num_sign_bit_copies_general): Likewise.
10676 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10677 (reg_nonzero_bits_general): Likewise.
10679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10680 Alan Hayward <alan.hayward@arm.com>
10681 David Sherwood <david.sherwood@arm.com>
10683 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10684 than in subroutines. Return 1 for non-integer modes.
10685 (cached_num_sign_bit_copies): Change the type of the mode parameter
10686 to scalar_int_mode.
10687 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10688 classes. Handle CONST_INT_P first and then check whether X also
10689 has a scalar integer mode. Check the same thing for inner registers
10690 of a SUBREG and for values that are being extended or truncated.
10692 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10693 Alan Hayward <alan.hayward@arm.com>
10694 David Sherwood <david.sherwood@arm.com>
10696 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10697 in subroutines. Return the mode mask for non-integer modes.
10698 (cached_nonzero_bits): Change the type of the mode parameter
10699 to scalar_int_mode.
10700 (nonzero_bits1): Likewise. Remove early exit for other mode
10701 classes. Handle CONST_INT_P first and then check whether X
10702 also has a scalar integer mode.
10704 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10705 Alan Hayward <alan.hayward@arm.com>
10706 David Sherwood <david.sherwood@arm.com>
10708 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10709 Return a scalar_int_mode and assert that the size is greater than
10711 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10712 (op_by_pieces_d::op_by_pieces_d): Likewise.
10713 (op_by_pieces_d::run): Likewise.
10714 (can_store_by_pieces): Likewise.
10716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10717 Alan Hayward <alan.hayward@arm.com>
10718 David Sherwood <david.sherwood@arm.com>
10720 * combine.c (extract_left_shift): Add a mode argument and update
10722 (make_compound_operation_int): Change the type of the mode parameter
10723 to scalar_int_mode and update the call to extract_left_shift.
10725 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10726 Alan Hayward <alan.hayward@arm.com>
10727 David Sherwood <david.sherwood@arm.com>
10729 * combine.c (simplify_and_const_int): Change the type of the mode
10730 parameter to scalar_int_mode.
10731 (simplify_and_const_int_1): Likewise. Update recursive call.
10733 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10734 Alan Hayward <alan.hayward@arm.com>
10735 David Sherwood <david.sherwood@arm.com>
10737 * combine.c (simplify_compare_const): Check that the mode is a
10738 scalar_int_mode (rather than VOIDmode) before testing its
10740 (simplify_comparison): Move COMPARISON_P handling out of the
10741 loop and restrict the latter part of the loop to scalar_int_modes.
10742 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10743 and when considering SUBREG_REGs. Use is_int_mode instead of
10744 checking GET_MODE_CLASS against MODE_INT.
10746 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10747 Alan Hayward <alan.hayward@arm.com>
10748 David Sherwood <david.sherwood@arm.com>
10750 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10751 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10752 shift_unit_mode and for modes involved in scalar shifts.
10754 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10755 Alan Hayward <alan.hayward@arm.com>
10756 David Sherwood <david.sherwood@arm.com>
10758 * combine.c (force_int_to_mode): New function, split out from...
10759 (force_to_mode): ...here. Keep xmode up-to-date and use it
10760 instead of GET_MODE (x).
10762 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10763 Alan Hayward <alan.hayward@arm.com>
10764 David Sherwood <david.sherwood@arm.com>
10766 * optabs-query.h (extraction_insn::struct_mode): Change type to
10767 opt_scalar_int_mode and update comment.
10768 (extraction_insn::field_mode): Change type to scalar_int_mode.
10769 (extraction_insn::pos_mode): Likewise.
10770 * combine.c (make_extraction): Update accordingly.
10771 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10772 (get_optab_extraction_insn): Likewise.
10773 * recog.c (simplify_while_replacing): Likewise.
10774 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10775 parameter to opt_scalar_int_mode.
10777 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10778 Alan Hayward <alan.hayward@arm.com>
10779 David Sherwood <david.sherwood@arm.com>
10781 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10782 to a scalar_int_mode instead of a machine_mode.
10783 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10784 (get_best_mode): Return a boolean and use a pointer argument to store
10785 the selected mode. Replace the limit mode parameter with a bit limit.
10786 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10787 for the values returned by bit_field_mode_iterator::next_mode.
10788 (store_bit_field): Update call to get_best_mode.
10789 (store_fixed_bit_field): Likewise.
10790 (extract_fixed_bit_field): Likewise.
10791 * expr.c (optimize_bitfield_assignment_op): Likewise.
10792 * fold-const.c (optimize_bit_field_compare): Likewise.
10793 (fold_truth_andor_1): Likewise.
10794 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10795 Update for new type of m_mode.
10796 (get_best_mode): As above.
10798 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10799 Alan Hayward <alan.hayward@arm.com>
10800 David Sherwood <david.sherwood@arm.com>
10802 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10803 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10804 (store_bit_field): Check is_a <scalar_int_mode> before calling
10805 strict_volatile_bitfield_p.
10806 (extract_bit_field): Likewise.
10808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10809 Alan Hayward <alan.hayward@arm.com>
10810 David Sherwood <david.sherwood@arm.com>
10812 * target.def (cstore_mode): Return a scalar_int_mode.
10813 * doc/tm.texi: Regenerate.
10814 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10815 * targhooks.h (default_cstore_mode): Likewise.
10816 * targhooks.c (default_cstore_mode): Likewise, using a forced
10818 * expmed.c (emit_cstore): Expect the target of the cstore to be
10821 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10822 Alan Hayward <alan.hayward@arm.com>
10823 David Sherwood <david.sherwood@arm.com>
10825 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10827 (niter_desc): Likewise mode.
10828 (iv_analyze): Add a mode parameter.
10830 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10831 and change its type to scalar_int_mode.
10832 * loop-iv.c: Update commentary at head of file.
10833 (iv_constant): Pass the mode paraeter before the rtx it describes
10834 and change its type to scalar_int_mode. Remove VOIDmode handling.
10835 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10836 (iv_extend): Likewise.
10837 (shorten_into_mode): Likewise.
10838 (iv_add): Use scalar_int_mode.
10839 (iv_mult): Likewise.
10840 (iv_shift): Likewise.
10841 (canonicalize_iv_subregs): Likewise.
10842 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10843 it describes and change its mode to scalar_int_mode. Also change
10844 the type of the returned inner_mode to scalar_int_mode.
10845 (get_biv_step): Likewise, turning outer_mode from a pointer
10846 into a direct parameter. Update call to get_biv_step_1.
10847 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10848 iv_constant and get_biv_step.
10849 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10850 and change its type to scalar_int_mode. Don't initialise iv->mode
10851 to VOIDmode and remove later checks for its still being VOIDmode.
10852 Update calls to iv_analyze_op and iv_analyze_expr. Check
10853 is_a <scalar_int_mode> when changing the mode under consideration.
10854 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10855 Update call to iv_analyze_expr.
10856 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10857 inner register is not also a scalar_int_mode. Update call to
10859 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10860 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10861 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10862 separate mode class checks. Update calls to iv_analyze. Remove
10863 fix-up of VOIDmodes after iv_analyze_biv.
10864 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10865 don't have a scalar_int_mode. Update call to biv_p.
10867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10868 Alan Hayward <alan.hayward@arm.com>
10869 David Sherwood <david.sherwood@arm.com>
10871 * cfgexpand.c (convert_debug_memory_address): Use
10872 as_a <scalar_int_mode>.
10873 * combine.c (expand_compound_operation): Likewise.
10874 (make_extraction): Likewise.
10875 (change_zero_ext): Likewise.
10876 (simplify_comparison): Likewise.
10877 * cse.c (cse_insn): Likewise.
10878 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10879 (mem_loc_descriptor): Likewise.
10880 (loc_descriptor): Likewise.
10881 * expmed.c (init_expmed_one_mode): Likewise.
10882 (synth_mult): Likewise.
10883 (emit_store_flag_1): Likewise.
10884 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10885 of a comparison with size.
10886 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10887 (reduce_to_bit_field_precision): Likewise.
10888 * function.c (expand_function_end): Likewise.
10889 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10890 * loop-doloop.c (doloop_modify): Likewise.
10891 * optabs.c (expand_binop): Likewise.
10892 (expand_unop): Likewise.
10893 (expand_copysign_absneg): Likewise.
10894 (prepare_cmp_insn): Likewise.
10895 (maybe_legitimize_operand): Likewise.
10896 * recog.c (const_scalar_int_operand): Likewise.
10897 * rtlanal.c (get_address_mode): Likewise.
10898 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10899 (simplify_cond_clz_ctz): Likewise.
10900 * tree-nested.c (get_nl_goto_field): Likewise.
10901 * tree.c (build_vector_type_for_mode): Likewise.
10902 * var-tracking.c (use_narrower_mode): Likewise.
10904 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10905 Alan Hayward <alan.hayward@arm.com>
10906 David Sherwood <david.sherwood@arm.com>
10908 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10909 * builtins.c (expand_builtin_signbit): Use it.
10910 * cfgexpand.c (expand_debug_expr): Likewise.
10911 * dojump.c (do_jump): Likewise.
10912 (do_compare_and_jump): Likewise.
10913 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
10914 * expmed.c (make_tree): Likewise.
10915 * expr.c (expand_expr_real_2): Likewise.
10916 (expand_expr_real_1): Likewise.
10917 (try_casesi): Likewise.
10918 * fold-const-call.c (fold_const_call_ss): Likewise.
10919 * fold-const.c (unextend): Likewise.
10920 (extract_muldiv_1): Likewise.
10921 (fold_single_bit_test): Likewise.
10922 (native_encode_int): Likewise.
10923 (native_encode_string): Likewise.
10924 (native_interpret_int): Likewise.
10925 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
10926 * internal-fn.c (expand_addsub_overflow): Likewise.
10927 (expand_neg_overflow): Likewise.
10928 (expand_mul_overflow): Likewise.
10929 (expand_arith_overflow): Likewise.
10930 * match.pd: Likewise.
10931 * stor-layout.c (layout_type): Likewise.
10932 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10933 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
10934 * tree-ssanames.c (get_range_info): Likewise.
10935 * tree-switch-conversion.c (array_value_type) Likewise.
10936 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
10937 (vect_recog_divmod_pattern): Likewise.
10938 (vect_recog_mixed_size_cond_pattern): Likewise.
10939 * tree-vrp.c (extract_range_basic): Likewise.
10940 (simplify_float_conversion_using_ranges): Likewise.
10941 * tree.c (int_fits_type_p): Likewise.
10942 * ubsan.c (instrument_bool_enum_load): Likewise.
10943 * varasm.c (mergeable_string_section): Likewise.
10944 (narrowing_initializer_constant_valid_p): Likewise.
10945 (output_constant): Likewise.
10947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10948 Alan Hayward <alan.hayward@arm.com>
10949 David Sherwood <david.sherwood@arm.com>
10951 * machmode.h (NARROWEST_INT_MODE): New macro.
10952 * expr.c (alignment_for_piecewise_move): Use it instead of
10953 GET_CLASS_NARROWEST_MODE (MODE_INT).
10954 (push_block): Likewise.
10955 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
10957 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10959 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10960 Alan Hayward <alan.hayward@arm.com>
10961 David Sherwood <david.sherwood@arm.com>
10963 * postreload.c (move2add_valid_value_p): Change the type of the
10964 mode parameter to scalar_int_mode.
10965 (move2add_use_add2_insn): Add a mode parameter and use it instead
10967 (move2add_use_add3_insn): Likewise.
10968 (reload_cse_move2add): Update accordingly.
10970 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10971 Alan Hayward <alan.hayward@arm.com>
10972 David Sherwood <david.sherwood@arm.com>
10974 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
10976 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
10977 * optabs.c (expand_unop): Likewise.
10979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10980 Alan Hayward <alan.hayward@arm.com>
10981 David Sherwood <david.sherwood@arm.com>
10983 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
10984 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
10985 (popcount_loc_descriptor): Likewise.
10986 (bswap_loc_descriptor): Likewise.
10987 (rotate_loc_descriptor): Likewise.
10988 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
10989 calling the functions above.
10991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10992 Alan Hayward <alan.hayward@arm.com>
10993 David Sherwood <david.sherwood@arm.com>
10995 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
10997 (try_combine): Likewise.
10998 (simplify_if_then_else): Likewise.
10999 * cse.c (cse_insn): Likewise.
11000 * dwarf2out.c (mem_loc_descriptor): Likewise.
11001 * emit-rtl.c (gen_lowpart_common): Likewise.
11002 * simplify-rtx.c (simplify_truncation): Likewise.
11003 (simplify_binary_operation_1): Likewise.
11004 (simplify_const_relational_operation): Likewise.
11005 (simplify_ternary_operation): Likewise.
11006 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11008 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11009 Alan Hayward <alan.hayward@arm.com>
11010 David Sherwood <david.sherwood@arm.com>
11012 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11013 * reload.c (push_reload): Likewise.
11014 (find_reloads): Likewise.
11016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11017 Alan Hayward <alan.hayward@arm.com>
11018 David Sherwood <david.sherwood@arm.com>
11020 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11021 (make_compound_operation_int): Likewise.
11022 (change_zero_ext): Likewise.
11023 * expr.c (convert_move): Likewise.
11024 (convert_modes): Likewise.
11025 * fwprop.c (forward_propagate_subreg): Likewise.
11026 * loop-iv.c (get_biv_step_1): Likewise.
11027 * optabs.c (widen_operand): Likewise.
11028 * postreload.c (move2add_valid_value_p): Likewise.
11029 * recog.c (simplify_while_replacing): Likewise.
11030 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11031 (simplify_binary_operation_1): Likewise. Remove redundant
11032 mode equality check.
11034 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11035 Alan Hayward <alan.hayward@arm.com>
11036 David Sherwood <david.sherwood@arm.com>
11038 * combine.c (combine_simplify_rtx): Add checks for
11039 is_a <scalar_int_mode>.
11040 (simplify_if_then_else): Likewise.
11041 (make_field_assignment): Likewise.
11042 (simplify_comparison): Likewise.
11043 * ifcvt.c (noce_try_bitop): Likewise.
11044 * loop-invariant.c (canonicalize_address_mult): Likewise.
11045 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
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 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11052 is_a <scalar_int_mode> instead of != BLKmode.
11054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11055 Alan Hayward <alan.hayward@arm.com>
11056 David Sherwood <david.sherwood@arm.com>
11058 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11059 instead of != VOIDmode.
11060 * combine.c (if_then_else_cond): Likewise.
11061 (change_zero_ext): Likewise.
11062 * dwarf2out.c (mem_loc_descriptor): Likewise.
11063 (loc_descriptor): Likewise.
11064 * rtlanal.c (canonicalize_condition): Likewise.
11065 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11067 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11068 Alan Hayward <alan.hayward@arm.com>
11069 David Sherwood <david.sherwood@arm.com>
11071 * simplify-rtx.c (simplify_binary_operation_1): Use
11072 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11075 Alan Hayward <alan.hayward@arm.com>
11076 David Sherwood <david.sherwood@arm.com>
11078 * wide-int.h (int_traits<unsigned char>) New class.
11079 (int_traits<unsigned short>) Likewise.
11080 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11081 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11083 * combine.c (set_nonzero_bits_and_sign_copies): Use
11084 is_a <scalar_int_mode>.
11085 (find_split_point): Likewise.
11086 (combine_simplify_rtx): Likewise.
11087 (simplify_logical): Likewise.
11088 (expand_compound_operation): Likewise.
11089 (expand_field_assignment): Likewise.
11090 (make_compound_operation): Likewise.
11091 (extended_count): Likewise.
11092 (change_zero_ext): Likewise.
11093 (simplify_comparison): Likewise.
11094 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11095 (ucompare_loc_descriptor): Likewise.
11096 (minmax_loc_descriptor): Likewise.
11097 (mem_loc_descriptor): Likewise.
11098 (loc_descriptor): Likewise.
11099 * expmed.c (init_expmed_one_mode): Likewise.
11100 * lra-constraints.c (lra_constraint_offset): Likewise.
11101 * optabs.c (prepare_libcall_arg): Likewise.
11102 * postreload.c (move2add_note_store): Likewise.
11103 * reload.c (operands_match_p): Likewise.
11104 * rtl.h (load_extend_op): Likewise.
11105 * rtlhooks.c (gen_lowpart_general): Likewise.
11106 * simplify-rtx.c (simplify_truncation): Likewise.
11107 (simplify_unary_operation_1): Likewise.
11108 (simplify_binary_operation_1): Likewise.
11109 (simplify_const_binary_operation): Likewise.
11110 (simplify_const_relational_operation): Likewise.
11111 (simplify_subreg): Likewise.
11112 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11113 * var-tracking.c (adjust_mems): Likewise.
11114 (prepare_call_arguments): Likewise.
11116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11117 Alan Hayward <alan.hayward@arm.com>
11118 David Sherwood <david.sherwood@arm.com>
11120 * machmode.h (is_int_mode): New fuction.
11121 * combine.c (find_split_point): Use it.
11122 (combine_simplify_rtx): Likewise.
11123 (simplify_if_then_else): Likewise.
11124 (simplify_set): Likewise.
11125 (simplify_shift_const_1): Likewise.
11126 (simplify_comparison): Likewise.
11127 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
11128 * cse.c (notreg_cost): Likewise.
11129 (cse_insn): Likewise.
11130 * cselib.c (cselib_lookup_1): Likewise.
11131 * dojump.c (do_jump_1): Likewise.
11132 (do_compare_rtx_and_jump): Likewise.
11133 * dse.c (get_call_args): Likewise.
11134 * dwarf2out.c (rtl_for_decl_init): Likewise.
11135 (native_encode_initializer): Likewise.
11136 * expmed.c (emit_store_flag_1): Likewise.
11137 (emit_store_flag): Likewise.
11138 * expr.c (convert_modes): Likewise.
11139 (store_field): Likewise.
11140 (expand_expr_real_1): Likewise.
11141 * fold-const.c (fold_read_from_constant_string): Likewise.
11142 * gimple-ssa-sprintf.c (get_format_string): Likewise.
11143 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
11144 * optabs.c (expand_binop): Likewise.
11145 (expand_unop): Likewise.
11146 (expand_abs_nojump): Likewise.
11147 (expand_one_cmpl_abs_nojump): Likewise.
11148 * simplify-rtx.c (mode_signbit_p): Likewise.
11149 (val_signbit_p): Likewise.
11150 (val_signbit_known_set_p): Likewise.
11151 (val_signbit_known_clear_p): Likewise.
11152 (simplify_relational_operation_1): Likewise.
11153 * tree.c (vector_type_mode): Likewise.
11155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11156 Alan Hayward <alan.hayward@arm.com>
11157 David Sherwood <david.sherwood@arm.com>
11159 * machmode.h (smallest_mode_for_size): Fix formatting.
11160 (smallest_int_mode_for_size): New function.
11161 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11162 instead of smallest_mode_for_size.
11163 * combine.c (make_extraction): Likewise.
11164 * config/arc/arc.c (arc_expand_movmem): Likewise.
11165 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11166 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11167 * config/s390/s390.c (s390_expand_insv): Likewise.
11168 * config/sparc/sparc.c (assign_int_registers): Likewise.
11169 * config/spu/spu.c (spu_function_value): Likewise.
11170 (spu_function_arg): Likewise.
11171 * coverage.c (get_gcov_type): Likewise.
11172 (get_gcov_unsigned_t): Likewise.
11173 * dse.c (find_shift_sequence): Likewise.
11174 * expmed.c (store_bit_field_1): Likewise.
11175 * expr.c (convert_move): Likewise.
11176 (store_field): Likewise.
11177 * internal-fn.c (expand_arith_overflow): Likewise.
11178 * optabs-query.c (get_best_extraction_insn): Likewise.
11179 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11180 * stor-layout.c (layout_type): Likewise.
11181 (initialize_sizetypes): Likewise.
11182 * targhooks.c (default_get_mask_mode): Likewise.
11183 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11185 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11186 Alan Hayward <alan.hayward@arm.com>
11187 David Sherwood <david.sherwood@arm.com>
11189 * machmode.h (opt_mode::else_blk): New function.
11190 (int_mode_for_mode): Declare.
11191 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11192 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11194 * cfgexpand.c (expand_debug_expr): Likewise.
11195 * combine.c (gen_lowpart_or_truncate): Likewise.
11196 (gen_lowpart_for_combine): Likewise.
11197 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11198 * config/avr/avr.c (avr_to_int_mode): Likewise.
11199 (avr_out_plus_1): Likewise.
11200 (avr_out_plus): Likewise.
11201 (avr_out_round): Likewise.
11202 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11203 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11204 (s390_expand_vcond): Likewise.
11205 * config/spu/spu.c (spu_split_immediate): Likewise.
11206 (spu_expand_mov): Likewise.
11207 * dse.c (get_stored_val): Likewise.
11208 * expmed.c (store_bit_field_1): Likewise.
11209 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11211 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11212 (extract_low_bits): Likewise.
11213 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11214 handling rather than repeating the check.
11215 (emit_group_store): Likewise.
11216 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11217 * optabs.c (expand_absneg_bit): Likewise.
11218 (expand_copysign_absneg): Likewise.
11219 (expand_copysign_bit): Likewise.
11220 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11221 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11222 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11223 * var-tracking.c (prepare_call_arguments): Likewise.
11224 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11225 int_mode_for_mode instead of mode_for_size.
11226 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11229 Alan Hayward <alan.hayward@arm.com>
11230 David Sherwood <david.sherwood@arm.com>
11232 * machmode.h (int_mode_for_size): New function.
11233 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11234 instead of mode_for_size.
11235 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11237 * combine.c (expand_field_assignment): Use int_mode_for_size
11238 instead of mode_for_size.
11239 (make_extraction): Likewise.
11240 (simplify_shift_const_1): Likewise.
11241 (simplify_comparison): Likewise.
11242 * dojump.c (do_jump): Likewise.
11243 * dwarf2out.c (mem_loc_descriptor): Likewise.
11244 * emit-rtl.c (init_derived_machine_modes): Likewise.
11245 * expmed.c (flip_storage_order): Likewise.
11246 (convert_extracted_bit_field): Likewise.
11247 * expr.c (copy_blkmode_from_reg): Likewise.
11248 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11249 * internal-fn.c (expand_mul_overflow): Likewise.
11250 * lower-subreg.c (simple_move): Likewise.
11251 * optabs-libfuncs.c (init_optabs): Likewise.
11252 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11253 * tree.c (vector_type_mode): Likewise.
11254 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11255 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11256 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11257 * tree-vect-stmts.c (vectorizable_load): Likewise.
11258 (vectorizable_store): Likewise.
11260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11261 Alan Hayward <alan.hayward@arm.com>
11262 David Sherwood <david.sherwood@arm.com>
11264 * coretypes.h (pod_mode): New type.
11265 (scalar_int_mode_pod): New typedef.
11266 * machmode.h (pod_mode): New class.
11267 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11268 * genmodes.c (emit_mode_int_n): Update accordingly.
11269 * lower-subreg.h (target_lower_subreg): Change type to
11270 scalar_int_mode_pod.
11271 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11272 scalar_int_mode_pod.
11274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11275 Alan Hayward <alan.hayward@arm.com>
11276 David Sherwood <david.sherwood@arm.com>
11278 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11279 machine_mode to scalar_int_mode.
11280 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11281 (rs6000_option_override_internal): Remove cast to int.
11282 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11283 machine_mode to scalar_int_mode.
11284 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11285 (rs6000_option_override_internal): Remove cast to int.
11286 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11287 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11289 * config/s390/s390.c (s390_expand_builtin): Likewise.
11290 * coretypes.h (scalar_int_mode): New type.
11291 (opt_scalar_int_mode): New typedef.
11292 * machmode.h (scalar_int_mode): New class.
11293 (scalar_int_mode::includes_p): New function.
11294 (byte_mode): Change type to scalar_int_mode.
11295 (word_mode): Likewise.
11296 (ptr_mode): Likewise.
11297 * emit-rtl.c (byte_mode): Likewise.
11298 (word_mode): Likewise.
11299 (ptr_mode): Likewise.
11300 (init_derived_machine_modes): Update accordingly.
11301 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11302 and MODE_PARTIAL_INT.
11303 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11304 opt_scalar_int_mode.
11306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11307 Alan Hayward <alan.hayward@arm.com>
11308 David Sherwood <david.sherwood@arm.com>
11310 * target.def (libgcc_floating_mode_supported_p): Take a
11312 * doc/tm.texi: Regenerate.
11313 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11315 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11316 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11320 Alan Hayward <alan.hayward@arm.com>
11321 David Sherwood <david.sherwood@arm.com>
11323 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11324 * doc/tm.texi: Regenerate.
11325 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11326 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11327 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11328 * targhooks.h (default_floatn_mode): Likewise.
11329 * targhooks.c (default_floatn_mode): Likewise.
11330 * tree.c (build_common_tree_nodes): Update accordingly.
11332 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11333 Alan Hayward <alan.hayward@arm.com>
11334 David Sherwood <david.sherwood@arm.com>
11336 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11337 (mode_iterator::iterate_p): Likewise.
11338 (mode_iterator::get_wider): Likewise.
11339 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11341 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11342 Alan Hayward <alan.hayward@arm.com>
11343 David Sherwood <david.sherwood@arm.com>
11345 * coretypes.h (opt_scalar_float_mode): New typedef.
11346 * machmode.h (float_mode_for_size): New function.
11347 * emit-rtl.c (double_mode): Delete.
11348 (init_emit_once): Use float_mode_for_size.
11349 * stor-layout.c (layout_type): Likewise.
11350 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11353 Alan Hayward <alan.hayward@arm.com>
11354 David Sherwood <david.sherwood@arm.com>
11356 * output.h (assemble_real): Take a scalar_float_mode.
11357 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11358 * config/arm/arm.md (consttable_4): Likewise.
11359 (consttable_8): Likewise.
11360 (consttable_16): Likewise.
11361 * config/mips/mips.md (consttable_float): Likewise.
11362 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11363 * varasm.c (assemble_real): Take a scalar_float_mode.
11364 (output_constant_pool_2): Update accordingly.
11365 (output_constant): Likewise.
11367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11368 Alan Hayward <alan.hayward@arm.com>
11369 David Sherwood <david.sherwood@arm.com>
11371 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11372 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11373 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11374 (native_encode_real): Likewise.
11375 (native_interpret_real): Likewise.
11376 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11377 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11379 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11380 Alan Hayward <alan.hayward@arm.com>
11381 David Sherwood <david.sherwood@arm.com>
11383 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11384 <scalar_float_mode>. Simplify.
11385 (gen_extend_conv_libfunc): Likewise.
11387 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11388 Alan Hayward <alan.hayward@arm.com>
11389 David Sherwood <david.sherwood@arm.com>
11391 * coretypes.h (scalar_float_mode): New type.
11392 * machmode.h (mode_traits::from_int): Use machine_mode if
11393 USE_ENUM_MODES is defined.
11394 (is_a): New function.
11396 (dyn_cast): Likewise.
11397 (scalar_float_mode): New class.
11398 (scalar_float_mode::includes_p): New function.
11399 (is_float_mode): Likewise.
11400 * gdbhooks.py (MachineModePrinter): New class.
11401 (build_pretty_printer): Use it for scalar_float_mode.
11402 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11403 (format_helper::format_helper): Turn into a template.
11404 * genmodes.c (get_mode_class): New function.
11405 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11406 or machine_mode if none.
11407 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11408 as_a <scalar_float_mode>.
11409 * dwarf2out.c (mem_loc_descriptor): Likewise.
11410 (insert_float): Likewise.
11411 (add_const_value_attribute): Likewise.
11412 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11413 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11414 (expand_unop): Update accordingly.
11415 (expand_abs_nojump): Likewise.
11416 (expand_copysign_absneg): Take a scalar_float_mode.
11417 (expand_copysign_bit): Likewise.
11418 (expand_copysign): Update accordingly.
11420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11421 Alan Hayward <alan.hayward@arm.com>
11422 David Sherwood <david.sherwood@arm.com>
11424 * coretypes.h (opt_mode): New class.
11425 * machmode.h (opt_mode): Likewise.
11426 (opt_mode::else_void): New function.
11427 (opt_mode::require): Likewise.
11428 (opt_mode::exists): Likewise.
11429 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11430 (GET_MODE_2XWIDER_MODE): Likewise.
11431 (mode_iterator::get_wider): Update accordingly.
11432 (mode_iterator::get_2xwider): Likewise.
11433 (mode_iterator::get_known_wider): Likewise, turning into a template.
11434 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11435 forcing a wider mode to exist.
11436 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11437 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11438 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11439 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11440 * lower-subreg.c (init_lower_subreg): Likewise.
11441 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11442 on the final iteration.
11443 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11444 a wider mode exists before asking for a move pattern.
11445 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11446 forcing a wider mode to exist.
11447 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11448 returning false if no such mode exists.
11449 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11450 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11451 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11452 Avoid checking for a MODE_INT if we already know the mode is not a
11454 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11455 forcing a wider mode to exist.
11456 (expmed_mult_highpart_optab): Likewise.
11457 (expmed_mult_highpart): Likewise.
11458 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11460 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11461 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11462 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11463 * internal-fn.c (expand_mul_overflow): Update use of
11464 GET_MODE_2XWIDER_MODE.
11465 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11466 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11467 GET_MODE_WIDER_MODE.
11468 (convert_plusminus_to_widen): Likewise.
11469 * tree-switch-conversion.c (array_value_type): Likewise.
11470 * var-tracking.c (emit_note_insn_var_location): Likewise.
11471 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11472 Return false inside rather than outside the loop if no wider mode
11474 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11475 and GET_MODE_2XWIDER_MODE
11476 (can_compare_p): Use else_void.
11477 * gdbhooks.py (OptMachineModePrinter): New class.
11478 (build_pretty_printer): Use it for opt_mode.
11480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11481 Alan Hayward <alan.hayward@arm.com>
11482 David Sherwood <david.sherwood@arm.com>
11484 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11485 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11488 Alan Hayward <alan.hayward@arm.com>
11489 David Sherwood <david.sherwood@arm.com>
11491 * machmode.h (mode_traits): New structure.
11492 (get_narrowest_mode): New function.
11493 (mode_iterator::start): Likewise.
11494 (mode_iterator::iterate_p): Likewise.
11495 (mode_iterator::get_wider): Likewise.
11496 (mode_iterator::get_known_wider): Likewise.
11497 (mode_iterator::get_2xwider): Likewise.
11498 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11499 (FOR_EACH_MODE): Likewise.
11500 (FOR_EACH_MODE_FROM): Likewise.
11501 (FOR_EACH_MODE_UNTIL): Likewise.
11502 (FOR_EACH_WIDER_MODE): Likewise.
11503 (FOR_EACH_2XWIDER_MODE): Likewise.
11504 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11505 * combine.c (simplify_comparison): Likewise
11506 * config/i386/i386.c (type_natural_mode): Likewise.
11507 * cse.c (cse_insn): Likewise.
11508 * dse.c (find_shift_sequence): Likewise.
11509 * emit-rtl.c (init_derived_machine_modes): Likewise.
11510 (init_emit_once): Likewise.
11511 * explow.c (hard_function_value): Likewise.
11512 * expmed.c (extract_fixed_bit_field_1): Likewise.
11513 (extract_bit_field_1): Likewise.
11514 (expand_divmod): Likewise.
11515 (emit_store_flag_1): Likewise.
11516 * expr.c (init_expr_target): Likewise.
11517 (convert_move): Likewise.
11518 (alignment_for_piecewise_move): Likewise.
11519 (widest_int_mode_for_size): Likewise.
11520 (emit_block_move_via_movmem): Likewise.
11521 (copy_blkmode_to_reg): Likewise.
11522 (set_storage_via_setmem): Likewise.
11523 (compress_float_constant): Likewise.
11524 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11525 * optabs-query.c (get_best_extraction_insn): Likewise.
11526 * optabs.c (expand_binop): Likewise.
11527 (expand_twoval_unop): Likewise.
11528 (expand_twoval_binop): Likewise.
11529 (widen_leading): Likewise.
11530 (widen_bswap): Likewise.
11531 (expand_parity): Likewise.
11532 (expand_unop): Likewise.
11533 (prepare_cmp_insn): Likewise.
11534 (prepare_float_lib_cmp): Likewise.
11535 (expand_float): Likewise.
11536 (expand_fix): Likewise.
11537 (expand_sfix_optab): Likewise.
11538 * postreload.c (move2add_use_add2_insn): Likewise.
11539 * reg-stack.c (reg_to_stack): Likewise.
11540 * reginfo.c (choose_hard_reg_mode): Likewise.
11541 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11542 * stor-layout.c (mode_for_size): Likewise.
11543 (smallest_mode_for_size): Likewise.
11544 (mode_for_vector): Likewise.
11545 (finish_bitfield_representative): Likewise.
11546 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11547 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11548 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11549 * var-tracking.c (prepare_call_arguments): Likewise.
11551 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11552 Alan Hayward <alan.hayward@arm.com>
11553 David Sherwood <david.sherwood@arm.com>
11555 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11556 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11557 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11558 * machmode.h (mode_size): Move earlier in file.
11559 (mode_precision): Likewise.
11560 (mode_inner): Likewise.
11561 (mode_nunits): Likewise.
11562 (mode_unit_size): Likewise.
11563 (unit_unit_precision): Likewise.
11564 (mode_wider): Likewise.
11565 (mode_2xwider): Likewise.
11566 (machine_mode): New class.
11567 (mode_to_bytes): New function.
11568 (mode_to_bits): Likewise.
11569 (mode_to_precision): Likewise.
11570 (mode_to_inner): Likewise.
11571 (mode_to_unit_size): Likewise.
11572 (mode_to_unit_precision): Likewise.
11573 (mode_to_nunits): Likewise.
11574 (GET_MODE_SIZE): Use mode_to_bytes.
11575 (GET_MODE_BITSIZE): Use mode_to_bits.
11576 (GET_MODE_PRECISION): Use mode_to_precision.
11577 (GET_MODE_INNER): Use mode_to_inner.
11578 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11579 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11580 (GET_MODE_NUNITS): Use mode_to_nunits.
11581 * system.h (ALWAYS_INLINE): New macro.
11582 * config/powerpcspe/powerpcspe-c.c
11583 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11584 int for arg1_mode and arg2_mode.
11586 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11587 Alan Hayward <alan.hayward@arm.com>
11588 David Sherwood <david.sherwood@arm.com>
11590 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11591 Prefix mode names with E_ in case statements.
11592 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11593 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11594 (aarch64_split_simd_move): Likewise.
11595 (aarch64_gen_storewb_pair): Likewise.
11596 (aarch64_gen_loadwb_pair): Likewise.
11597 (aarch64_gen_store_pair): Likewise.
11598 (aarch64_gen_load_pair): Likewise.
11599 (aarch64_get_condition_code_1): Likewise.
11600 (aarch64_constant_pool_reload_icode): Likewise.
11601 (get_rsqrte_type): Likewise.
11602 (get_rsqrts_type): Likewise.
11603 (get_recpe_type): Likewise.
11604 (get_recps_type): Likewise.
11605 (aarch64_gimplify_va_arg_expr): Likewise.
11606 (aarch64_simd_container_mode): Likewise.
11607 (aarch64_emit_load_exclusive): Likewise.
11608 (aarch64_emit_store_exclusive): Likewise.
11609 (aarch64_expand_compare_and_swap): Likewise.
11610 (aarch64_gen_atomic_cas): Likewise.
11611 (aarch64_emit_bic): Likewise.
11612 (aarch64_emit_atomic_swap): Likewise.
11613 (aarch64_emit_atomic_load_op): Likewise.
11614 (aarch64_evpc_trn): Likewise.
11615 (aarch64_evpc_uzp): Likewise.
11616 (aarch64_evpc_zip): Likewise.
11617 (aarch64_evpc_ext): Likewise.
11618 (aarch64_evpc_rev): Likewise.
11619 (aarch64_evpc_dup): Likewise.
11620 (aarch64_gen_ccmp_first): Likewise.
11621 (aarch64_gen_ccmp_next): Likewise.
11622 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11623 (alpha_emit_xfloating_libcall): Likewise.
11624 (emit_insxl): Likewise.
11625 (alpha_arg_type): Likewise.
11626 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11627 (arc_preferred_simd_mode): Likewise.
11628 (arc_secondary_reload): Likewise.
11629 (get_arc_condition_code): Likewise.
11630 (arc_print_operand): Likewise.
11631 (arc_legitimate_constant_p): Likewise.
11632 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11633 * config/arc/arc.md (casesi_load): Likewise.
11634 (casesi_compact_jump): Likewise.
11635 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11636 (cc_use_register): Likewise.
11637 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11638 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11639 (arm_init_iwmmxt_builtins): Likewise.
11640 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11641 (neon_expand_vector_init): Likewise.
11642 (arm_attr_length_move_neon): Likewise.
11643 (maybe_get_arm_condition_code): Likewise.
11644 (arm_emit_vector_const): Likewise.
11645 (arm_preferred_simd_mode): Likewise.
11646 (arm_output_iwmmxt_tinsr): Likewise.
11647 (thumb1_output_casesi): Likewise.
11648 (thumb2_output_casesi): Likewise.
11649 (arm_emit_load_exclusive): Likewise.
11650 (arm_emit_store_exclusive): Likewise.
11651 (arm_expand_compare_and_swap): Likewise.
11652 (arm_evpc_neon_vuzp): Likewise.
11653 (arm_evpc_neon_vzip): Likewise.
11654 (arm_evpc_neon_vrev): Likewise.
11655 (arm_evpc_neon_vtrn): Likewise.
11656 (arm_evpc_neon_vext): Likewise.
11657 (arm_validize_comparison): Likewise.
11658 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11659 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11660 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11661 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11662 (c6x_preferred_simd_mode): Likewise.
11663 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11664 (epiphany_rtx_costs): Likewise.
11665 * config/epiphany/predicates.md (proper_comparison_operator):
11667 * config/frv/frv.c (condexec_memory_operand): Likewise.
11668 (frv_emit_move): Likewise.
11669 (output_move_single): Likewise.
11670 (output_condmove_single): Likewise.
11671 (frv_hard_regno_mode_ok): Likewise.
11672 (frv_matching_accg_mode): Likewise.
11673 * config/h8300/h8300.c (split_adds_subs): Likewise.
11674 (h8300_rtx_costs): Likewise.
11675 (h8300_print_operand): Likewise.
11676 (compute_mov_length): Likewise.
11677 (output_logical_op): Likewise.
11678 (compute_logical_op_length): Likewise.
11679 (compute_logical_op_cc): Likewise.
11680 (h8300_shift_needs_scratch_p): Likewise.
11681 (output_a_shift): Likewise.
11682 (compute_a_shift_length): Likewise.
11683 (compute_a_shift_cc): Likewise.
11684 (expand_a_rotate): Likewise.
11685 (output_a_rotate): Likewise.
11686 * config/i386/i386.c (classify_argument): Likewise.
11687 (function_arg_advance_32): Likewise.
11688 (function_arg_32): Likewise.
11689 (function_arg_64): Likewise.
11690 (function_value_64): Likewise.
11691 (ix86_gimplify_va_arg): Likewise.
11692 (ix86_legitimate_constant_p): Likewise.
11693 (put_condition_code): Likewise.
11694 (split_double_mode): Likewise.
11695 (ix86_avx256_split_vector_move_misalign): Likewise.
11696 (ix86_expand_vector_logical_operator): Likewise.
11697 (ix86_split_idivmod): Likewise.
11698 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11699 (ix86_build_const_vector): Likewise.
11700 (ix86_build_signbit_mask): Likewise.
11701 (ix86_match_ccmode): Likewise.
11702 (ix86_cc_modes_compatible): Likewise.
11703 (ix86_expand_branch): Likewise.
11704 (ix86_expand_sse_cmp): Likewise.
11705 (ix86_expand_sse_movcc): Likewise.
11706 (ix86_expand_int_sse_cmp): Likewise.
11707 (ix86_expand_vec_perm_vpermi2): Likewise.
11708 (ix86_expand_vec_perm): Likewise.
11709 (ix86_expand_sse_unpack): Likewise.
11710 (ix86_expand_int_addcc): Likewise.
11711 (ix86_split_to_parts): Likewise.
11712 (ix86_vectorize_builtin_gather): Likewise.
11713 (ix86_vectorize_builtin_scatter): Likewise.
11714 (avx_vpermilp_parallel): Likewise.
11715 (inline_memory_move_cost): Likewise.
11716 (ix86_tieable_integer_mode_p): Likewise.
11717 (x86_maybe_negate_const_int): Likewise.
11718 (ix86_expand_vector_init_duplicate): Likewise.
11719 (ix86_expand_vector_init_one_nonzero): Likewise.
11720 (ix86_expand_vector_init_one_var): Likewise.
11721 (ix86_expand_vector_init_concat): Likewise.
11722 (ix86_expand_vector_init_interleave): Likewise.
11723 (ix86_expand_vector_init_general): Likewise.
11724 (ix86_expand_vector_set): Likewise.
11725 (ix86_expand_vector_extract): Likewise.
11726 (emit_reduc_half): Likewise.
11727 (ix86_emit_i387_round): Likewise.
11728 (ix86_mangle_type): Likewise.
11729 (ix86_expand_round_sse4): Likewise.
11730 (expand_vec_perm_blend): Likewise.
11731 (canonicalize_vector_int_perm): Likewise.
11732 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11733 (expand_vec_perm_1): Likewise.
11734 (expand_vec_perm_interleave3): Likewise.
11735 (expand_vec_perm_even_odd_pack): Likewise.
11736 (expand_vec_perm_even_odd_1): Likewise.
11737 (expand_vec_perm_broadcast_1): Likewise.
11738 (ix86_vectorize_vec_perm_const_ok): Likewise.
11739 (ix86_expand_vecop_qihi): Likewise.
11740 (ix86_expand_mul_widen_hilo): Likewise.
11741 (ix86_expand_sse2_abs): Likewise.
11742 (ix86_expand_pextr): Likewise.
11743 (ix86_expand_pinsr): Likewise.
11744 (ix86_preferred_simd_mode): Likewise.
11745 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11746 * config/i386/sse.md (*andnot<mode>3): Likewise.
11747 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11748 (*<code><mode>3): Likewise.
11749 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11750 (ia64_expand_atomic_op): Likewise.
11751 (ia64_arg_type): Likewise.
11752 (ia64_mode_to_int): Likewise.
11753 (ia64_scalar_mode_supported_p): Likewise.
11754 (ia64_vector_mode_supported_p): Likewise.
11755 (expand_vec_perm_broadcast): Likewise.
11756 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11757 (iq2000_function_arg_advance): Likewise.
11758 (iq2000_function_arg): Likewise.
11759 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11760 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11761 (m68k_libcall_value): Likewise.
11762 (m68k_function_value): Likewise.
11763 (sched_attr_op_type): Likewise.
11764 * config/mcore/mcore.c (mcore_output_move): Likewise.
11765 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11767 (microblaze_function_arg): Likewise.
11768 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11769 (mips_print_operand): Likewise.
11770 (mips_mode_ok_for_mov_fmt_p): Likewise.
11771 (mips_vector_mode_supported_p): Likewise.
11772 (mips_preferred_simd_mode): Likewise.
11773 (mips_expand_vpc_loongson_even_odd): Likewise.
11774 (mips_expand_vec_unpack): Likewise.
11775 (mips_expand_vi_broadcast): Likewise.
11776 (mips_expand_vector_init): Likewise.
11777 (mips_expand_vec_reduc): Likewise.
11778 (mips_expand_msa_cmp): Likewise.
11779 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11780 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11781 (cc_flags_for_mode): Likewise.
11782 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11783 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11784 (nds32_output_casesi_pc_relative): Likewise.
11785 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11786 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11787 (nvptx_gen_unpack): Likewise.
11788 (nvptx_gen_pack): Likewise.
11789 (nvptx_gen_shuffle): Likewise.
11790 (nvptx_gen_wcast): Likewise.
11791 (nvptx_preferred_simd_mode): Likewise.
11792 * config/pa/pa.c (pa_secondary_reload): Likewise.
11793 * config/pa/predicates.md (base14_operand): Likewise.
11794 * config/powerpcspe/powerpcspe-c.c
11795 (altivec_resolve_overloaded_builtin): Likewise.
11796 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11798 (rs6000_preferred_simd_mode): Likewise.
11799 (output_vec_const_move): Likewise.
11800 (rs6000_expand_vector_extract): Likewise.
11801 (rs6000_split_vec_extract_var): Likewise.
11802 (reg_offset_addressing_ok_p): Likewise.
11803 (rs6000_legitimate_offset_address_p): Likewise.
11804 (rs6000_legitimize_address): Likewise.
11805 (rs6000_emit_set_const): Likewise.
11806 (rs6000_const_vec): Likewise.
11807 (rs6000_emit_move): Likewise.
11808 (spe_build_register_parallel): Likewise.
11809 (rs6000_darwin64_record_arg_recurse): Likewise.
11810 (swap_selector_for_mode): Likewise.
11811 (spe_init_builtins): Likewise.
11812 (paired_init_builtins): Likewise.
11813 (altivec_init_builtins): Likewise.
11814 (do_load_for_compare): Likewise.
11815 (rs6000_generate_compare): Likewise.
11816 (rs6000_expand_float128_convert): Likewise.
11817 (emit_load_locked): Likewise.
11818 (emit_store_conditional): Likewise.
11819 (rs6000_output_function_epilogue): Likewise.
11820 (rs6000_handle_altivec_attribute): Likewise.
11821 (rs6000_function_value): Likewise.
11822 (emit_fusion_gpr_load): Likewise.
11823 (emit_fusion_p9_load): Likewise.
11824 (emit_fusion_p9_store): Likewise.
11825 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11826 (fusion_gpr_mem_load): Likewise.
11827 (fusion_addis_mem_combo_load): Likewise.
11828 (fusion_addis_mem_combo_store): Likewise.
11829 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11830 (fusion_gpr_mem_load): Likewise.
11831 (fusion_addis_mem_combo_load): Likewise.
11832 (fusion_addis_mem_combo_store): Likewise.
11833 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11835 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11836 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11837 (rs6000_preferred_simd_mode): Likewise.
11838 (output_vec_const_move): Likewise.
11839 (rs6000_expand_vector_extract): Likewise.
11840 (rs6000_split_vec_extract_var): Likewise.
11841 (reg_offset_addressing_ok_p): Likewise.
11842 (rs6000_legitimate_offset_address_p): Likewise.
11843 (rs6000_legitimize_address): Likewise.
11844 (rs6000_emit_set_const): Likewise.
11845 (rs6000_const_vec): Likewise.
11846 (rs6000_emit_move): Likewise.
11847 (rs6000_darwin64_record_arg_recurse): Likewise.
11848 (swap_selector_for_mode): Likewise.
11849 (paired_init_builtins): Likewise.
11850 (altivec_init_builtins): Likewise.
11851 (rs6000_expand_float128_convert): Likewise.
11852 (emit_load_locked): Likewise.
11853 (emit_store_conditional): Likewise.
11854 (rs6000_output_function_epilogue): Likewise.
11855 (rs6000_handle_altivec_attribute): Likewise.
11856 (rs6000_function_value): Likewise.
11857 (emit_fusion_gpr_load): Likewise.
11858 (emit_fusion_p9_load): Likewise.
11859 (emit_fusion_p9_store): Likewise.
11860 * config/rx/rx.c (rx_gen_move_template): Likewise.
11861 (flags_from_mode): Likewise.
11862 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11863 (s390_slb_comparison): Likewise.
11864 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11865 (s390_vector_mode_supported_p): Likewise.
11866 (s390_cc_modes_compatible): Likewise.
11867 (s390_match_ccmode_set): Likewise.
11868 (s390_canonicalize_comparison): Likewise.
11869 (s390_emit_compare_and_swap): Likewise.
11870 (s390_branch_condition_mask): Likewise.
11871 (s390_rtx_costs): Likewise.
11872 (s390_secondary_reload): Likewise.
11873 (__SECONDARY_RELOAD_CASE): Likewise.
11874 (s390_expand_cs): Likewise.
11875 (s390_preferred_simd_mode): Likewise.
11876 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11877 * config/sh/sh.c (sh_print_operand): Likewise.
11878 (dump_table): Likewise.
11879 (sh_secondary_reload): Likewise.
11880 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11881 * config/sh/sh.md (casesi_worker_1): Likewise.
11882 (casesi_worker_2): Likewise.
11883 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11884 (fcc_comparison_operator): Likewise.
11885 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11886 (emit_soft_tfmode_cvt): Likewise.
11887 (sparc_preferred_simd_mode): Likewise.
11888 (output_cbranch): Likewise.
11889 (sparc_print_operand): Likewise.
11890 (sparc_expand_vec_perm_bmask): Likewise.
11891 (vector_init_bshuffle): Likewise.
11892 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11893 (spu_vector_mode_supported_p): Likewise.
11894 (spu_expand_insv): Likewise.
11895 (spu_emit_branch_or_set): Likewise.
11896 (spu_handle_vector_attribute): Likewise.
11897 (spu_builtin_splats): Likewise.
11898 (spu_builtin_extract): Likewise.
11899 (spu_builtin_promote): Likewise.
11900 (spu_expand_sign_extend): Likewise.
11901 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11902 (tilegx_simd_int): Likewise.
11903 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11904 (tilepro_simd_int): Likewise.
11905 * config/v850/v850.c (const_double_split): Likewise.
11906 (v850_print_operand): Likewise.
11907 (ep_memory_offset): Likewise.
11908 * config/vax/vax.c (vax_rtx_costs): Likewise.
11909 (vax_output_int_move): Likewise.
11910 (vax_output_int_add): Likewise.
11911 (vax_output_int_subtract): Likewise.
11912 * config/visium/predicates.md (visium_branch_operator): Likewise.
11913 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
11914 (visium_print_operand_address): Likewise.
11915 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11916 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
11917 (xtensa_expand_conditional_branch): Likewise.
11918 (xtensa_copy_incoming_a7): Likewise.
11919 (xtensa_output_literal): Likewise.
11920 * dfp.c (decimal_real_maxval): Likewise.
11921 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11923 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11924 Alan Hayward <alan.hayward@arm.com>
11925 David Sherwood <david.sherwood@arm.com>
11927 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
11928 (mode_nunits_inline): Likewise.
11929 (mode_inner_inline): Likewise.
11930 (mode_unit_size_inline): Likewise.
11931 (mode_unit_precision_inline): Likewise.
11932 (emit_insn_modes_h): Likewise. Also emit a #define of the
11934 (emit_mode_wider): Add an E_ prefix to mode names.
11935 (emit_mode_complex): Likewise.
11936 (emit_mode_inner): Likewise.
11937 (emit_mode_adjustments): Likewise.
11938 (emit_mode_int_n): Likewise.
11939 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
11940 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
11941 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
11942 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
11943 (CRC32_BUILTIN, ENTRY): Likewise.
11944 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
11945 (aarch64_pop_regs): Likewise.
11946 (aarch64_process_components): Likewise.
11947 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
11948 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
11949 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
11950 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
11951 * config/arm/arm.c (arm_init_libfuncs): Likewise.
11952 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
11954 * config/i386/i386-builtin.def (pcmpestr): Likewise.
11955 (pcmpistr): Likewise.
11956 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
11957 * config/mmix/mmix.c (mmix_output_condition): Likewise.
11958 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
11960 * config/rl78/rl78.c (mduc_regs): Likewise.
11961 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11962 (htm_expand_builtin): Likewise.
11963 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
11964 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
11965 * config/xtensa/xtensa.c (print_operand): Likewise.
11966 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
11967 (NUM_MODE_VECTOR_INT): Likewise.
11968 * genoutput.c (null_operand): Likewise.
11969 (output_operand_data): Likewise.
11970 * genrecog.c (print_parameter_value): Likewise.
11971 * lra.c (debug_operand_data): Likewise.
11973 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11975 * dwarf2out.c (scompare_loc_descriptor_wide)
11976 (scompare_loc_descriptor_narrow): New functions, split out from...
11977 (scompare_loc_descriptor): ...here.
11978 * expmed.c (emit_store_flag_int): New function, split out from...
11979 (emit_store_flag): ...here.
11981 2017-08-30 Richard Biener <rguenther@suse.de>
11983 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
11984 (dwarf2out_early_finish): Move setting of AT_pubnames from
11985 early debug output to early finish.
11987 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11989 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
11990 and -mdata-region to the assembler.
11992 2017-08-30 Richard Biener <rguenther@suse.de>
11994 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
11996 (gen_subprogram_die): Add DW_AT_object_pointer only early.
11997 (dwarf2out_early_global_decl): Only generate a DIE for the
11998 abstract origin if it doesn't already exist or is a declaration DIE.
11999 (resolve_addr): Do not add the linkage name twice when
12000 generating a stub DIE for the DW_TAG_GNU_call_site target.
12002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12004 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12005 Use machine_mode rather than int for arg1_mode.
12007 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12010 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12011 that the second argument of the built-in functions to unpack
12012 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12013 switch statement instead a lot of if statements.
12014 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12015 Allow 64-bit values to be in Altivec registers as well as
12016 traditional floating point registers.
12017 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12019 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12021 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12022 MAX_REGS_PER_ADDRESS == 1.
12024 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12026 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12027 * config/i386/i386.c (ix86_option_override_internal): Simplify
12028 setting of opts->x_flag_entry.
12030 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12031 Jakub Jelinek <jakub@redhat.com>
12032 Richard Biener <rguenther@suse.de>
12034 PR tree-optimization/81503
12035 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12036 folded constant fits in the target type; reorder tests for clarity.
12038 2017-08-29 Martin Liska <mliska@suse.cz>
12040 * passes.def: Include pass_lower_switch.
12041 * stmt.c (dump_case_nodes): Remove and move to
12042 tree-switch-conversion.
12043 (case_values_threshold): Likewise.
12044 (expand_switch_as_decision_tree_p): Likewise.
12045 (emit_case_decision_tree): Likewise.
12046 (expand_case): Likewise.
12047 (balance_case_nodes): Likewise.
12048 (node_has_low_bound): Likewise.
12049 (node_has_high_bound): Likewise.
12050 (node_is_bounded): Likewise.
12051 (emit_case_nodes): Likewise.
12052 (struct simple_case_node): New struct.
12053 (add_case_node): Remove.
12054 (emit_case_dispatch_table): Use vector instead of case_list.
12055 (reset_out_edges_aux): Remove.
12056 (compute_cases_per_edge): Likewise.
12057 (expand_case): Build list of simple_case_node.
12058 (expand_sjlj_dispatch_table): Use it.
12059 * tree-switch-conversion.c (struct case_node): Moved from
12060 stmt.c and adjusted.
12061 (emit_case_nodes): Likewise.
12062 (node_has_low_bound): Likewise.
12063 (node_has_high_bound): Likewise.
12064 (node_is_bounded): Likewise.
12065 (case_values_threshold): Likewise.
12066 (reset_out_edges_aux): Likewise.
12067 (compute_cases_per_edge): Likewise.
12068 (add_case_node): Likewise.
12069 (dump_case_nodes): Likewise.
12070 (balance_case_nodes): Likewise.
12071 (expand_switch_as_decision_tree_p): Likewise.
12072 (emit_jump): Likewise.
12073 (emit_case_decision_tree): Likewise.
12074 (try_switch_expansion): Likewise.
12075 (do_jump_if_equal): Likewise.
12076 (emit_cmp_and_jump_insns): Likewise.
12077 (fix_phi_operands_for_edge): New function.
12078 (record_phi_operand_mapping): Likewise.
12079 (class pass_lower_switch): New pass.
12080 (pass_lower_switch::execute): New function.
12081 (make_pass_lower_switch): Likewise.
12082 (conditional_probability):
12083 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12084 * tree-pass.h: Add make_pass_lower_switch.
12086 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12089 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12092 2017-08-29 Richard Biener <rguenther@suse.de>
12094 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12095 we do not add a DW_AT_inline attribute twice.
12096 (gen_subprogram_die): Remove code setting DW_AT_inline on
12097 DECL_ABSTRACT_P nodes.
12099 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
12101 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
12102 calls to internal functions.
12103 (gimplify_modify_expr): Likewise.
12104 * tree-call-cdce.c (use_internal_fn): Likewise.
12105 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
12106 (convert_to_divmod): Set the nothrow flag.
12107 * tree-if-conv.c (predicate_mem_writes): Likewise.
12108 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12109 (vectorizable_call): Likewise.
12110 (vectorizable_store): Likewise.
12111 (vectorizable_load): Likewise.
12112 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
12113 (vect_recog_mask_conversion_pattern): Likewise.
12115 2017-08-29 Martin Liska <mliska@suse.cz>
12118 * gcc.c (driver_handle_option): Add new argument.
12119 * opts-common.c (handle_option): Pass
12120 target_option_override_hook.
12121 * opts-global.c (lang_handle_option): Add new option.
12122 (set_default_handlers): Add new argument.
12123 (decode_options): Likewise.
12124 * opts.c (target_handle_option): Likewise.
12125 (common_handle_option): Call target_option_override_hook.
12126 * opts.h (struct cl_option_handler_func): Add hook for
12127 target option override.
12128 (struct cl_option_handlers): Likewise.
12129 (set_default_handlers): Add new argument.
12130 (decode_options): Likewise.
12131 (common_handle_option): Likewise.
12132 (target_handle_option): Likewise.
12133 * toplev.c (toplev::main): Pass targetm.target_option.override
12136 2017-08-29 Richard Biener <rguenther@suse.de>
12137 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
12139 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
12140 life to the active subtree.
12142 2017-08-28 Jeff Law <law@redhat.com>
12144 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
12145 derive_equivalences.
12146 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
12147 Code moved into....
12148 (edge_info::derive_equivalences): New private member function
12150 * tree-ssa-dom.c (class edge_info): Changed from a struct
12151 to a class. Add ctor/dtor, methods and data members.
12152 (edge_info::edge_info): Renamed from allocate_edge_info.
12153 Initialize additional members.
12154 (edge_info::~edge_info): New.
12155 (free_dom_edge_info): Delete the edge info.
12156 (record_edge_info): Use new class & associated member functions.
12157 Tighten forms for testing for edge equivalences.
12158 (record_temporary_equivalences): Iterate over the simple
12159 equivalences rather than assuming there's only one per edge.
12160 (cprop_into_successor_phis): Iterate over the simple
12161 equivalences rather than assuming there's only one per edge.
12162 (optimize_stmt): Use operand_equal_p rather than pointer
12163 equality for mini-DSE code.
12165 2017-08-28 Nathan Sidwell <nathan@acm.org>
12167 * gcc.c (execute): Fold SIGPIPE handling into switch
12168 statement. Adjust internal error message.
12170 2017-08-28 Richard Biener <rguenther@suse.de>
12173 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12174 Do nothing for removed DIEs.
12176 2017-08-28 Richard Biener <rguenther@suse.de>
12178 PR tree-optimization/81977
12179 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12182 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12185 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12186 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12187 using targetm.gen_mem_thread_fence.
12189 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12192 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12193 predicate to register_operand. Reorder operands.
12194 (*btr<mode>): Ditto.
12195 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12196 (*btr<mode>_mask): Ditto.
12198 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12200 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12201 * config/rs6000/xmmintrin.h: New file.
12202 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12204 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12207 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12208 parameter and_insn and return it.
12209 (recombine_lvx_pattern): Insert a copy to ensure availability of
12210 the base register of the copied masking operation at the point of
12211 the instruction replacement.
12212 (recombine_stvx_pattern): Likewise.
12214 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12216 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12217 undocumented switches.
12218 (-mpower9-dform-vector): Likewise.
12219 (-mpower9-dform): Likewise.
12220 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12221 comments to delete references to -mpower9-dform* switches.
12222 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12223 Delete reference to -mpower9-dform* switches, test for
12224 -mpower9-vector instead.
12225 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12226 (OTHER_P9_VECTOR_MASKS): Likewise.
12227 (POWERPC_MASKS): Likewise.
12228 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12229 tests against -mpower9-dform* to -mpower9-vector. Delete code
12230 that checked for -mpower9-dform* consistancy with other options.
12231 Add test for -mpower9-misc to enable other power9 switches.
12232 (rs6000_init_hard_regno_mode_ok): Likewise.
12233 (rs6000_option_override_internal): Likewise.
12234 (rs6000_emit_prologue): Likewise.
12235 (rs6000_emit_epilogue): Likewise.
12236 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12237 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12238 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12240 (emit_fusion_p9_store): Likewise.
12241 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12242 resetting these macros if the assembler does not support ISA 3.0
12244 (TARGET_P9_DFORM_VECTOR): Likewise.
12245 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12246 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12248 2017-08-25 Alan Modra <amodra@gmail.com>
12250 PR rtl-optimization/81747
12251 * cse.c (cse_extended_basic_block): Don't attempt to record
12252 equivalences for degenerate conditional branches that branch
12253 to their fall-through.
12255 2017-08-24 Martin Sebor <msebor@redhat.com>
12257 PR middle-end/81908
12258 * gimple-fold.c (size_must_be_zero_p): New function.
12259 (gimple_fold_builtin_memory_op): Call it.
12261 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12263 * config/rs6000/mm_malloc.h: New file.
12265 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12267 PR tree-optimization/81913
12268 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12269 analysis when either IVs in condition can wrap.
12271 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12273 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12274 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12276 2017-08-24 Richard Biener <rguenther@suse.de>
12279 * targhooks.c (default_target_can_inline_p): Properly
12280 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12281 is present and always compare.
12282 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12283 infer -mfpmath=sse from TARGET_SSE_P.
12284 (ix86_can_inline_p): Properly use target_option_default_node when
12285 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12287 2017-08-24 Richard Biener <rguenther@suse.de>
12290 * dwarf2out.c (output_die): Handle flag_generate_offload like
12292 (output_comp_unit): Likewise.
12293 (gen_array_type_die): Likewise.
12294 (dwarf2out_early_finish): Likewise.
12295 (note_variable_value_in_expr): Likewise.
12296 (dwarf2out_finish): Likewise. Adjust assert.
12297 * cgraphunit.c (symbol_table::compile): Move setting of
12298 flag_generate_offload earlier ...
12299 (symbol_table::finalize_compilation_unit): ... here, before
12300 early debug finalization.
12302 2017-08-24 Richard Biener <rguenther@suse.de>
12304 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12305 and ipa-fnsummary.h.
12306 (ix86_can_inline_p): When ix86_fpmath flags do not match
12307 check whether the callee uses FP math at all.
12309 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12311 PR middle-end/81931
12312 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12313 instead of TYPE_PRECISION.
12315 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12316 Alan Hayward <alan.hayward@arm.com>
12317 David Sherwood <david.sherwood@arm.com>
12319 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12321 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12323 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12325 (ix86_frame::stack_realign_allocate): New field.
12326 (struct machine_frame_state): Modify comments.
12327 (machine_frame_state::sp_realigned_fp_end): New field.
12328 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12329 layout calculation.
12330 (sp_valid_at): Add assertion to assure no attempt to access invalid
12331 offset of a realigned stack.
12332 (fp_valid_at): Likewise.
12333 (choose_baseaddr): Modify comments.
12334 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12335 ix86_expand_prologue.
12336 (ix86_expand_prologue): Modify stack realignment and allocation.
12337 (ix86_expand_epilogue): Modify comments.
12338 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12339 avx2_runtime, avx512f, and avx512f_runtime.
12341 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12343 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12344 (mstackrealign): Do not init to -1.
12345 * config/i386/i386.c (ix86_option_override_internal):
12346 Check opts_set, not opts when setting default value of
12347 opts->x_ix86_force_align_arg_pointer.
12349 2017-08-23 Richard Biener <rguenther@suse.de>
12351 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12352 lang_hooks.decl_printable_name.
12353 * print-rtl-function.c (print_rtx_function): Likewise.
12354 * tree-pretty-print.c (dump_function_header): Likewise.
12356 2017-08-23 Richard Biener <rguenther@suse.de>
12359 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12360 -g0 at compile-time.
12362 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12364 PR middle-end/19706
12365 * doc/sourcebuild.texi (Other hardware attributes):
12368 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12370 PR middle-end/19706
12371 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12372 Fix single-use check.
12374 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12376 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12378 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12380 * doc/install.texi: Modify to add more details on running selected
12383 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12385 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12386 is combined with -mabi=ms.
12387 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12390 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12392 PR tree-optimization/81488
12393 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12394 and cached_basis fields.
12395 (MAX_SPREAD): New constant.
12396 (alloc_cand_and_find_basis): Initialize new fields.
12397 (clear_visited): New function.
12398 (create_phi_basis_1): Rename from create_phi_basis, set visited
12399 and cached_basis fields.
12400 (create_phi_basis): New wrapper function.
12401 (phi_add_costs_1): Rename from phi_add_costs, add spread
12402 parameter, set visited field, short-circuit when limits reached.
12403 (phi_add_costs): New wrapper function.
12404 (record_phi_increments_1): Rename from record_phi_increments, set
12406 (record_phi_increments): New wrapper function.
12407 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12408 (phi_incr_cost): New wrapper function.
12409 (all_phi_incrs_profitable_1): Rename from
12410 all_phi_incrs_profitable, set visited field.
12411 (all_phi_incrs_profitable): New wrapper function.
12413 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12414 Alan Hayward <alan.hayward@arm.com>
12415 David Sherwood <david.sherwood@arm.com>
12417 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12418 that takes the outer and inner modes.
12419 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12420 comparison as the canonical test for a paradoxical subreg.
12421 * combine.c (simplify_set): Use paradoxical_subreg_p.
12422 (make_extraction): Likewise.
12423 (force_to_mode): Likewise.
12424 (rtx_equal_for_field_assignment_p): Likewise.
12425 (gen_lowpart_for_combine): Likewise.
12426 (simplify_comparison): Likewise.
12427 * cse.c (equiv_constant): Likewise.
12428 * expmed.c (store_bit_field_1): Likewise.
12429 * final.c (alter_subreg): Likewise.
12430 * fwprop.c (propagate_rtx): Likewise.
12431 (forward_propagate_subreg): Likewise.
12432 * ira-conflicts.c (ira_build_conflicts): Likewise.
12433 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12434 * lra-constraints.c (curr_insn_transform): Likewise.
12435 (split_reg): Likewise.
12436 * lra-eliminations.c (move_plus_up): Likewise.
12437 (lra_eliminate_regs_1): Likewise.
12438 * recog.c (general_operand): Likewise.
12439 * ree.c (combine_reaching_defs): Likewise.
12440 * reload.c (push_reload): Likewise.
12441 (find_reloads): Likewise.
12442 * reload1.c (elimination_effects): Likewise.
12443 (compute_reload_subreg_offset): Likewise.
12444 (choose_reload_regs): Likewise.
12445 * rtlanal.c (subreg_lsb_1): Likewise.
12446 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12447 (simplify_subreg): Likewise.
12448 * var-tracking.c (track_loc_p): Likewise.
12449 * emit-rtl.c (byte_lowpart_offset): Likewise.
12450 (paradoxical_subreg_p): Delete out-of-line definition.
12452 2017-08-22 Jeff Law <law@redhat.com>
12454 PR tree-optimization/81741
12455 PR tree-optimization/71947
12456 * tree-ssa-dom.c: Include tree-inline.h.
12457 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12458 equivalences if one is more expensive to compute than the other.
12459 * tree-ssa-scopedtables.h (class const_or_copies): Make
12460 record_const_or_copy_raw method private.
12461 (class avail_exprs_stack): New method simplify_binary_operation.
12462 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12463 avail_exprs_stack::simplify_binary_operation as needed.
12464 (avail_exprs_stack::simplify_binary_operation): New function.
12466 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12468 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12469 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12470 (DOT_SYMBOLS): Likewise.
12471 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12472 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12473 (RS6000_ABI_NAME): Likewise.
12474 (TARGET_CMODEL): Likewise.
12475 (TOC_SECTION_ASM_OP): Likewise.
12476 (SET_CMODEL): New macro.
12477 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12479 2017-08-22 Richard Biener <rguenther@suse.de>
12481 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12482 to allow for free-lang-data replacements similar to verify_type_variant.
12484 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12485 Alan Hayward <alan.hayward@arm.com>
12486 David Sherwood <david.sherwood@arm.com>
12488 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12489 VOIDmode for the LABEL_REF.
12491 2017-08-22 Richard Biener <rguenther@suse.de>
12493 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12495 2017-08-22 Richard Biener <rguenther@suse.de>
12497 * common.opt (feliminate-dwarf2-dups): Ignore.
12498 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12499 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12500 same_die_p_wrap, compute_section_prefix,
12501 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12502 (comdat_symbol_id, comdat_symbol_number): Likewise.
12503 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12505 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12506 (output_die): Mark unreachable path unreachable.
12507 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12508 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12509 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12510 (dwarf2out_early_finish): Likewise.
12512 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12514 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12516 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12519 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12520 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12521 (avr_attribute_table) <io, io_low, address>: Initialize
12522 .decl_required with true.
12524 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12526 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12527 undocumented debugging options.
12528 (-mvsx-scalar-double): Likewise.
12529 (-mallow-df-permute): Likewise.
12530 (-mvectorize-builtins): Likewise.
12531 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12532 (rs6000_builtin_vectorized_function): Likewise.
12533 (rs6000_builtin_md_vectorized_function): Likewise.
12534 (rs6000_opt_vars): Likewise.
12536 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12539 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12540 (*btrq_imm): Rename from *btrq.
12541 (*btcq_imm): Rename from *btcq.
12542 (btsc): New code attribute.
12543 (*<btsc><mode>): New insn pattern.
12544 (*btr<mode>): Ditto.
12545 (*<btsc><mode>_mask): New insn_and_split pattern.
12546 (*btr<mode>_mask): Ditto.
12548 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12549 Alan Hayward <alan.hayward@arm.com>
12550 David Sherwood <david.sherwood@arm.com>
12552 * function.c (pad_below): Simplify padding calculation.
12554 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12555 Alan Hayward <alan.hayward@arm.com>
12556 David Sherwood <david.sherwood@arm.com>
12558 * target.def (function_prologue): Remove frame size argument.
12559 (function_epilogue): Likewise.
12560 * doc/tm.texi: Regenerate.
12561 * final.c (final_start_function): Update call to function_prologue.
12562 (final_end_function): Update call to function_epilogue.
12563 (default_function_pro_epilogue): Remove frame size argument.
12564 * output.h (default_function_pro_epilogue): Likewise.
12565 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12566 (arm_output_function_prologue): Likewise.
12567 * config/frv/frv.c (frv_function_prologue): Likewise.
12568 (frv_function_epilogue): Likewise.
12569 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12570 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12571 (ia64_output_function_epilogue): Likewise.
12572 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12573 (m32r_output_function_epilogue): Likewise.
12574 * config/microblaze/microblaze.c (microblaze_function_prologue)
12575 (microblaze_function_epilogue): Likewise.
12576 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12577 (mips_output_function_epilogue): Likewise.
12578 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12579 (mmix_target_asm_function_epilogue): Likewise.
12580 * config/msp430/msp430.c (msp430_start_function): Likewise.
12581 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12582 (nds32_asm_function_epilogue): Likewise.
12583 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12584 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12585 (pa_output_function_epilogue): Likewise.
12586 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12587 (rs6000_output_function_epilogue): Likewise.
12588 * config/rl78/rl78.c (rl78_start_function): Likewise.
12589 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12590 (rs6000_output_function_epilogue): Likewise.
12591 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12592 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12593 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12594 (sparc_asm_function_epilogue): Likewise.
12596 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12598 * tree.h (type_has_mode_precision_p): New function.
12599 * convert.c (convert_to_integer_1): Use it.
12600 * expr.c (expand_expr_real_2): Likewise.
12601 (expand_expr_real_1): Likewise.
12602 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
12603 * match.pd: Likewise.
12604 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
12605 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
12606 * tree-tailcall.c (process_assignment): Likewise.
12607 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12608 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
12609 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
12610 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12611 (vectorizable_assignment): Likewise.
12612 (vectorizable_shift): Likewise.
12613 (vectorizable_operation): Likewise.
12614 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12616 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12618 * match.pd: Add pow (C, x) simplification.
12620 2017-08-21 Richard Biener <rguenther@suse.de>
12622 PR tree-optimization/81900
12623 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12624 for blocks with abnormal predecessors.
12625 (compute_antic): Do not set visited flag prematurely.
12627 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12630 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12632 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12634 * stor-layout.h (vector_type_mode): Move to...
12635 * tree.h (vector_type_mode): ...here.
12636 * stor-layout.c (vector_type_mode): Move to...
12637 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12639 2017-08-21 Richard Biener <rguenther@suse.de>
12641 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12642 register_external_die hooks.
12643 (debug_false_tree_charstarstar_uhwistar): Declare.
12644 (debug_nothing_tree_charstar_uhwi): Likewise.
12645 * debug.c (do_nothing_debug_hooks): Adjust.
12646 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12647 (debug_nothing_tree_charstar_uhwi): Likewise.
12648 * dbxout.c (dbx_debug_hooks): Adjust.
12649 (xcoff_debug_hooks): Likewise.
12650 * sdbout.c (sdb_debug_hooks): Likewise.
12651 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12652 * dwarf2out.c (macinfo_label_base): New global.
12653 (dwarf2out_register_external_die): New function for the
12654 register_external_die hook.
12655 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12656 (dwarf2_debug_hooks): Use them.
12657 (dwarf2_lineno_debug_hooks): Adjust.
12658 (struct die_struct): Add with_offset flag.
12659 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12660 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12661 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12662 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12663 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12664 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12665 defining section names for the early LTO debug variants.
12666 (reset_indirect_string): New helper.
12667 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12668 (print_dw_val): Add support for offsetted symbol references.
12669 (get_ultimate_context): Split out from is_cxx.
12670 (is_cxx): Use get_ultimate_context.
12671 (is_fortran): Add decl overload.
12672 (compute_comp_unit_symbol): Split out worker from
12673 compute_section_prefix.
12674 (compute_section_prefix): Call compute_comp_unit_symbol and
12675 set comdat_type_p here.
12676 (output_die): Skip DIE symbol output for the LTO added one.
12677 Handle DIE symbol references with offset.
12678 (output_comp_unit): Guard section name mangling properly.
12679 For LTO debug sections emit a symbol at the section beginning
12680 which we use to refer to its DIEs.
12681 (add_abstract_origin_attribute): For DIEs registered via
12682 dwarf2out_register_external_die directly refer to the early
12683 DIE rather than indirectly through the shadow one we created.
12684 Remove obsolete call to dwarf2out_abstract_function for
12685 non-function/block origins.
12686 (gen_array_type_die): When generating early LTO debug do
12687 not emit DW_AT_string_length.
12688 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12689 late when in LTO. As suggested place a gcc_unreachable for
12690 the DECL_ABSTRACT_P case.
12691 (gen_subprogram_die): Avoid another specification DIE
12692 for early built declarations/definitions for the late LTO case.
12693 (gen_variable_die): Add type references for late duplicated VLA dies
12695 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12696 we have the abstract instance already.
12697 (process_scope_var): Adjust decl DIE contexts in LTO which
12698 first puts them in limbo.
12699 (gen_decl_die): Do not generate type DIEs late apart from
12700 types for VLAs or for decls we do not yet have a DIE. Do not
12701 call dwarf2out_abstract_function late.
12702 (dwarf2out_early_global_decl): Make sure to create DIEs
12703 for abstract instances of a decl first.
12704 (dwarf2out_late_global_decl): Adjust comment.
12705 (output_macinfo_op): With multiple macro sections use
12706 macinfo_label_base to distinguish labels.
12707 (output_macinfo): Likewise. Update macinfo_label_base.
12708 Pass in the line info label.
12709 (note_variable_value_in_expr): When generating LTO resolve
12710 all variable values here by generating DIEs as needed.
12711 (init_sections_and_labels): Add early LTO debug flag parameter
12712 and generate different sections and names if set. Add generation
12713 counter for the labels so we can have multiple of them.
12714 (reset_dies): Helper to allow DIEs to be output multiple times.
12715 (dwarf2out_finish): When outputting DIEs to the fat part of an
12716 LTO object first reset DIEs.
12717 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12718 (modified_type_die): Check for decl_ultimate_origin being self
12720 (gen_type_die_with_usage): Likewise.
12721 (gen_typedef_die): Allow decl_ultimate_origin being self.
12722 (set_decl_abstract_flags): Remove.
12723 (set_block_abstract_flags): Likewise.
12724 (dwarf2out_abstract_function): Treat the early generated DIEs
12725 as the abstract copy and only add DW_AT_inline and
12726 DW_AT_artificial here and call set_decl_origin_self.
12727 If the DIE has an abstract origin don't do anything.
12728 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12729 if we have none yet (Go fails to build one, PR78628).
12730 (variably_modified_type_p): Prevent endless recursion for Ada
12731 cyclic pointer types.
12732 * lto-streamer-in.c: Include debug.h.
12733 (dref_queue): New global.
12734 (lto_read_tree_1): Stream in DIE references.
12735 (lto_input_tree): Register DIE references.
12736 (input_function): Stream DECL_DEBUG_ARGS.
12737 * lto-streamer-out.c: Include debug.h.
12738 (lto_write_tree_1): Output DIE references.
12739 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12740 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12741 (output_function): Stream DECL_DEBUG_ARGS.
12742 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12743 Stream DECL_ABSTRACT_ORIGIN.
12744 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12745 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12746 DECL_CONTEXT for file-scope decls.
12747 * lto-streamer.h (struct dref_entry): Declare.
12748 (dref_queue): Likewise.
12749 * cfgexpand.c (pass_expand::execute): Do not call the
12750 outlining_inline_function hook here.
12751 * lto-wrapper.c (debug_obj): New global.
12752 (tool_cleanup): Unlink it if required.
12753 (debug_objcopy): New function.
12754 (run_gcc): Handle early debug sections in the IL files by
12755 extracting them to separate files, partially linkin them and
12756 feeding the result back as result to the linker.
12757 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12758 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12759 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12760 sections into a separate segment.
12761 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12763 (darwin_asm_dwarf_section): Likewise.
12764 (darwin_asm_output_dwarf_offset): Likewise.
12765 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12767 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12768 Alan Hayward <alan.hayward@arm.com>
12769 David Sherwood <david.sherwood@arm.com>
12771 * read-md.h (md_reader::record_potential_iterator_use): Replace
12772 pointer argument with an rtx and an index.
12773 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12774 (apply_mode_iterator): Likewise.
12775 (apply_code_iterator): Likewise.
12776 (apply_int_iterator): Likewise.
12777 (apply_subst_iterator): Likewise.
12778 (record_iterator_use): Likewise.
12779 (record_attribute_use): Likewise.
12780 (md_reader::record_potential_iterator_use): Likewise. Update calls
12781 to record_iterator_use and apply_iterator.
12782 (iterator_use): Replace ptr with x and index.
12783 (attribute_use): Likewise.
12784 (apply_attribute_uses): Update calls to apply_iterator.
12785 (apply_iterators): Likewise. Update initialization of iterator_use.
12786 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12787 and record_potential_iterator_use.
12788 (rtx_reader::read_rtx_operand): Likewise.
12790 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12791 Alan Hayward <alan.hayward@arm.com>
12792 David Sherwood <david.sherwood@arm.com>
12794 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12797 2017-08-21 Richard Biener <rguenther@suse.de>
12799 PR middle-end/81884
12800 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12801 at struct end conservatively when comparing common bases.
12803 2017-08-21 Richard Biener <rguenther@suse.de>
12805 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12806 (mem_ref_in_stmt): Remove.
12807 (determine_max_movement): Use ref index to get at the reference.
12808 (invariantness_dom_walker::before_dom_children): Deal with
12809 lim data already initialized.
12810 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12812 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12814 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12815 (TARGET_ISA_ROUND): Ditto.
12816 (TARGET_ROUND): Ditto.
12817 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12818 * config/i386/i386.md: Ditto.
12819 * config/i386/sse.md: Ditto.
12820 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12821 with OPTION_MASK_ISA_SSE4_1.
12823 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12826 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12827 __builtin_ia32_lzcnt_u16.
12829 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12832 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12833 (rs6000_set_current_function): Rewrite function to use it.
12835 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12838 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12839 and TYPE_WARN_IF_NOT_ALIGN.
12840 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12841 (handle_warn_if_not_align): New.
12842 (place_union_field): Call handle_warn_if_not_align.
12843 (place_field): Call handle_warn_if_not_align.
12844 Copy TYPE_WARN_IF_NOT_ALIGN.
12845 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12846 (layout_type): Likewise.
12847 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12849 (tree_decl_common): Add warn_if_not_align.
12850 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12851 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12852 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12853 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12854 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12855 * doc/extend.texi: Document warn_if_not_aligned attribute.
12856 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12858 2017-08-17 Martin Liska <mliska@suse.cz>
12861 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12862 (get_data_dependence): Use it as pointer type.
12863 (distribute_loop): Likewise.
12865 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12867 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12868 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12869 (p8_vmrgow_<mode>_direct): New define_insn.
12870 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12871 handle endianness for vmrgew and vmrgow permute patterns.
12873 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12875 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12876 * config/rs6000/rs6000-cpus.def: Remove comment.
12877 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12878 (POWERPC_MASKS): Likewise.
12879 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12880 use of TARGET_VSX_TIMODE.
12881 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12882 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12883 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12884 (rs6000_option_override_internal): Remove dead code.
12885 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12886 (rs6000_legitimize_reload_address): Likewise.
12887 (rs6000_legitimate_address_p): Likewise.
12888 (rs6000_opt_masks): Delete "vsx-timode".
12889 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12890 from function comment.
12891 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12892 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12893 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12895 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12896 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12897 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12898 (VSX_TI): Likewise.
12900 (define_peephole2): Likewise.
12902 2017-08-17 Martin Sebor <msebor@redhat.com>
12905 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12906 past the end of an array.
12907 (test_pp_format): Add test cases.
12909 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
12911 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
12912 missing ECF_NOTHROW flags.
12914 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12917 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
12918 operands residing in integer registers.
12919 (*vsx_le_perm_load_<mode>): Likewise.
12920 (*vsx_le_perm_store_<mode>): Likewise.
12921 (define_peephole2): Add peepholes to optimize the above.
12923 2017-08-17 Marek Polacek <polacek@redhat.com>
12925 PR middle-end/81814
12926 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
12927 to mimic what shorten_compare did. Change the return type to bool.
12928 (fold_cond_expr_with_comparison): Update call to
12929 operand_equal_for_comparison_p.
12930 (fold_ternary_loc): Likewise.
12932 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
12934 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
12936 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
12938 2017-08-17 Richard Biener <rguenther@suse.de>
12940 * tree-ssa-structalias.c (solve_graph): When propagating
12941 to successors update the graphs succ edges and avoid duplicate work.
12943 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
12946 * config/i386/i386.c (ix86_option_override_internal): Save target
12947 specific options after ix86_stack_protector_guard_reg was changed.
12949 2017-08-17 Richard Biener <rguenther@suse.de>
12951 PR tree-optimization/81827
12952 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
12953 (new_var_info): Initialize it conservatively.
12954 (get_call_vi): Mark register vars.
12955 (new_scalar_tmp_constraint_exp): Likewise.
12956 (handle_rhs_call): Likewise.
12957 (handle_const_call): Likewise.
12958 (create_function_info_for): Likewise.
12959 (solve_constraints): Sort varinfos to separate register from
12960 non-register vars to pack points-to solution bitmaps during
12963 2017-08-17 Marek Polacek <polacek@redhat.com>
12965 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
12967 2017-08-17 Richard Biener <rguenther@suse.de>
12969 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
12970 to true when overflow is undefined and we saturated the result.
12972 2017-08-17 Alan Modra <amodra@gmail.com>
12975 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
12976 Don't use store multiple if only one reg needs saving.
12977 (interesting_frame_related_regno): New function.
12978 (rs6000_frame_related): Don't emit frame info for regs that
12980 (rs6000_emit_epilogue): Likewise.
12982 2017-08-16 Nathan Sidwell <nathan@acm.org>
12984 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
12985 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
12986 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
12987 * tree.c (free_lang_data_in_type): Use else-if chain. Always
12988 clear TYPE_LANG_1. Remove obsolete member-function stripping.
12989 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
12990 (verify_type): Adjust for TYPE_BINFO move.
12991 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
12992 process TYPE_BINFO directly.
12993 (hash_tree): Likewise.
12994 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
12996 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
12999 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13001 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13003 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13006 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13007 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13008 Add ix86_binary_operator_ok to insn constraint.
13009 (*iordi_1_bts): Ditto.
13010 (*xordi_1_btc): Ditto.
13011 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13012 Update corresponding peephole2 pattern.
13016 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13018 PR tree-optimization/81832
13019 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13020 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13022 2017-08-16 Marek Polacek <polacek@redhat.com>
13025 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13026 regarding size_zero_node.
13028 2017-08-16 Martin Liska <mliska@suse.cz>
13031 * config.gcc: Respect previously set extra_objs in case
13034 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13036 PR tree-optimization/81835
13037 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13038 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13039 not depend on the phi.
13041 2017-08-16 Alan Modra <amodra@gmail.com>
13043 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13046 2017-08-16 Alan Modra <amodra@gmail.com>
13048 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13049 (save_reg_p): ..into this. Update all callers.
13050 (first_reg_to_save): Simplify.
13052 2017-08-16 Alan Modra <amodra@gmail.com>
13054 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13057 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13061 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13062 constant count if that count is less than 32.
13064 2017-08-15 Nathan Sidwell <nathan@acm.org>
13066 * gcc.c (execute): Emit friendlier message if inferior is killed
13067 by an external cause.
13069 2017-08-15 Richard Biener <rguenther@suse.de>
13071 PR tree-optimization/81790
13072 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13073 CONSTRUCTORs from simplifying and VN.
13075 2017-08-14 Martin Sebor <msebor@redhat.com>
13077 * builtin-attrs.def: Add comments.
13079 2017-08-14 Martin Sebor <msebor@redhat.com>
13082 * doc/extend.texi (attribute nonstring): Document new attribute.
13084 2017-08-14 Martin Sebor <msebor@redhat.com>
13087 * tree-diagnostic.c (default_tree_printer): Handle %G.
13088 * gimple-pretty-print.h (percent_G_format): Declare new function.
13089 * gimple-pretty-print.c (percent_G_format): Define.
13090 * tree-pretty-print.c (percent_K_format): Add argument.
13092 2017-08-14 Martin Sebor <msebor@redhat.com>
13094 PR translation/79998
13095 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13096 Remove a stray space.
13098 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
13101 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
13102 (*iordi_1_bts): Ditto.
13103 (*xordi_1_btc): Ditto.
13105 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13108 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
13109 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13111 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
13112 quoted strings, and make more translator-friendly.
13113 (darwin_rs6000_override_options): Likewise.
13114 (rs6000_option_override_internal): Likewise.
13115 (rs6000_return_in_memory): Fix overlong line.
13116 (init_cmulative_args): Use quoted strings, and make more
13117 translator-friendly.
13118 (rs6000_pass_by_reference): Fix overlong line.
13119 (def_builtin): Use quoted strings.
13120 (altivec_expand_predicate_builtin): Use quoted strings, and make
13121 more translator-friendly.
13122 (htm_expand_builtin): Use quoted strings.
13123 (cpu_expand_builtin): Use quoted strings, and make more
13124 translator-friendly.
13125 (altivec_expand_builtin): Likewise.
13126 (paired_expand_predicate_builtin): Likewise.
13127 (rs6000_invalid_builtin): Likewise.
13128 (builtin_function_type): Use quoted strings.
13129 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
13130 more translator-friendly.
13131 (rs6000_trampoline_init): Likewise.
13132 (rs6000_handle_altivec_attribute): Likewise.
13133 (rs6000_inner_target_options): Use quoted strings.
13134 (rs6000_disable_incompatible_switches): Likewise.
13135 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
13136 strings, and make more translator-friendly.
13137 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
13139 2017-08-14 Bin Cheng <bin.cheng@arm.com>
13141 PR tree-optimization/81799
13142 * tree-loop-distribution.c (version_loop_by_alias_check): Force
13143 cond_expr to simple gimple operand.
13145 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
13147 PR middle-end/46932
13148 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
13150 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
13154 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
13155 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
13156 TARGET_GASISR_PROLOGUES.
13157 * config/avr/avr.c (avr_option_override): Same.
13158 (avr_pass_pre_proep::execute): Same.
13160 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13163 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13164 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13166 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13168 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13169 define_insn_and_split. Split to a memory load from 0 in
13170 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13171 using PTR mode iterator.
13172 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13173 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13174 (*add_tp_<mode>): Redefine as define_insn_and_split.
13175 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13176 address space. Merge with *add_tp_x32 using PTR mode iterator.
13177 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13178 Split to an add with a memory load from 0 in
13179 DEFAULT_TLS_SEG_REG address space.
13181 2017-08-12 Andrew Pinski <apinski@cavium.com>
13183 * config/aarch64/aarch64-option-extensions.def (rdma):
13184 Fix feature string to what Linux prints out in /proc/cpuinfo.
13186 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13189 * dwarf2out.c (modified_type_die): For C typedef types that have
13190 an ultimate origin, process the ultimate origin instead of the
13192 (gen_typedef_die): Assert that input DECLs have no ultimate
13194 (gen_type_die_with_usage): For typedef variants that have an
13195 ultimate origin, just call gen_decl_die on the original DECL.
13196 (process_scope_var): Avoid creating DIEs for local typedefs and
13197 concrete static variables.
13199 2017-08-12 Alan Modra <amodra@gmail.com>
13203 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13204 match gnu-user.h startfile.
13205 (ENDFILE_LINUX_SPEC): Similarly.
13207 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13210 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13212 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13214 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13215 * config/aarch64/aarch64.md (mov<mode>): Change.
13216 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13217 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13218 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13220 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13222 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13223 for storage order barriers.
13225 2017-08-11 Martin Liska <mliska@suse.cz>
13228 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13229 variables of void type.
13231 2017-08-11 Martin Liska <mliska@suse.cz>
13233 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13234 TARGET_SUPPORTS_ALIASES.
13235 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13236 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13237 (optimize_weakref): Likewise.
13238 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13239 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13240 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13242 2017-08-11 Martin Liska <mliska@suse.cz>
13245 * config/i386/i386.c (make_resolver_func): Do complete
13246 refactoring of the function.
13248 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13251 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13252 * config/i386/i386.c (ix86_stack_protect_guard): Use
13253 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13254 * doc/invoke.texi (x86 Options): Document
13255 -mstack-protector-guard-symbol= option.
13257 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13259 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13260 * config/i386/i386.c (ix86_split_stack_guard): New function.
13261 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13262 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13263 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13264 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13265 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13266 (split_stack_space_check): Call ix86_split_stack_guard.
13268 2017-08-10 Martin Sebor <msebor@redhat.com>
13270 * print-tree.c (print_node): Print location using the established
13272 Replace spaces with colons.
13273 (debug_raw, debug): Ditto.
13275 2017-08-10 Martin Sebor <msebor@redhat.com>
13278 * pretty-print.c (pp_format): Correct the handling of %s precision.
13280 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13283 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13285 (ix86_finalize_stack_frame_flags): This. Also clear
13286 frame_pointer_needed if -fno-omit-frame-pointer is used without
13288 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13289 with ix86_finalize_stack_frame_flags.
13290 (ix86_expand_epilogue): Likewise.
13291 (ix86_expand_split_stack_prologue): Likewise.
13292 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13294 2017-08-10 Martin Liska <mliska@suse.cz>
13297 * c-attribs.c (handle_target_attribute):
13298 Report warning for an empty string argument of target attribute.
13300 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13303 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13305 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13306 or DECL_NONLOCAL labels.
13307 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13308 or DECL_NONLOCAL labels here.
13310 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13312 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13313 to indicate when early gimple folding has been disabled.
13314 (rs6000_gimple_fold_builtin): Add debug content.
13315 (rs6000_invalid_builtin): Fix whitespace.
13316 (rs6000_expand_builtin): Fix whitespace.
13317 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13319 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13322 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13323 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13325 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13327 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13328 (qdf24xx): Likewise.
13329 * config/aarch64/aarch64.md: Include falkor.md.
13330 * config/aarch64/falkor.md: New.
13332 2017-08-09 Marek Polacek <polacek@redhat.com>
13335 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13336 * diagnostic.c (emit_diagnostic): Add a comment.
13337 (emit_diagnostic_valist): New function.
13339 2017-08-09 Marek Polacek <polacek@redhat.com>
13342 * input.c (make_location): New overload.
13343 * input.h (make_location): Declare.
13345 2017-08-08 Alan Modra <amodra@gmail.com>
13346 H.J. Lu <hongjiu.lu@intel.com>
13349 * gcc.c (NO_PIE_SPEC): Delete.
13350 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13352 (LINK_PIE_SPEC): ..to here.
13353 (LINK_COMMAND_SPEC): Support -no-pie.
13354 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13355 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13357 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13358 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13359 (ENDFILE_CRTEND_SPEC): Similarly.
13361 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13364 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13365 (mstack-protector-guard-offset=): Ditto.
13366 * config/i386/i386.c (ix86_option_override): Handle
13367 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13369 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13370 ix86_stack_protect_guard_offset variables.
13371 (TARGET_STACK_PROTECT_GUARD): Always define.
13372 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13373 and -mstack-protector-guard-offset= options.
13375 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13377 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13378 boundary case for the last candidate.
13380 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13382 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13383 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13385 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13387 PR middle-end/19706
13388 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13389 * config/aarch64/aarch64-builtins.c
13390 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13391 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13392 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13394 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13395 Andrew Pinski <pinskia@gmail.com>
13397 PR middle-end/19706
13398 * internal-fn.def (XORSIGN): New.
13399 * optabs.def (xorsign_optab): New.
13400 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13401 (convert_expand_mult_copysign): New.
13402 (pass_optimize_widening_mul::execute): Call
13403 convert_expand_mult_copysign.
13405 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13407 PR tree-optimization/81354
13408 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13409 Insert on edges rather than explicitly creating landing pads.
13410 (analyze_candidates_and_replace): Commit edge inserts.
13412 2017-08-08 Richard Biener <rguenther@suse.de>
13414 PR middle-end/81719
13415 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13416 (expand_simple_operations): Also look through ADDR_EXPRs with
13417 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13419 2017-08-08 Richard Biener <rguenther@suse.de>
13421 PR tree-optimization/81723
13422 * tree-vect-slp.c (struct bst_traits): New hash traits.
13423 (bst_fail): New global.
13424 (vect_build_slp_tree_2): New worker, split out from ...
13425 (vect_build_slp_tree): ... this now wrapping it with using
13426 bst_fail set to cache SLP tree build fails. Properly handle
13428 (vect_analyze_slp_instance): Allocate and free bst_fail.
13430 2017-08-08 Martin Liska <mliska@suse.cz>
13433 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13434 LABEL_DECLs that can be from a different function.
13436 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13438 PR tree-optimization/81744
13439 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13440 loop's number of iterations.
13442 2017-08-08 Martin Liska <mliska@suse.cz>
13444 * asan.c: Include header files.
13445 * attribs.c (build_decl_attribute_variant): New function moved
13447 (build_type_attribute_qual_variant): Likewise.
13448 (cmp_attrib_identifiers): Likewise.
13449 (simple_cst_list_equal): Likewise.
13450 (omp_declare_simd_clauses_equal): Likewise.
13451 (attribute_value_equal): Likewise.
13452 (comp_type_attributes): Likewise.
13453 (build_type_attribute_variant): Likewise.
13454 (lookup_ident_attribute): Likewise.
13455 (remove_attribute): Likewise.
13456 (merge_attributes): Likewise.
13457 (merge_type_attributes): Likewise.
13458 (merge_decl_attributes): Likewise.
13459 (merge_dllimport_decl_attributes): Likewise.
13460 (handle_dll_attribute): Likewise.
13461 (attribute_list_equal): Likewise.
13462 (attribute_list_contained): Likewise.
13463 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13464 (lookup_attribute_by_prefix): Likewise.
13465 * bb-reorder.c: Include header files.
13466 * builtins.c: Likewise.
13467 * calls.c: Likewise.
13468 * cfgexpand.c: Likewise.
13469 * cgraph.c: Likewise.
13470 * cgraphunit.c: Likewise.
13471 * convert.c: Likewise.
13472 * dwarf2out.c: Likewise.
13473 * final.c: Likewise.
13474 * fold-const.c: Likewise.
13475 * function.c: Likewise.
13476 * gimple-expr.c: Likewise.
13477 * gimple-fold.c: Likewise.
13478 * gimple-pretty-print.c: Likewise.
13479 * gimple.c: Likewise.
13480 * gimplify.c: Likewise.
13481 * hsa-common.c: Likewise.
13482 * hsa-gen.c: Likewise.
13483 * internal-fn.c: Likewise.
13484 * ipa-chkp.c: Likewise.
13485 * ipa-cp.c: Likewise.
13486 * ipa-devirt.c: Likewise.
13487 * ipa-fnsummary.c: Likewise.
13488 * ipa-inline.c: Likewise.
13489 * ipa-visibility.c: Likewise.
13491 * lto-cgraph.c: Likewise.
13492 * omp-expand.c: Likewise.
13493 * omp-general.c: Likewise.
13494 * omp-low.c: Likewise.
13495 * omp-offload.c: Likewise.
13496 * omp-simd-clone.c: Likewise.
13497 * opts-global.c: Likewise.
13498 * passes.c: Likewise.
13499 * predict.c: Likewise.
13500 * sancov.c: Likewise.
13501 * sanopt.c: Likewise.
13502 * symtab.c: Likewise.
13503 * toplev.c: Likewise.
13504 * trans-mem.c: Likewise.
13505 * tree-chkp.c: Likewise.
13506 * tree-eh.c: Likewise.
13507 * tree-into-ssa.c: Likewise.
13508 * tree-object-size.c: Likewise.
13509 * tree-parloops.c: Likewise.
13510 * tree-profile.c: Likewise.
13511 * tree-ssa-ccp.c: Likewise.
13512 * tree-ssa-live.c: Likewise.
13513 * tree-ssa-loop.c: Likewise.
13514 * tree-ssa-sccvn.c: Likewise.
13515 * tree-ssa-structalias.c: Likewise.
13516 * tree-ssa.c: Likewise.
13517 * tree-streamer-in.c: Likewise.
13518 * tree-vectorizer.c: Likewise.
13519 * tree-vrp.c: Likewise.
13520 * tsan.c: Likewise.
13521 * ubsan.c: Likewise.
13522 * varasm.c: Likewise.
13523 * varpool.c: Likewise.
13524 * tree.c: Remove functions moved to attribs.[ch].
13525 * tree.h: Likewise.
13526 * config/aarch64/aarch64.c: Add attrs.h header file.
13527 * config/alpha/alpha.c: Likewise.
13528 * config/arc/arc.c: Likewise.
13529 * config/arm/arm.c: Likewise.
13530 * config/avr/avr.c: Likewise.
13531 * config/bfin/bfin.c: Likewise.
13532 * config/c6x/c6x.c: Likewise.
13533 * config/cr16/cr16.c: Likewise.
13534 * config/cris/cris.c: Likewise.
13535 * config/darwin.c: Likewise.
13536 * config/epiphany/epiphany.c: Likewise.
13537 * config/fr30/fr30.c: Likewise.
13538 * config/frv/frv.c: Likewise.
13539 * config/ft32/ft32.c: Likewise.
13540 * config/h8300/h8300.c: Likewise.
13541 * config/i386/winnt.c: Likewise.
13542 * config/ia64/ia64.c: Likewise.
13543 * config/iq2000/iq2000.c: Likewise.
13544 * config/lm32/lm32.c: Likewise.
13545 * config/m32c/m32c.c: Likewise.
13546 * config/m32r/m32r.c: Likewise.
13547 * config/m68k/m68k.c: Likewise.
13548 * config/mcore/mcore.c: Likewise.
13549 * config/microblaze/microblaze.c: Likewise.
13550 * config/mips/mips.c: Likewise.
13551 * config/mmix/mmix.c: Likewise.
13552 * config/mn10300/mn10300.c: Likewise.
13553 * config/moxie/moxie.c: Likewise.
13554 * config/msp430/msp430.c: Likewise.
13555 * config/nds32/nds32-isr.c: Likewise.
13556 * config/nds32/nds32.c: Likewise.
13557 * config/nios2/nios2.c: Likewise.
13558 * config/nvptx/nvptx.c: Likewise.
13559 * config/pa/pa.c: Likewise.
13560 * config/pdp11/pdp11.c: Likewise.
13561 * config/powerpcspe/powerpcspe.c: Likewise.
13562 * config/riscv/riscv.c: Likewise.
13563 * config/rl78/rl78.c: Likewise.
13564 * config/rx/rx.c: Likewise.
13565 * config/s390/s390.c: Likewise.
13566 * config/sh/sh.c: Likewise.
13567 * config/sol2.c: Likewise.
13568 * config/sparc/sparc.c: Likewise.
13569 * config/spu/spu.c: Likewise.
13570 * config/stormy16/stormy16.c: Likewise.
13571 * config/tilegx/tilegx.c: Likewise.
13572 * config/tilepro/tilepro.c: Likewise.
13573 * config/v850/v850.c: Likewise.
13574 * config/vax/vax.c: Likewise.
13575 * config/visium/visium.c: Likewise.
13576 * config/xtensa/xtensa.c: Likewise.
13578 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13581 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13582 constraints since the -mupper-regs-* switches have been
13584 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13585 into a vector from a double word element that was extracted from
13586 another vector, and eliminate extra XXPERMDI instructions.
13587 (vsx_concat_<mode>_2): Likewise.
13588 (vsx_concat_<mode>_3): Likewise.
13589 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13590 concat to allow optimizing inserts from previous extracts.
13592 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13594 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13595 memory reference to a SSP offset in TLS address space.
13596 (ix86_print_operand) <case '@'>: Remove.
13597 (ix86_print_operand_punct_valid_p): Remove '@' code.
13598 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
13599 UNSPEC_SP_TLS_TEST.
13600 (stack_tls_protect_set_<mode>): Remove.
13601 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
13602 (stack_tls_protect_test_<mode>): Remove.
13603 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
13605 2017-08-07 Olivier Hainque <hainque@adacore.com>
13608 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
13610 2017-08-07 Douglas Rupp <rupp@adacore.com>
13612 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
13613 variable was referenced as multidir in command.
13615 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13618 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13621 2017-08-07 Martin Liska <mliska@suse.cz>
13623 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13624 * config/rl78/rl78.c: Add include of attribs.h.
13625 * config/sh/sh.c: Likewise.
13626 * config/v850/v850.c: Likewise.
13628 2017-08-07 Tom de Vries <tom@codesourcery.com>
13630 PR middle-end/78266
13631 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13633 2017-08-07 Martin Liska <mliska@suse.cz>
13635 * config/mips/mips.c: Include attribs.h.
13637 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13640 * doc/invoke.texi: Document change in behvaior for -Ofast for
13643 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13645 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13647 (aarch64_pop_regs): Likewise.
13648 (aarch64_gen_load_pair): Likewise.
13649 (aarch64_save_callee_saves): Likewise.
13650 (aarch64_restore_callee_saves): Likewise.
13652 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13654 * config/i386/i386.c: Revert the last change.
13656 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13659 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13661 (ix86_finalize_stack_frame_flags): This. Also clear
13662 frame_pointer_needed if -fno-omit-frame-pointer is used without
13664 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13665 with ix86_finalize_stack_frame_flags.
13666 (ix86_expand_epilogue): Likewise.
13667 (ix86_expand_split_stack_prologue): Likewise.
13669 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13672 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13673 to P_AES for Westmere.
13675 2017-08-07 Jonathan Yong <10walls@gmail.com>
13677 * config/i386/mingw.opt (fset-stack-executable): Removed.
13678 * config/i386/cygming.opt (fset-stack-executable): Moved
13680 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13682 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13684 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13686 2017-08-07 Marek Polacek <polacek@redhat.com>
13688 PR middle-end/81737
13689 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13691 2017-08-07 Martin Liska <mliska@suse.cz>
13693 * attribs.h (canonicalize_attr_name): New function.
13694 (cmp_attribs): Move from c-format.c and adjusted.
13695 (is_attribute_p): Moved from tree.h.
13696 * tree-inline.c: Add new includes.
13697 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13698 (private_is_attribute_p): Remove.
13699 (private_lookup_attribute): Likewise.
13700 (private_lookup_attribute_by_prefix): Simplify.
13701 (remove_attribute): Use is_attribute_p.
13702 * tree.h: Remove removed declarations.
13704 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13706 PR middle-end/81698
13707 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13708 instead of computing it in the function. Formatting fix.
13709 (expand_case): Don't rely on default_edge being the first edge,
13710 clear it if removing it, pass default_edge to
13711 emit_case_dispatch_table.
13712 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13715 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13717 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13718 insn in the function, emit NOP after the insn.
13720 2017-08-06 Tom de Vries <tom@codesourcery.com>
13722 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13725 2017-08-06 Tom de Vries <tom@codesourcery.com>
13727 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13730 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13732 PR tree-optimization/57371
13733 * match.pd: New pattern.
13735 2017-08-04 Marek Polacek <polacek@redhat.com>
13737 PR middle-end/81695
13738 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13739 perform the computation in offset_int.
13741 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13743 PR tree-optimization/81136
13744 * tree-vectorizer.h: Include tree-hash-traits.h.
13745 (vec_base_alignments): New typedef.
13746 (vec_info): Add a base_alignments field.
13747 (vect_record_base_alignments): Declare.
13748 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13750 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13751 (create_data_ref): Add an is_conditional_in_stmt argument.
13752 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13753 the is_conditional_in_stmt field.
13754 (data_ref_loc): Add an is_conditional_in_stmt field.
13755 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13756 (find_data_references_in_stmt): Update call to create_data_ref.
13757 (graphite_find_data_references_in_stmt): Likewise.
13758 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13759 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13760 (vect_record_base_alignment): New function.
13761 (vect_record_base_alignments): Likewise.
13762 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13763 for nested statements even if we fail to compute a misalignment.
13764 Use pooled base alignments for unconditional references.
13765 (vect_find_same_alignment_drs): Compare base addresses instead
13767 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13768 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13770 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13772 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13773 Add an explicit name for the enum. Use auto_vec for slp_instances
13774 and grouped_stores.
13775 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13777 (_bb_vec_info): Add a constructor and destructor.
13778 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13779 (destroy_loop_vec_info): Delete.
13780 (vect_destroy_datarefs): Likewise.
13781 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13782 (vec_info::vec_info): New function.
13783 (vec_info::~vec_info): Likewise.
13784 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13785 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13786 destroy_loop_vec_info.
13787 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13788 (_loop_vec_info::_loop_vec_info): ...this.
13789 (destroy_loop_vec_info): Replace with...
13790 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13791 the stmt_vec_infos. Leave handling of vec_info information to its
13792 destructor. Remove explicit vector releases.
13793 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13794 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13795 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13796 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13797 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13798 (destroy_bb_vec_info): Replace with...
13799 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13800 information to its destructor.
13801 (vect_slp_analyze_bb_1): Use new and delete instead of
13802 new_bb_vec_info and destroy_bb_vec_info.
13803 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13806 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13808 * tree-data-ref.h (subscript): Add access_fn field.
13809 (data_dependence_relation): Add could_be_independent_p.
13810 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13811 (same_access_functions): Move to tree-data-ref.c.
13812 * tree-data-ref.c (ref_contains_union_access_p): New function.
13813 (access_fn_component_p): Likewise.
13814 (access_fn_components_comparable_p): Likewise.
13815 (dr_analyze_indices): Add a reference to access_fn_component_p.
13816 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13818 (constant_access_functions): Likewise.
13819 (add_other_self_distances): Likewise.
13820 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13821 (initialize_data_dependence_relation): Use XCNEW and remove
13822 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13823 of access functions that have the same type. Allow the
13824 subsequence to end with different bases in some circumstances.
13825 Record the chosen access functions in SUB_ACCESS_FN.
13826 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13827 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13828 (subscript_dependence_tester_1): Likewise dra and drb.
13829 (build_classic_dist_vector): Update calls accordingly.
13830 (subscript_dependence_tester): Likewise.
13831 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13832 DDR_COULD_BE_INDEPENDENT_P.
13833 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13834 comp_alias_ddrs instead of may_alias_ddrs.
13835 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13837 (vect_analyze_data_ref_dependence): Use it if
13838 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13839 distance vectors if that fails.
13840 (dependence_distance_ge_vf): New function.
13841 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13842 LOOP_VINFO_MAY_ALIAS_DDRS.
13844 2017-08-04 Richard Biener <rguenther@suse.de>
13846 PR middle-end/81705
13847 * fold-const.c (fold_binary_loc): Properly restrict
13848 minus_var0 && minus_var1 case when associating undefined overflow
13851 2017-08-04 Tom de Vries <tom@codesourcery.com>
13853 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13855 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13857 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13858 Don't start diagnostic messages with a capital letter.
13859 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13861 (rs6000_invalid_builtin): Likewise.
13862 (rs6000_trampoline_init): Likewise.
13864 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13867 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13868 after setting changeable df flags.
13870 2017-08-03 Richard Biener <rguenther@suse.de>
13872 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13873 up if the use is in USE - X.
13875 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13877 * toplev.c (dumpfile.h): New include.
13878 (internal_error_reentered): New static function. Use it...
13879 (internal_error_function): ...here to handle reentered internal_error.
13881 2017-08-03 Richard Biener <rguenther@suse.de>
13883 PR middle-end/81148
13884 * fold-const.c (split_tree): Add minus_var and minus_con
13885 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13886 here but always use minus_*.
13887 (associate_trees): Assert we never associate with MINUS_EXPR
13888 and NULL first operand. Do not recurse for PLUS_EXPR operands
13889 when associating as MINUS_EXPR either.
13890 (fold_binary_loc): Track minus_var and minus_con.
13892 2017-08-03 Tom de Vries <tom@codesourcery.com>
13895 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13896 ACCEL_COMPILER, apply finish_options on
13897 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13899 2017-08-03 Tom de Vries <tom@codesourcery.com>
13902 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13903 function_entry_patch_area_size > 0.
13905 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13908 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13909 instead of 10??LU, perform unit multiplication in wide_int,
13910 don't change alloc_object_size_limit if the limit is larger
13913 PR tree-optimization/81655
13914 PR tree-optimization/81588
13915 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
13916 the case when ranges[i].low and high are 1 for unsigned type with
13919 PR middle-end/81052
13920 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
13921 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
13923 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13925 * tree-vrp.h: Add include guard.
13927 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
13930 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
13931 (ud2): New insn pattern.
13932 * config/i386/i386.c (ix86_expand_epilogue):
13933 For naked functions, generate ud2 instead of trap insn.
13935 2017-08-02 Marek Polacek <polacek@redhat.com>
13938 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
13940 2017-08-02 Tom de Vries <tom@codesourcery.com>
13941 Cesar Philippidis <cesar@codesourcery.com>
13943 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
13944 Add missing edge probabilities.
13946 2017-08-02 Tamar Christina <tamar.christina@arm.com>
13948 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
13949 Correct endianness.
13951 2017-08-02 Jakub Jelinek <jakub@redhat.com>
13953 PR middle-end/79499
13954 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
13955 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
13956 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
13957 prologue_seq sequences - if any.
13959 2017-08-02 Richard Biener <rguenther@suse.de>
13961 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
13962 via vectors if supported, integer extracts via punning if supported
13963 or otherwise vector extracts.
13965 2017-08-02 Richard Biener <rguenther@suse.de>
13967 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
13969 (bitmap_insert_into_set): ... this.
13971 2017-08-02 Richard Biener <rguenther@suse.de>
13973 PR tree-optimization/81633
13975 2015-08-17 Alan Hayward <alan.hayward@arm.com>
13977 PR tree-optimization/71752
13978 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
13980 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
13982 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
13983 (machine_function::call_ms2sysv_pad_out): Remove field.
13984 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
13985 (ix86_compute_frame_layout): Likewise.
13987 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
13990 * config/i386/i386.c (ix86_set_func_type): Disallow naked
13991 attribute with interrupt attribute.
13993 2017-08-01 Andrew Pinski <apinski@cavium.com>
13995 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
13996 BIT_INSERT_EXPR's operand 1
13997 to see if the types precision matches.
13999 2017-08-01 Martin Liska <mliska@suse.cz>
14001 PR middle-end/70140
14002 * builtins.c (expand_builtin_memcpy_args): Remove.
14003 (expand_builtin_memcpy): Call newly added function
14004 expand_builtin_memory_copy_args.
14005 (expand_builtin_memcpy_with_bounds): Likewise.
14006 (expand_builtin_mempcpy): Remove last argument.
14007 (expand_builtin_mempcpy_with_bounds): Likewise.
14008 (expand_builtin_memory_copy_args): New function created from
14009 expand_builtin_mempcpy_args with small modifications.
14010 (expand_builtin_mempcpy_args): Remove.
14011 (expand_builtin_stpcpy): Remove unused argument.
14012 (expand_builtin): Likewise.
14013 (expand_builtin_with_bounds): Likewise.
14015 2017-08-01 Martin Liska <mliska@suse.cz>
14018 Make mempcpy more optimal (PR middle-end/70140).
14020 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14023 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14024 __builtin_vec_cmpne verify both arguments are compatible vectors
14025 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14026 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14027 move computation of aligned to after checking the argument types.
14031 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14032 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14035 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14036 Doug Rupp <rupp@adacore.com>
14037 Olivier Hainque <hainque@adacore.com>
14039 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14040 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14042 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14043 for TARGET_OS_CPP_BUILTIN.
14044 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14045 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14047 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14048 passing required abi options to the assembler for EABI configurations.
14049 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14050 of .text.hot and .text.unlikely sections for kernel modules when
14051 using ARM style exceptions.
14052 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14053 options. Add EXTRA_CC1_SPEC.
14054 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14056 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14058 (ARM_TARGET2_DWARF_FORMAT): Define.
14059 * config/arm/t-vxworks: Adjust multilib control to removal of the
14060 Diab command line options.
14062 2017-08-01 Martin Liska <mliska@suse.cz>
14064 PR gcov-profile/81561
14065 * gcov.c (unblock): Make unblocking safe as we need to preserve
14066 index correspondence of blocks and block_lists.
14068 2017-08-01 Richard Biener <rguenther@suse.de>
14070 PR tree-optimization/81181
14071 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14072 (compute_antic): ... end of iteration here.
14074 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14076 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14077 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14078 (ftree-slp-vectorize): Likewise.
14079 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14080 can no longer be set independent of flag_tree_loop_vectorize.
14081 * omp-general.c (emp_max_vf): Likewise.
14082 * opts.c (enable_fdo_optimizations): Remove references to
14083 flag_tree_vectorize, these are now implicit.
14084 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14085 and leave it for the options machinery.
14087 2017-08-01 Martin Liska <mliska@suse.cz>
14089 PR middle-end/70140
14090 * builtins.c (expand_builtin_memcpy_args): Remove.
14091 (expand_builtin_memcpy): Call newly added function
14092 expand_builtin_memory_copy_args.
14093 (expand_builtin_memcpy_with_bounds): Likewise.
14094 (expand_builtin_mempcpy): Remove last argument.
14095 (expand_builtin_mempcpy_with_bounds): Likewise.
14096 (expand_builtin_memory_copy_args): New function created from
14097 expand_builtin_mempcpy_args with small modifications.
14098 (expand_builtin_mempcpy_args): Remove.
14099 (expand_builtin_stpcpy): Remove unused argument.
14100 (expand_builtin): Likewise.
14101 (expand_builtin_with_bounds): Likewise.
14103 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14106 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
14107 print "ds:" only for immediates in generic address space.
14109 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14112 * config/i386/i386.c (ix86_funciton_naked): New prototype.
14113 (ix86_function_ok_for_sibcall): Return false for naked functions.
14115 2017-08-01 Richard Biener <rguenther@suse.de>
14117 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
14118 (compute_antic): Seed worklist with exit block predecessors.
14119 * cfganal.c (dfs_find_deadend): For a cycle return the source
14120 of the edge closing it.
14122 2017-08-01 Tamar Christina <tamar.christina@arm.com>
14124 * config/aarch64/aarch64.c
14125 (aarch64_can_const_movi_rtx_p): Move 0 check.
14127 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14129 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
14130 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
14132 * match.pd: Ditto in address comparison pattern.
14134 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14136 PR tree-optimization/81627
14137 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
14138 closed ssa form for store-store chain.
14140 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14142 PR tree-optimization/81620
14143 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
14144 for store-store chain.
14146 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14148 PR tree-optimization/81588
14149 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
14150 ranges[i].in_p, invert comparison code ccode. For >/>=,
14151 swap rhs1 and rhs2 and comparison code unconditionally,
14152 for </<= don't do that. Don't swap rhs1/rhs2 again if
14153 ranges[i].in_p, instead invert comparison code ccode if
14154 opcode or oe->rank is BIT_IOR_EXPR.
14157 * optabs.def (vec_extract_optab, vec_init_optab): Change from
14158 a direct optab to conversion optab.
14159 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
14160 with GET_MODE_INNER as last argument instead of optab_handler.
14161 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14162 vector extraction if possible and optab is available.
14163 * expr.c (store_constructor): Use convert_optab_handler instead
14164 of optab_handler. Use vector initialization from smaller
14165 vectors if possible and optab is available.
14166 * tree-vect-stmts.c (vectorizable_load): Likewise.
14167 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14168 now have two modes.
14169 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14170 of vec_init from half-sized vectors with the same element mode.
14171 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14172 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14173 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14174 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14175 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14176 after mode in gen_vec_extract* calls.
14177 (vec_extract<mode>): Renamed to ...
14178 (vec_extract<mode><ssescalarmodelower>): ... this.
14179 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14180 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14181 element mode after mode in gen_vec_init* calls.
14182 (VEC_INIT_HALF_MODE): New mode iterator.
14183 (vec_init<mode>): Renamed to ...
14184 (vec_init<mode><ssescalarmodelower>): ... this.
14185 (vec_init<mode><ssehalfvecmodelower>): New expander.
14186 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14187 (vec_extractv2sfsf): ... this.
14188 (vec_initv2sf): Renamed to ...
14189 (vec_initv2sfsf): ... this.
14190 (vec_extractv2si): Renamed to ...
14191 (vec_extractv2sisi): ... this.
14192 (vec_initv2si): Renamed to ...
14193 (vec_initv2sisi): ... this.
14194 (vec_extractv4hi): Renamed to ...
14195 (vec_extractv4hihi): ... this.
14196 (vec_initv4hi): Renamed to ...
14197 (vec_initv4hihi): ... this.
14198 (vec_extractv8qi): Renamed to ...
14199 (vec_extractv8qiqi): ... this.
14200 (vec_initv8qi): Renamed to ...
14201 (vec_initv8qiqi): ... this.
14202 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14203 (vec_init<mode>): Renamed to ...
14204 (vec_init<mode><VEC_base_l>): ... this.
14205 (vec_extract<mode>): Renamed to ...
14206 (vec_extract<mode><VEC_base_l>): ... this.
14207 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14208 (vec_initv2sfsf): ... this.
14209 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14210 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14211 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14212 element mode after mode in gen_vec_init* calls.
14213 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14214 (vec_init<mode><Vel>): ... this.
14215 (vec_extract<mode>): Renamed to ...
14216 (vec_extract<mode><Vel>): ... this.
14217 * config/aarch64/iterators.md (Vel): New mode attribute.
14218 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14219 Add element mode after mode in gen_vec_extract* calls.
14220 * config/s390/vector.md (non_vec_l): New mode attribute.
14221 (vec_extract<mode>): Renamed to ...
14222 (vec_extract<mode><non_vec_l>): ... this.
14223 (vec_init<mode>): Renamed to ...
14224 (vec_init<mode><non_vec_l>): ... this.
14225 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14226 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14228 * config/arm/iterators.md (V_elem_l): New mode attribute.
14229 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14230 (vec_extract<mode><V_elem_l>): ... this.
14231 (vec_extractv2di): Renamed to ...
14232 (vec_extractv2didi): ... this.
14233 (vec_init<mode>): Renamed to ...
14234 (vec_init<mode><V_elem_l>): ... this.
14235 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14236 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14237 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14238 Add element mode after gen_vec_extract* calls.
14239 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14240 (vec_init<mode><unitmode>): ... this.
14241 (vec_extract<mode>): Renamed to ...
14242 (vec_extract<mode><unitmode>): ... this.
14243 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14244 (vec_init<mode><unitmode>): ... this.
14245 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14246 (vec_initv2sfsf): ... this.
14247 (vec_extractv2sf): Renamed to ...
14248 (vec_extractv2sfsf): ... this.
14249 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14250 Add element mode after gen_vec_extract* calls.
14251 * config/mips/mips.md (unitmode): New mode iterator.
14252 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14253 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14254 * config/spu/spu.md (inner_l): New mode attribute.
14255 (vec_init<mode>): Renamed to ...
14256 (vec_init<mode><inner_l>): ... this.
14257 (vec_extract<mode>): Renamed to ...
14258 (vec_extract<mode><inner_l>): ... this.
14259 * config/sparc/sparc.md (veltmode): New mode iterator.
14260 (vec_init<VMALL:mode>): Renamed to ...
14261 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14262 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14263 (vec_initv2sisi): ... this.
14264 (vec_initv2sf): Renamed to ...
14265 (vec_initv2sfsf): ... this.
14266 (vec_extractv2sf): Renamed to ...
14267 (vec_extractv2sfsf): ... this.
14268 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14269 (vec_init<mode>): Renamed to ...
14270 (vec_init<mode><VEC_base_l>): ... this.
14271 (vec_extract<mode>): Renamed to ...
14272 (vec_extract<mode><VEC_base_l>): ... this.
14273 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14274 (vec_initv2sfsf): ... this.
14275 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14276 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14277 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14278 gen_vec_init* calls.
14280 2017-08-01 Richard Biener <rguenther@suse.de>
14282 PR tree-optimization/81297
14283 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14284 TREE_OVERFLOW from INTEGER_CSTs.
14286 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14288 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14290 2017-07-31 Carl Love <cel@us.ibm.com>
14292 * config/rs6000/rs6000-c: Add support for built-in functions
14293 vector signed char vec_xl_be (signed long long, signed char *);
14294 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14295 vector signed int vec_xl_be (signed long long, signed int *);
14296 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14297 vector signed long long vec_xl_be (signed long long, signed long long *);
14298 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14299 vector signed short vec_xl_be (signed long long, signed short *);
14300 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14301 vector double vec_xl_be (signed long long, double *);
14302 vector float vec_xl_be (signed long long, float *);
14303 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14304 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14305 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14307 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14308 (altivec_expand_builtin): Add switch statement to call
14309 altivec_expand_xl_be for each builtin.
14310 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14311 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14312 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14313 __builtin_vsx_le_be_v16qi.
14314 * doc/extend.texi: Update the built-in documentation file for the
14315 new built-in functions.
14317 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14320 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14322 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14324 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14326 * config.gcc: Add z14.
14327 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14328 CPU model numbers for z13s and z14.
14329 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14331 * config/s390/s390-opts.h (enum processor_type): Rename
14332 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14333 * config/s390/s390.c (processor_table): Add field for CPU name to
14334 be passed to Binutils.
14335 (s390_asm_output_machine_for_arch): Use the new field in
14336 processor_table for Binutils.
14337 (s390_expand_builtin): Replace arch12 with z14.
14338 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14339 (s390_get_sched_attrmask): Likewise.
14340 (s390_get_unit_mask): Likewise.
14341 * config/s390/s390.opt: Add z14 to processor_type enum.
14343 2017-07-31 Martin Jambor <mjambor@suse.cz>
14346 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14347 regardless of optimization level.
14349 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14350 Martin Liska <mliska@suse.cz>
14352 * predict.def: Remove old comment and adjust probability.
14353 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14354 PREDICT statements.
14356 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14359 * config/i386/i386.c (ix86_function_naked): New function.
14360 (ix86_can_use_return_insn_p): Return false for naked functions.
14361 (ix86_expand_prologue): Skip prologue for naked functions.
14362 (ix86_expand_epilogue): Skip epilogue for naked functions
14363 and emit trap instruction.
14364 (ix86_warn_func_return): New function.
14365 (ix86_attribute_table): Add "naked" attribute specification.
14366 (TARGET_WARN_FUNC_RETURN): Define.
14367 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14369 2017-07-31 Martin Liska <mliska@suse.cz>
14371 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14372 (dump_gimple_bb_header): Always dump BB info.
14373 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14375 2017-07-31 Martin Liska <mliska@suse.cz>
14378 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14379 also with current_function_decl non-null equality.
14381 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14384 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14385 change type to the element type, instead add eltype variable and
14386 use it where we are interested in the element type.
14388 PR tree-optimization/81603
14389 * ipa-polymorphic-call.c
14390 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14391 offset arithmetic in offset_int, bail out if the resulting bit offset
14392 doesn't fit into shwi.
14394 2017-07-31 Martin Liska <mliska@suse.cz>
14396 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14397 (gimplify_save_expr): Fix comment.
14399 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14402 * config/i386/i386.c (ix86_function_arg): Update arguments for
14404 (ix86_compute_frame_layout): Set the initial stack offset to
14405 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14406 INCOMING_FRAME_SP_OFFSET.
14407 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14408 stack before exception handler returns.
14409 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14410 the 'ERROR_CODE' for exception handler.
14412 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14414 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14415 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14416 (ASM_OUTPUT_REG_POP): Ditto.
14417 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14418 instead of asm_fprintf to output pure string.
14420 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14422 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14423 to imported_module_or_decl hook.
14424 (debug_nothing_tree_tree_tree_bool): Remove.
14425 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14426 * debug.c (do_nothing_debug_hooks): Use
14427 debug_nothing_tree_tree_tree_bool_bool instead of
14428 debug_nothing_tree_tree_tree_bool.
14429 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14430 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14431 * sdbout.c (sdb_debug_hooks): Likewise.
14432 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14433 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14435 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14436 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14437 attribute, don't add anything.
14439 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14441 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14442 (fold_build2_stat_loc): Likewise.
14443 (fold_build3_stat_loc): Likewise.
14444 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14445 (fold_build1_loc): Remove macro.
14446 (fold_build2_loc): Likewise.
14447 (fold_build3_loc): Likewise.
14449 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14451 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14452 (gimple_build_debug_bind_source_stat): Likewise.
14453 * gimple.h (gimple_build_debug_bind): Remove macro.
14454 (gimple_build_debug_bind_source): Likewise.
14456 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14458 * bitmap.c (bitmap_alloc): Adjust.
14459 (bitmap_gc_alloc): Likewise.
14460 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14462 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14464 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14465 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14466 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14467 (bitmap_gc_alloc_stat): Likewise.
14468 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14470 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14472 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14473 * rtl.h (shallow_copy_rtx): Remove macro.
14475 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14477 * emit-rtl.c (gen_raw_REG): Adjust.
14478 * gengenrtl.c (gendef): Likewise.
14479 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14480 * rtl.h (rtx_alloc): Remove macro.
14482 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14484 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14485 (build_tree_list_stat): Likewise.
14486 * tree.h (build_tree_list): Remove macro.
14487 (build_tree_list_vec): Likewise.
14489 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14491 * tree.c (make_vector_stat): Remove _stat from name.
14492 (build_vector_stat): Likewise.
14493 * tree.h (make_vector_stat): Remove macro.
14494 (build_vector_stat): Likewise.
14496 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14498 * tree.h (build_var_debug_value): Remove prototype.
14500 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14502 * tree.c (tree_cons_stat): Remove _stat from name.
14503 * tree.h (tree_cons): Remove macro.
14505 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14507 * tree.c (build_vl_exp_stat): Remove _stat from name.
14508 * tree.h (build_vl_exp): Remove macro.
14510 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14512 * tree.c (build_decl_stat): Remove _stat from name.
14513 * tree.h (build_decl): Remove macro.
14515 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14517 * gimple.c (gimple_build_with_ops_stat): Adjust.
14518 (gimple_alloc_stat): Remove _stat from name.
14519 * gimple.h (gimple_alloc): Remove macro.
14521 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14523 * tree.c (make_tree_vec_stat): Remove _stat from name.
14524 (grow_tree_vec_stat): Likewise.
14525 * tree.h (make_tree_vec_stat): Adjust prototype.
14526 (grow_tree_vec_stat): Likewise.
14527 (make_tree_vec): Remove macro.
14528 (grow_tree_vec): Likewise.
14530 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14532 * fold-const.c (fold_build1_stat_loc): Adjust.
14533 (fold_build2_stat_loc): Likewise.
14534 (fold_build3_stat_loc): Likewise.
14535 * tree.c (build0_stat): Remove _stat from name.
14536 (build1_stat): Likewise.
14537 (build2_stat): Likewise.
14538 (build3_stat): Likewise.
14539 (build4_stat): Likewise.
14540 (build5_stat): Likewise.
14541 * tree.h (build1_loc): Remove macro, and rename _stat function
14543 (build2_loc): Likewise.
14544 (build3_loc): Likewise.
14545 (build4_loc): Likewise.
14546 (build5_loc): Likewise.
14548 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14550 * tree.c (make_int_cst_stat): Remove _stat from name.
14551 * tree.h (make_int_cst_stat): Adjust prototype.
14552 (make_int_cst): Remove macro.
14554 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14556 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14557 * tree.h (make_tree_binfo_stat): Adjust prototype.
14558 (make_tree_binfo): Remove.
14560 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14562 * tree.c (copy_node_stat): Rename to copy_node.
14563 (build_distinct_type_copy): Adjust.
14564 * tree.h (copy_node_stat): Adjust prototype.
14565 (copy_node): Remove macro.
14567 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14569 * tree.c (make_node_stat): rename to make_node.
14570 (build_tree_list_stat): Adjust.
14571 (build0_stat): Likewise.
14572 (build2_stat): Likewise.
14573 (build3_stat): Likewise.
14574 (build4_stat): Likewise.
14575 (build5_stat): Likewise.
14576 (build_decl_stat): Likewise.
14577 * tree.h (make_node_stat): Adjust prototype.
14578 (make_node): remove macro.
14580 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14582 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14583 (PPC_FEATURE2_SCV): Likewise.
14584 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14586 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14588 * config/aarch64/aarch64.c
14589 (aarch64_internal_mov_immediate): Add new special pattern.
14590 * config/aarch64/aarch64.md (*movdi_aarch64):
14591 Add reg/32bit const mov case.
14593 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14594 Richard Sandiford <richard.sandiford@linaro.org>
14596 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14597 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14598 Add integer and movi cases.
14599 (movi-split-hf-df-sf split, fp16): New.
14600 (enabled): Added TARGET_FP_F16INST.
14601 * config/aarch64/iterators.md (GPF_HF): New.
14602 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
14604 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14606 * config/aarch64/aarch64.c
14607 (aarch64_simd_container_mode): Add prototype.
14608 (aarch64_expand_mov_immediate): Add HI support.
14609 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
14610 (aarch64_can_const_movi_rtx_p): New.
14611 (aarch64_preferred_reload_class):
14612 Remove restrictions of using FP registers for certain SIMD operations.
14613 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
14614 (aarch64_valid_floating_const): Add integer move validation.
14615 (aarch64_simd_imm_scalar_p): Remove.
14616 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14617 (aarch64_legitimate_constant_p): Expand list of supported cases.
14618 * config/aarch64/aarch64-protos.h
14619 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14620 (aarch64_reinterpret_float_as_int): New.
14621 (aarch64_simd_imm_scalar_p): Remove.
14622 * config/aarch64/constraints.md (Uvi): New.
14623 (Dd): Split into Ds and new Dd.
14624 * config/aarch64/aarch64.md (*movsi_aarch64):
14626 (*movdi_aarch64): Add SIMD mov case.
14628 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14630 * tree-predcom.c: (struct chain): Handle store-store chain in which
14631 stores for elimination only store loop invariant values.
14632 (execute_pred_commoning_chain): Ditto.
14633 (prepare_initializers_chain_store_elim): Ditto.
14634 (prepare_finalizers): Ditto.
14635 (is_inv_store_elimination_chain): New function.
14636 (initialize_root_vars_store_elim_1): New function.
14638 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14640 * tree-predcom.c: Revise general description of the pass.
14641 (enum chain_type): New enum type for store elimination.
14642 (struct chain): New field supporting store elimination.
14643 (struct component): Ditto.
14644 (dump_chain): Dump store-stores chain.
14645 (release_chain): Release resources.
14646 (split_data_refs_to_components): Compute and create component
14647 contains only stores for elimination.
14648 (get_chain_last_ref_at): New function.
14649 (make_invariant_chain): Initialization.
14650 (make_rooted_chain): Specify chain type in parameter and record it.
14651 (add_looparound_copies): Skip for store-stores chain.
14652 (determine_roots_comp): Compute type of chain and pass it to
14654 (initialize_root_vars_store_elim_2): New function.
14655 (finalize_eliminated_stores): New function.
14656 (remove_stmt): Handle store for elimination.
14657 (execute_pred_commoning_chain): Execute predictive commoning on
14658 store-store chains.
14659 (determine_unroll_factor): Skip unroll for store-stores chain.
14660 (prepare_initializers_chain_store_elim): New function.
14661 (prepare_initializers_chain): Hanlde store-store chain.
14662 (prepare_finalizers_chain, prepare_finalizers): New function.
14663 (tree_predictive_commoning_loop): Return integer value indicating
14664 if loop is unrolled or lcssa form is corrupted.
14665 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14667 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14669 * tree-predcom.c (initialize_root): Delete.
14670 (execute_pred_commoning_chain): Initialize root vars and replace
14671 reference of non-combined chain directly, rather than call above
14674 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14676 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14677 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14679 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14681 * tree-predcom.c (struct chain): New field init_seq.
14682 (release_chain): Release init_seq.
14683 (prepare_initializers_chain): Record intialization stmts in above
14685 (insert_init_seqs): New function.
14686 (tree_predictive_commoning_loop): Call insert_init_seqs.
14688 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14690 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14692 2017-07-28 Richard Biener <rguenther@suse.de>
14694 * match.pd: Remove superfluous :c.
14695 * genmatch.c (simplify::id): Add member.
14696 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14698 (current_id): New global.
14699 (dt_node::parent): Move from ...
14700 (dt_operand::parent): ... here. Add for_id member.
14701 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14702 (decision_tree::find_node): Relax order requirement when
14703 merging DT_TRUE nodes to ones inbetween the current simplify
14704 and the one we try to merge with. Add diagnostic whenever
14705 we need to enforce pattern order by not merging.
14706 (decision_tree::insert): Set current_id.
14707 (decision_tree::print_node): Dump parent node and for_id.
14708 (parser::last_id): Add member.
14709 (parser::push_simplify): Assign unique id.
14710 (parser::parser): Initialize last_id.
14712 2017-07-28 Martin Liska <mliska@suse.cz>
14715 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14716 gimple_build_debug_bind.
14718 2017-07-28 Richard Biener <rguenther@suse.de>
14720 PR tree-optimization/81502
14721 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14723 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14725 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14726 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14727 for BIT_FIELD_REF args.
14728 * fold-const.c (make_bit_field_ref): Likewise.
14729 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14731 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14734 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14735 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14736 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14737 Or it into SANITIZER_UNDEFINED.
14738 * ubsan.c: Include gimple-fold.h and varasm.h.
14739 (ubsan_expand_ptr_ifn): New function.
14740 (instrument_pointer_overflow): New function.
14741 (maybe_instrument_pointer_overflow): New function.
14742 (instrument_object_size): Formatting fix.
14743 (pass_ubsan::execute): Call instrument_pointer_overflow
14744 and maybe_instrument_pointer_overflow.
14745 * internal-fn.c (expand_UBSAN_PTR): New function.
14746 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14747 * sanitizer.def (__ubsan_handle_pointer_overflow,
14748 __ubsan_handle_pointer_overflow_abort): New builtins.
14749 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14750 * internal-fn.def (UBSAN_PTR): New internal function.
14751 * opts.c (sanitizer_opts): Add pointer-overflow.
14752 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14753 * fold-const.c (build_range_check): Compute pointer range check in
14754 integral type if pointer arithmetics would be needed. Formatting
14757 2017-07-28 Martin Liska <mliska@suse.cz>
14760 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14761 parameters that are of a variable-length.
14763 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14765 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14767 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14768 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14769 (CRT_CALL_STATIC_FUNCTION): Likewise.
14770 (ASM_DEFAULT_SPEC): New define.
14771 (ASM_SPEC32): Likewise.
14772 (ASM_SPEC64): Likewise.
14773 (ASM_SPEC_COMMON): Likewise.
14774 (ASM_SPEC): Likewise.
14775 (INVALID_64BIT): Likewise.
14776 (LINK_OS_DEFAULT_SPEC): Likewise.
14777 (LINK_OS_SPEC32): Likewise.
14778 (LINK_OS_SPEC64): Likewise.
14779 (POWERPC_LINUX): Likewise.
14780 (PTRDIFF_TYPE): Likewise.
14781 (RESTORE_FP_PREFIX): Likewise.
14782 (RESTORE_FP_SUFFIX): Likewise.
14783 (SAVE_FP_PREFIX): Likewise.
14784 (SAVE_FP_SUFFIX): Likewise.
14785 (SIZE_TYPE): Likewise.
14786 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14787 (TARGET_64BIT): Likewise.
14788 (TARGET_64BIT): Likewise.
14789 (TARGET_AIX): Likewise.
14790 (WCHAR_TYPE_SIZE): Likewise.
14791 (WCHAR_TYPE): Undefine.
14792 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14793 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14794 (CPP_OS_RTEMS_SPEC): Delete.
14795 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14796 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14798 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14800 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14802 PR tree-optimization/81578
14803 * tree-parloops.c (build_new_reduction): Bail out if
14804 reduction_code isn't one of the standard OpenMP reductions.
14805 Move the details printing after that decision.
14807 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14809 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14810 related to reload_in_progress.
14811 (splat_input_operand): Likewise.
14812 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14814 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14816 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14817 (TARGET_INSTANTIATE_DECLS): Likewise.
14818 (legitimate_indexed_address_p): Delete reload_in_progress code.
14819 (rs6000_debug_legitimate_address_p): Likewise.
14820 (rs6000_eliminate_indexed_memrefs): Likewise.
14821 (rs6000_emit_le_vsx_store): Likewise.
14822 (rs6000_emit_move_si_sf_subreg): Likewise.
14823 (rs6000_emit_move): Likewise.
14824 (register_to_reg_type): Likewise.
14825 (rs6000_pre_atomic_barrier): Likewise.
14826 (rs6000_machopic_legitimize_pic_address): Likewise.
14827 (rs6000_allocate_stack_temp): Likewise.
14828 (rs6000_address_for_fpconvert): Likewise.
14829 (rs6000_address_for_altivec): Likewise.
14830 (rs6000_secondary_memory_needed_rtx): Delete function.
14831 (rs6000_check_sdmode): Likewise.
14832 (rs6000_alloc_sdmode_stack_slot): Likewise.
14833 (rs6000_instantiate_decls): Likewise.
14834 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14835 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14836 Delete reload_in_progress.
14837 (*vec_reload_and_plus_<mptrsize>): Likewise.
14838 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14839 (vsx_div_v2di): Likewise.
14840 (vsx_udiv_v2di): Likewise.
14842 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14844 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14845 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14846 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14847 (rs6000_debug_reg_global): Delete print of LRA status.
14848 (rs6000_option_override_internal): Delete dead LRA related code.
14849 (rs6000_lra_p): Delete function.
14850 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14852 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14854 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14855 * config/riscv/rtems.h: New file.
14857 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14858 Sudakshina Das <sudi.das@arm.com>
14860 * config/aarch64/aarch64.md
14861 (define_split for and<mode>3nr_compare): Move
14862 non aarch64_logical_operand to a register.
14863 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14864 register immediate operand to a register.
14865 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14867 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14869 PR middle-end/81564
14870 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14872 2017-07-27 Richard Biener <rguenther@suse.de>
14874 PR tree-optimization/81573
14875 PR tree-optimization/81494
14876 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14877 multi defuse cycle case.
14879 2017-07-27 Richard Biener <rguenther@suse.de>
14881 PR tree-optimization/81571
14882 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14885 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14887 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14888 earlier and only if MASK_FPU is set. Adjust formatting.
14890 2017-07-27 Martin Liska <mliska@suse.cz>
14892 * opt-functions.awk: Add validation of value of Init.
14893 * optc-gen.awk: Pass new argument.
14895 2017-07-27 Martin Liska <mliska@suse.cz>
14897 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14898 Fix wrong condition.
14900 2017-07-27 Martin Liska <mliska@suse.cz>
14902 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14903 BBs and edges seen by autoFDO.
14905 2017-07-27 Richard Biener <rguenther@suse.de>
14907 PR tree-optimization/81502
14908 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14909 with incompatible but same sized type.
14910 (execute_update_addresses_taken): Likewise.
14912 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
14914 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
14915 flag_tree_loop_vectorize rather than flag_tree_vectorize.
14917 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14920 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
14921 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
14922 Change s_operand to memory_operand.
14924 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
14926 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
14927 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
14928 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
14929 Emit instructions rather than returning an expression. Handle TFmode
14930 and KFmode by casting to TImode.
14931 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
14932 (rs6000_emit_le_vsx_store): Likewise.
14933 * config/rs6000/vsx.md (VSX_TI): New iterator.
14934 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
14935 (*vsx_le_undo_permute_<mode>): Likewise.
14936 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
14937 emit the split sequence.
14938 (*vsx_le_perm_store_<mode>): Likewise.
14940 2017-07-27 Jakub Jelinek <jakub@redhat.com>
14942 PR tree-optimization/81555
14943 PR tree-optimization/81556
14944 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
14945 if true, force CHANGED for the recursive invocation.
14946 (reassociate_bb): Remember original length of ops array, pass
14947 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
14949 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
14950 attributes for noipa attribute. For naked attribute use
14951 lookup_attribute first before lookup_attribute_spec.
14952 * final.c (rest_of_handle_final): Disable IPA RA for functions with
14954 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
14955 for functions with noipa attribute.
14956 (cgraph_externally_visible_p): Return true for functions with noipa
14958 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
14959 for functions with noipa attribute.
14960 * doc/extend.texi: Document noipa function attribute.
14961 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
14962 also for functions with noipa attribute.
14963 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
14965 2017-07-26 Andrew Pinski <apinski@cavium.com>
14967 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
14968 vec_unalign_load_cost and vec_unalign_store_cost.
14970 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14972 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
14973 -mvsx-small-integer option.
14974 (ISA_3_0_MASKS_IEEE): Likewise.
14975 (OTHER_VSX_VECTOR_MASKS): Likewise.
14976 (POWERPC_MASKS): Likewise.
14977 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
14978 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
14979 code, only testing for DImode being allowed in non-VSX floating
14981 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
14982 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
14984 (rs6000_option_override_internal): Delete -mvsx-small-integer.
14985 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
14986 TARGET_P8_VECTOR test.
14987 (rs6000_secondary_reload_simple_move): Likewise.
14988 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
14989 since TARGET_P9_VECTOR was already tested.
14990 (rs6000_opt_masks): Remove -mvsx-small-integer.
14991 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
14992 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
14994 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
14995 test for TARGET_VEXTRACTUB was used, and that uses
14997 (p9 extract splitter): Likewise.
14998 (vsx_extract_<mode>_di_p9): Likewise.
14999 (vsx_extract_<mode>_store_p9): Likewise.
15000 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15001 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15002 the elimination of TARGET_VSX_SMALL_INTEGER.
15003 (vsx_extract_<mode>_p8): Likewise.
15004 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15005 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15006 (vsx_set_<mode>_p9): Likewise.
15007 (vsx_set_v4sf_p9): Likewise.
15008 (vsx_set_v4sf_p9_zero): Likewise.
15009 (vsx_insert_extract_v4sf_p9): Likewise.
15010 (vsx_insert_extract_v4sf_p9_2): Likewise.
15011 * config/rs6000/rs6000.md (sign extend splitter): Change
15012 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15013 (floatsi<mode>2_lfiwax_mem): Likewise.
15014 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15015 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15016 since a test for TARGET_P9_VECTOR was used.
15017 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15018 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15019 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15020 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15021 TARGET_P8_VECTOR test.
15022 (fix_trunc<mode>si2_stfiwx): Likewise.
15023 (fix_trunc<mode>si2_internal): Likewise.
15024 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15025 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15027 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15028 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15029 TARGET_P8_VECTOR test.
15030 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15031 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15032 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15034 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15035 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15036 since a test for TARGET_P9_VECTOR was used.
15037 (splitter for loading small constants): Likewise.
15039 2017-07-26 Andrew Pinski <apinski@cavium.com>
15041 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15044 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15047 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15048 (fp_valid_at): Likewise.
15050 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15052 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15053 (qdf24xx_addrcost_table): Likewise.
15054 (cortexa57_tunings): Update to use generic_branch_cost.
15055 (cortexa72_tunings): Likewise.
15056 (cortexa73_tunings): Likewise.
15057 (qdf24xx_tunings): Likewise.
15059 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15061 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15062 (thunderx2t99_branch_cost): Likewise.
15063 (cortexa35_tunings): Update to use generic_branch_cost.
15064 (cortexa53_tunings): Likewise.
15065 (cortexa57_tunings): Likewise.
15066 (cortexa72_tunings): Likewise.
15067 (cortexa73_tunings): Likewise.
15068 (thunderx2t99_tunings): Likewise.
15070 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15072 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15073 (sparc_option_override): Honour MASK_FSMULD.
15074 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15075 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15076 * config/sparc/sparc.opt (mfsmuld): New option.
15077 * doc/invoke.texi (mfsmuld): Document option.
15079 2017-07-26 Marek Polacek <polacek@redhat.com>
15081 PR middle-end/70992
15082 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15084 2017-07-26 Richard Biener <rguenther@suse.de>
15086 * gimple-match-head.c (do_valueize): Return OP if valueize
15088 (get_def): New helper to get at the def stmt of a SSA name
15089 if valueize allows.
15090 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15091 do_valueize to get at the def stmt.
15092 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15094 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15096 PR middle-end/46932
15097 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
15099 2017-07-26 Martin Liska <mliska@suse.cz>
15102 * function.c (expand_function_start): Make expansion of
15103 nonlocal_goto_save_area after parm_birth_insn.
15105 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15107 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
15108 from all CPU target flags enable members.
15110 2017-07-26 Richard Biener <rguenther@suse.de>
15112 * genmatch.c (dt_simplify::gen): Make iterator vars const.
15113 (decision_tree::gen): Make 'type' const.
15114 (write_predicate): Likewise.
15116 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15118 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15119 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
15120 (rs6000_option_override_internal): Likewise.
15121 (rs6000_expand_vector_set): Likewise.
15122 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
15123 (TARGET_UPPER_REGS_SF): Likewise.
15124 (TARGET_UPPER_REGS_DI): Likewise.
15125 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
15126 (TARGET_DIRECT_MOVE_64BIT): Likewise.
15127 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
15128 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15129 (Splitters for DI constants in Altivec registers): Likewise.
15130 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
15131 (vsx_set_v4sf_p9): Likewise.
15132 (vsx_set_v4sf_p9_zero): Likewise.
15133 (vsx_insert_extract_v4sf_p9): Likewise.
15134 (vsx_insert_extract_v4sf_p9_2): Likewise.
15136 2017-07-25 Carl Love <cel@us.ibm.com>
15138 * doc/extend.texi: Update the built-in documentation file for the
15139 existing built-in functions
15140 vector signed char vec_cnttz (vector signed char);
15141 vector unsigned char vec_cnttz (vector unsigned char);
15142 vector signed short vec_cnttz (vector signed short);
15143 vector unsigned short vec_cnttz (vector unsigned short);
15144 vector signed int vec_cnttz (vector signed int);
15145 vector unsigned int vec_cnttz (vector unsigned int);
15146 vector signed long long vec_cnttz (vector signed long long);
15147 vector unsigned long long vec_cnttz (vector unsigned long long);
15149 2017-07-25 Andrew Pinski <apinski@cavium.com>
15151 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
15152 accesses where the use is for the first operand of a BIT_INSERT.
15154 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15157 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15158 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15160 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15162 * config/i386/gstabs.h: Delete.
15163 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15165 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15167 * config/i386/i386.c (ix86_decompose_address): Do not check for
15168 register RTX when looking at index_reg or base_reg.
15169 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15171 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15173 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15174 to update EH info here.
15176 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15178 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15180 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15182 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15184 2017-07-25 Torsten Duwe <duwe@suse.de>
15186 * common.opt: Introduce -fpatchable-function-entry
15187 command line option, and its variables function_entry_patch_area_size
15188 and function_entry_patch_area_start.
15189 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15190 including a two-value parser.
15191 * target.def (print_patchable_function_entry): New target hook.
15192 * targhooks.h (default_print_patchable_function_entry): New function.
15193 * targhooks.c (default_print_patchable_function_entry): Likewise.
15194 * toplev.c (process_options): Switch off IPA-RA if
15195 patchable function entries are being generated.
15196 * varasm.c (assemble_start_function): Look at the
15197 patchable-function-entry command line switch and current
15198 function attributes and maybe generate NOP instructions by
15199 calling the print_patchable_function_entry hook.
15200 * doc/extend.texi: Document patchable_function_entry attribute.
15201 * doc/invoke.texi: Document -fpatchable_function_entry
15202 command line option.
15203 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15205 * doc/tm.texi: Re-generate.
15207 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15210 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15211 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15213 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15215 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15217 2017-07-25 Richard Biener <rguenther@suse.de>
15219 PR tree-optimization/81455
15220 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15221 not walk in cycles when looking for guards.
15223 2017-07-25 Richard Biener <rguenther@suse.de>
15225 PR tree-optimization/81529
15226 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15227 when optimizing backedge uses.
15229 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15231 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15233 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15234 to dl_section_ref. On AIX, append an expression to subtract
15235 the size of the section length to dl_section_ref.
15237 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15239 * configure.ac: If any of the config.* scripts fail, exit 1.
15240 * configure: Regenerate.
15242 2017-07-25 Richard Biener <rguenther@suse.de>
15244 PR middle-end/81546
15245 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15246 of immediate uses, be more verbose on errors.
15248 2017-07-25 Richard Biener <rguenther@suse.de>
15250 PR tree-optimization/81510
15251 * tree-vect-loop.c (vect_is_simple_reduction): When the
15252 reduction stmt is not inside the loop bail out.
15254 2017-07-25 Richard Biener <rguenther@suse.de>
15256 PR tree-optimization/81303
15257 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15258 profitability check against LOOP_VINFO_NITERSM1.
15260 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15262 * domwalk.c (cmp_bb_postorder): Simplify.
15263 (sort_bbs_postorder): New function. Use it...
15264 (dom_walker::walk): ...here to optimize common cases.
15266 2017-07-25 Martin Liska <mliska@suse.cz>
15269 * ipa-visibility.c (function_and_variable_visibility): Make the
15270 redirection just on target that supports aliasing.
15271 Fix GNU coding style.
15273 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15276 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15278 * config/arm/rtems.h: Likewise.
15279 * config/bfin/rtems.h: Likewise.
15280 * config/i386/rtemself.h: Likewise.
15281 * config/lm32/rtems.h: Likewise.
15282 * config/m32c/rtems.h: Likewise.
15283 * config/m68k/rtemself.h: Likewise.
15284 * config/microblaze/rtems.h: Likewise.
15285 * config/mips/rtems.h: Likewise.
15286 * config/moxie/rtems.h: Likewise.
15287 * config/nios2/rtems.h: Likewise.
15288 * config/powerpcspe/rtems.h: Likewise.
15289 * config/rs6000/rtems.h: Likewise.
15290 * config/rtems.h: Likewise.
15291 * config/sh/rtems.h: Likewise.
15292 * config/sh/rtemself.h: Likewise.
15293 * config/sparc/rtemself.h: Likewise.
15295 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15298 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15299 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15300 * tree-ssa-structalias.c (alias_get_name): Same.
15302 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15305 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15306 instructions if no du chain is found.
15308 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15310 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15312 2017-07-25 Richard Biener <rguenther@suse.de>
15314 PR middle-end/81505
15315 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15318 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15320 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15321 upper-regs options.
15322 (ISA_2_7_MASKS_SERVER): Likewise.
15323 (ISA_3_0_MASKS_IEEE): Likewise.
15324 (OTHER_P8_VECTOR_MASKS): Likewise.
15325 (OTHER_VSX_VECTOR_MASKS): Likewise.
15326 (POWERPC_MASKS): Likewise.
15327 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15328 duplicate list of options.
15329 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15330 explicit -mupper-regs options.
15331 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15332 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15333 alias for -mupper-regs-df.
15334 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15335 (rs6000_init_hard_regno_mode_ok): Likewise.
15336 (rs6000_option_override_internal): Likewise.
15337 (rs6000_opt_masks): Likewise.
15338 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15339 options in terms of whether -mvsx or -mpower8-vector was used.
15340 (TARGET_UPPER_REGS_DI): Likewise.
15341 (TARGET_UPPER_REGS_SF): Likewise.
15342 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15343 -mupper-regs-* options.
15345 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15347 * passes.c (emergency_dump_function): Print some empty lines and a
15348 header before the RTL dump.
15350 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15352 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15354 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15357 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15358 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15360 2017-07-24 Carl Love <cel@us.ibm.com>
15362 * config/rs6000/rs6000-c.c: Add support for built-in functions
15363 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15364 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15365 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15366 vec_extract_fp_from_shortl): Add defines for the two builtins.
15367 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15368 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15370 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15371 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15372 * doc/extend.texi: Update the built-in documentation file for the
15373 new built-in function.
15375 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15378 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15380 * doc/generic.texi: Likewise.
15381 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15382 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15384 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15386 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15387 (aarch64_mls_elt_merge<mode>): Likewise.
15389 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15391 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15392 having __cxa_atexit.
15394 2017-07-23 Michael Collison <michael.collison@arm.com>
15396 * config/arm/arm.c (arm_option_override): Deprecate
15397 use of -mstructure-size-boundary.
15398 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15399 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15401 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15404 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15405 Reduce cost estimate for direct moves.
15407 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15410 * config/i386/i386.c (ix86_option_override_internal): Disable
15411 BMI, BMI2 and TBM instructions for -m16.
15413 2017-07-21 Carl Love <cel@us.ibm.com>
15415 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15416 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15417 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15418 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15419 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15420 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15421 VMULOSW): New enum "unspec" values.
15422 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15423 altivec_vmulosw): New patterns.
15424 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15425 VMULOSW): Add definitions.
15427 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15429 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15430 (qdf24xx): Likewise.
15431 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15432 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15433 (AARCH64_FL_V8_1): Renumber.
15434 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15435 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15436 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15437 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15438 rdma to feature modifiers list.
15440 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15442 PR middle-end/56727
15443 * ipa-visibility (function_and_variable_visibility): Convert
15444 recursive PLT call to direct call if appropriate.
15446 2017-07-21 Andrew Pinski <apinski@cavium.com>
15448 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15449 operand 1 to see if the types precision matches.
15450 * fold-const.c (operand_equal_p): Likewise.
15452 2017-07-21 Richard Biener <rguenther@suse.de>
15454 PR tree-optimization/81303
15455 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15456 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15457 (vect_peeling_hash_get_lowest_cost): Adjust.
15458 (vect_enhance_data_refs_alignment): Likewise. Use
15459 vect_get_peeling_costs_all_drs to compute the penalty for no
15460 peeling to match up costs.
15462 2017-07-21 Richard Biener <rguenther@suse.de>
15464 PR tree-optimization/81500
15465 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15466 we didn't identify a reduction path.
15468 2017-07-21 Tom de Vries <tom@codesourcery.com>
15469 Cesar Philippidis <cesar@codesourcery.com>
15471 PR gcov-profile/81442
15472 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15475 2017-07-21 Tom de Vries <tom@codesourcery.com>
15478 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15480 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15481 nvptx_override_options_after_change.
15483 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15485 * dwarf2out.c (output_file_names): Avoid double testing for
15486 dwarf_version >= 5.
15488 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15490 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15492 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15494 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15496 (try_crossjump_to_edge): Do not punt on partitioned functions.
15498 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15500 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15501 Put all BBs reachable only via paths crossing cold region to cold
15503 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15505 2016-07-21 Richard Biener <rguenther@suse.de>
15507 PR tree-optimization/81303
15508 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15509 into account prologue and epilogue iterations when raising
15510 min_profitable_iters to sth at least covering one vector iteration.
15512 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15514 * config/arm/arm.c (arm_test_cpu_arch_dat):
15517 2017-07-20 Nathan Sidwell <nathan@acm.org>
15519 Remove TYPE_METHODS.
15520 * tree.h (TYPE_METHODS): Delete.
15521 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15522 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15523 (dbxout_type_methods): Scan TYPE_FIELDS.
15524 (dbxout_type): Don't check TYPE_METHODS here.
15525 * function.c (use_register_for_decl): Always ignore register for
15526 class types when not optimizing.
15527 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15528 * tree.c (free_lang_data_in_type): Stitch out member functions and
15529 templates from TYPE_FIELDS.
15530 (build_distinct_type_copy, verify_type_variant,
15531 verify_type): Member fns are on TYPE_FIELDS.
15532 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15533 * tree-pretty-print.c (dump_generic_node): Likewise.
15535 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15538 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15539 V2TImode and V4TImode.
15540 (ix86_expand_vector_extract): Likewise.
15541 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15542 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15543 (ssescalarmode): Handle V4TImode and V2TImode.
15544 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15545 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15546 (VEXTRACTI128_MODE): New mode iterator.
15547 (splitter for *vec_extractv?ti first element): New.
15548 (VEC_INIT_MODE): New mode iterator.
15549 (vec_init<mode>): Consolidate 3 expanders into one using
15550 VEC_INIT_MODE mode iterator.
15552 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15554 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15555 non_spilled_static_chain_regno_p.
15557 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15559 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15561 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15563 * bb-reorder.c (connect_traces): Allow copying of blocks within
15566 2017-07-20 Richard Biener <rguenther@suse.de>
15568 * gimple.h (gimple_phi_result): Add gphi * overload.
15569 (gimple_phi_result_ptr): Likewise.
15570 (gimple_phi_arg): Likewise. Adjust index assert to only
15571 allow actual argument accesses rather than all slots available
15573 (gimple_phi_arg_def): Add gphi * overload.
15574 * tree-phinodes.c (make_phi_node): Initialize only actual
15576 (resize_phi_node): Clear memory not covered by old node,
15577 do not initialize excess argument slots.
15578 (reserve_phi_args_for_new_edge): Initialize new argument slot
15581 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15583 PR tree-optimization/81388
15585 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15587 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15588 by removing computation of may_be_zero.
15590 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15591 Tom de Vries <tom@codesourcery.com>
15593 PR middle-end/81030
15594 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15595 when gimple level profile disagrees with what RTL expander did.
15597 2017-07-20 Richard Biener <rguenther@suse.de>
15599 PR tree-optimization/61171
15600 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
15601 (vect_analyze_stmt): Add slp instance parameter.
15602 (vectorizable_reduction): Likewise.
15603 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
15604 (vect_is_simple_reduction): Deal with chains not detected
15605 as SLP reduction chain, specifically not properly associated
15606 chains containing a mix of plus/minus.
15607 (get_reduction_op): Remove.
15608 (get_initial_defs_for_reduction): Simplify, pass in whether
15609 this is a reduction chain, pass in the SLP node for the PHIs.
15610 (vect_create_epilog_for_reduction): Get the SLP instance as
15612 (vectorizable_reduction): Get the SLP instance as arg.
15613 During analysis remember the SLP node with the PHIs in the
15614 instance. Simplify getting at the vectorized reduction PHIs.
15615 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15616 through SLP instance.
15617 (vect_slp_analyze_operations): Likewise.
15618 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15619 (vect_transform_stmt): Likewise.
15621 2017-07-20 Tom de Vries <tom@codesourcery.com>
15623 PR tree-optimization/81489
15624 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15625 read of phi arg location to before loop that modifies phi.
15627 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15629 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15632 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15634 PR middle-end/81331
15635 * except.c (execute): Fix ordering issue.
15637 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15639 PR rtl-optimization/81423
15640 * combine.c (make_compound_operation_int): Don't try to optimize
15641 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15643 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15645 PR rtl-optimization/81423
15646 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15647 with a constant that is -1 in the truncated to mode.
15649 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15651 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15652 (determine_unlikely_bbs): ... here.
15653 * predict.h (propagate_unlikely_bbs_forward): Declare.
15654 * cfgexpand.c (pass_expand::execute): Use it.
15655 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15657 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15658 propagate_unlikely_bbs_forward.
15660 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15662 PR middle-end/81331
15663 * except.c (maybe_add_nop_after_section_switch): New function.
15666 2017-07-19 Tom de Vries <tom@codesourcery.com>
15668 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15670 2017-07-19 Tom de Vries <tom@codesourcery.com>
15672 * gimple.h (gimple_phi_arg): Make assert more strict.
15674 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15676 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15677 * config/rs6000/mmintrin.h: New file.
15678 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15680 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15682 PR tree-optimization/81346
15683 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15685 2017-07-19 Tom de Vries <tom@codesourcery.com>
15687 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15689 2017-07-19 Tom de Vries <tom@codesourcery.com>
15691 * config/nvptx/nvptx-modes.def: Add V2DImode.
15692 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15693 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15694 (nvptx_output_mov_insn): Handle lack of mov.b128.
15695 (nvptx_print_operand): Handle 'H' and 'L' codes.
15696 (nvptx_vector_mode_supported): Allow V2DImode.
15697 (nvptx_preferred_simd_mode): New function.
15698 (nvptx_data_alignment): New function.
15699 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15700 nvptx_preferred_simd_mode.
15701 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15703 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15705 2017-07-19 Tom de Vries <tom@codesourcery.com>
15707 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15708 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15709 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15710 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15711 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15712 (mov<VECIM>_insn): New define_insn.
15713 (define_expand "mov<VECIM>): New define_expand.
15715 2017-07-19 Tom de Vries <tom@codesourcery.com>
15717 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15719 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15721 PR tree-optimization/81346
15722 * fold-const.h (fold_div_compare, range_check_type): Declare.
15723 * fold-const.c (range_check_type): New function.
15724 (build_range_check): Use range_check_type.
15725 (fold_div_compare): No longer static, rewritten into
15726 a match.pd helper function.
15727 (fold_comparison): Don't call fold_div_compare here.
15728 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15729 as helper function.
15731 2017-07-19 Nathan Sidwell <nathan@acm.org>
15733 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15734 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15735 * tree.c (find_decls_types_r, verify_type): Use
15736 TYPE_{MIN,MAX}_VALUE_RAW.
15737 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15738 (hash_tree): Likewise.
15739 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15741 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15744 2017-07-18 Tom de Vries <tom@codesourcery.com>
15746 PR middle-end/81464
15747 * omp-expand.c (expand_omp_for_static_chunk): Handle
15748 equal-argument loop exit phi.
15750 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15753 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15754 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15755 operand 2 predicate.
15756 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15757 operand 2 predicate.
15758 (ror,rol -> rorx splitters): Use const_int_operand as
15759 operand 2 predicate.
15761 2017-06-18 Richard Biener <rguenther@suse.de>
15763 PR tree-optimization/81410
15764 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15765 the gap in the ! slp_perm SLP case after each group.
15767 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15769 PR middle-end/81463
15770 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15773 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15775 PR middle-end/81462
15776 * predict.c (set_even_probabilities): Cleanup; do not affect
15777 probabilities that are already known.
15778 (combine_predictions_for_bb): Call even when count is set.
15780 2017-07-18 Nathan Sidwell <nathan@acm.org>
15782 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15785 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15788 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15789 optimization for loop niter analysis.
15791 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15794 * config/avr/avr.c (avr_optimize_casesi): Don't use
15795 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15797 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15799 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15800 body_cost_vec from _vect_peel_extended_info.
15801 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15802 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15805 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15807 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15809 2017-07-18 Richard Biener <rguenther@suse.de>
15811 PR tree-optimization/80620
15812 PR tree-optimization/81403
15813 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15814 info when re-using a VN table entry.
15816 2017-07-18 Richard Biener <rguenther@suse.de>
15818 PR tree-optimization/81418
15819 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15820 vectype_in. Verify that with lane-reducing reduction operations
15821 we have a single def-use cycle.
15823 2017-07-17 Carl Love <cel@us.ibm.com>
15825 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15827 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15828 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15829 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15830 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15831 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15832 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15833 VMULOSW): New enum "unspec" values.
15834 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15835 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15836 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15837 altivec_vmulosw): New patterns.
15838 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15839 VMULOSW): Add definitions.
15840 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15842 * config/alpha/alpha.c: Include predict.h.
15844 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15846 * tree-vrp.c (compare_assert_loc): Fix comparison function
15847 to return predictable results.
15849 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15851 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15853 (subdi3): Likewise.
15854 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15855 * doc/invoke.texi (mexpand-adddi): Update text.
15857 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15859 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15860 that also clobbers the CC register. The old expand code is moved
15862 (*arc_clzsi2): ... here.
15863 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15864 the CC register. The old expand code is moved to ...
15865 (arc_ctzsi2): ... here.
15867 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15869 * config/arc/arc.opt (mindexed-loads): Use initial value
15870 TARGET_INDEXED_LOADS_DEFAULT.
15871 (mauto-modify-reg): Use initial value
15872 TARGET_AUTO_MODIFY_REG_DEFAULT.
15873 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15874 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15875 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15876 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15878 2017-07-17 Martin Liska <mliska@suse.cz>
15881 * opts.c (finish_options): Do not allow -fgnu-tm
15882 w/ -fsanitize={kernel-,}address. Say sorry.
15884 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15887 * tree-loop-distribution.c (classify_partition): Only assert on
15888 numer of iterations.
15889 (merge_dep_scc_partitions): Delete prameter. Update function call.
15890 (distribute_loop): Remove code handling loop with unknown niters.
15891 (pass_loop_distribution::execute): Skip loop with unknown niters.
15893 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15896 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15897 function sort_partitions_by_post_order.
15899 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15901 PR tree-optimization/81374
15902 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15903 the max index of basic blocks, rather than number of basic blocks.
15905 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15907 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15909 (arc_legitimate_pic_operand_p): Likewise.
15910 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
15912 (arc_needs_pcl_p): Likewise.
15913 (arc_legitimate_pc_offset_p): Likewise.
15914 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
15915 function is also used in constrains.md.
15916 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
15917 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
15918 PLUS. Only return true/false in known cases, otherwise assert.
15919 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
15920 is already called in arc_legitimate_constant_p.
15921 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
15923 (LEGITIMATE_PIC_OPERAND_P): Use
15924 arc_raw_symbolic_reference_mentioned_p function.
15925 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
15930 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15931 Andrew Burgess <andrew.burgess@embecosm.com>
15933 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
15934 (arc_return_address_register): New function.
15935 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
15936 (arc_handle_fndecl_attribute): Add naked attribute.
15937 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15938 (TARGET_WARN_FUNC_RETURN): Likewise.
15939 (arc_allocate_stack_slots_for_args): New function.
15940 (arc_warn_func_return): Likewise.
15941 (machine_function): Change type fn_type.
15942 (arc_compute_function_type): Consider new naked function type,
15943 change function return type.
15944 (arc_must_save_register): Adapt to handle new
15945 arc_compute_function_type's return type.
15946 (arc_expand_prologue): Likewise.
15947 (arc_expand_epilogue): Likewise.
15948 (arc_return_address_regs): Delete.
15949 (arc_return_address_register): New function.
15950 (arc_epilogue_uses): Use above function.
15951 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
15952 (arc_function_type): Change encoding, add naked type.
15953 (ARC_INTERRUPT_P): Change to handle the new encoding.
15954 (ARC_FAST_INTERRUPT_P): Likewise.
15955 (ARC_NORMAL_P): Define.
15956 (ARC_NAKED_P): Likewise.
15957 (arc_compute_function_type): Delete prototype.
15958 * config/arc/arc.md (in_ret_delay_slot): Use
15959 arc_return_address_register function.
15960 (simple_return): Likewise.
15961 (p_return_i): Likewise.
15963 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15965 PR tree-optimization/81428
15966 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
15967 can't be built for those types.
15969 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15971 Remove stuff dead since r239246.
15973 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
15974 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
15975 (avr_inform_devices): Remove dead stuff.
15977 2017-07-17 Tamar Christina <tamar.christina@arm.com>
15979 * config/arm/arm_neon.h: Fix softp typo.
15981 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15983 PR tree-optimization/81365
15984 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
15985 aggregate moves onto bb predecessor edges, make sure there are no
15986 loads that could alias the lhs in between the start of bb and the
15989 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
15992 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
15993 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
15994 [LSHIFTRT, outer_code = TRUNCATE]: Same.
15996 2017-07-17 Jakub Jelinek <jakub@redhat.com>
15998 PR tree-optimization/81396
15999 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16000 (init_symbolic_number): Initialize it to 1.
16001 (perform_symbolic_merge): Add n_ops from both operands into the new
16003 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16004 without base_addr as useless if they need more than one operation.
16005 (bswap_replace): Handle !bswap case for NULL base_addr.
16007 2017-07-17 Tom de Vries <tom@codesourcery.com>
16010 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16013 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16015 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16016 conditional builtin define __FIX_LEON3FT_B2BST.
16018 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16020 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16021 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16024 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16026 PR rtl-optimization/81424
16027 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16028 to remove potential trapping from operands if -fnon-call-exceptions.
16030 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16032 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16033 profile_proability for scalling.
16034 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16036 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16038 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16040 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16042 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16043 fixpoint arithmetics.
16045 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16047 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16048 fixpoint arithmetics.
16050 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16052 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16053 fixpoint arithmetics.
16055 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16057 * profile-count.h (profile_probability::from_reg_br_prob_note,
16058 profile_probability::to_reg_br_prob_note): New functions.
16059 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16060 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16061 * predict.c (probability_reliable_p): Update.
16062 (edge_probability_reliable_p): Update.
16063 (br_prob_note_reliable_p): Update.
16064 (invert_br_probabilities): Update.
16065 (add_reg_br_prob_note): New function.
16066 (combine_predictions_for_insn): Update.
16067 * asan.c (asan_clear_shadow): Update.
16068 * cfgbuild.c (compute_outgoing_frequencies): Update.
16069 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16070 (update_br_prob_note): Update.
16071 (rtl_verify_edges): Update.
16072 (purge_dead_edges): Update.
16073 (fixup_reorder_chain): Update.
16074 * emit-rtl.c (try_split): Update.
16075 * ifcvt.c (cond_exec_process_insns): Update.
16076 (cond_exec_process_if_block): Update.
16077 (dead_or_predicable): Update.
16078 * internal-fn.c (expand_addsub_overflow): Update.
16079 (expand_neg_overflow): Update.
16080 (expand_mul_overflow): Update.
16081 * loop-doloop.c (doloop_modify): Update.
16082 * loop-unroll.c (compare_and_jump_seq): Update.
16083 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16084 * predict.h: Update.
16085 * reorg.c (mostly_true_jump): Update.
16087 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16088 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16089 * config/arc/arc.c: (emit_unlikely_jump): Update.
16090 * config/arm/arm.c: (emit_unlikely_jump): Update.
16091 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16092 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16093 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16094 (ix86_print_operand): Update.
16095 (ix86_split_fp_branch): Update.
16096 (predict_jump): Update.
16097 * config/ia64/ia64.c (ia64_print_operand): Update.
16098 * config/mmix/mmix.c (mmix_print_operand): Update.
16099 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
16100 (rs6000_expand_split_stack_prologue): Update.
16101 * config/rs6000/rs6000.c: Update.
16102 * config/s390/s390.c (s390_expand_vec_strlen): Update.
16103 (s390_expand_vec_movstr): Update.
16104 (s390_expand_cs_tdsi): Update.
16105 (s390_expand_split_stack_prologue): Update.
16106 * config/sh/sh.c (sh_print_operand): Update.
16107 (expand_cbranchsi4): Update.
16108 (expand_cbranchdi4): Update.
16109 * config/sparc/sparc.c (output_v9branch): Update.
16110 * config/spu/spu.c (get_branch_target): Update.
16111 (ea_load_store_inline): Update.
16112 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
16113 * config/tilepro/tilepro.c: Update.
16115 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16117 * gimplify.c (mostly_copy_tree_r): Revert latest change.
16118 (gimplify_save_expr): Likewise.
16120 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16122 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
16124 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16126 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
16128 * timevar.def (TV_IPA_FNSUMMARY): Define.
16130 2017-07-16 Daniel Cederman <cederman@gaisler.com>
16132 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
16133 to back store errata sensitive sequence from being generated.
16134 (sqrtdf2_fix): Likewise.
16136 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16138 * tree-ssa-threadupdate.c (compute_path_counts,
16139 update_joiner_offpath_counts): Use profile_probability.
16141 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
16144 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16146 * config/arm/arm-c.c (arm_cpu_builtins): Define
16147 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16149 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16151 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16152 array entries to represent __ieee128 versions of the
16153 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
16154 scalar_extract_sig, and scalar_insert_exp built-in functions.
16155 (altivec_resolve_overloaded_builtin): Add special case handling
16156 for the __builtin_scalar_insert_exp function, as represented by
16157 the P9V_BUILTIN_VEC_VSIEDP constant.
16158 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
16159 exponent support for __ieee128 argument.
16160 (VSESQP): Add scalar extract signature support for __ieee128
16162 (VSTDCNQP): Add scalar test negative support for __ieee128
16164 (VSIEQP): Add scalar insert exponent support for __int128 argument
16165 with __ieee128 result.
16166 (VSIEQPF): Add scalar insert exponent support for __ieee128
16167 argument with __ieee128 result.
16168 (VSTDCQP): Add scalar test data class support for __ieee128
16170 (VSTDCNQP): Add overload support for scalar test negative with
16171 __ieee128 argument.
16172 (VSTDCQP): Add overload support for scalar test data class
16173 __ieee128 argument.
16174 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16175 UNSPEC_VSX_SXSIGDP.
16176 (UNSPEC_VSX_SIEXPQP): New constant.
16177 (xsxexpqp): New insn for VSX scalar extract exponent quad
16179 (xsxsigqp): New insn for VSX scalar extract significand quad
16181 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16182 precision with floating point argument.
16183 (xststdcqp): New expand for VSX scalar test data class quad
16185 (xststdcnegqp): New expand for VSX scalar test negative quad
16187 (xststdcqp): New insn to match expansions for VSX scalar test data
16188 class quad precision and VSX scalar test negative quad precision.
16189 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16190 special case operand checking to enforce that second operand of
16191 VSX scalar test data class with quad precision argument is a 7-bit
16193 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16194 prototypes and descriptions of __ieee128 versions of
16195 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16196 scalar_test_data_class, and scalar_test_neg built-in functions.
16198 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16200 PR tree-optimization/81162
16201 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16202 replace a negate with an add.
16204 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16206 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16208 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16210 * config/arm/arm-c.c (arm_cpu_builtins): Define
16211 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16213 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16215 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16216 (armv8-r): Set ARM Cortex-R52 as default CPU.
16217 * config/arm/arm-tables.opt: Regenerate.
16218 * config/arm/arm-tune.md: Regenerate.
16219 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16221 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16222 extension for -mcpu=cortex-r52.
16224 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16226 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16227 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16228 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16229 (fp-armv8): Define it as FP_ARMv8 only.
16230 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16231 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16233 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16234 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16235 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16236 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16237 than TARGET_FPU_ARMV8.
16238 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16239 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16240 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16242 * config/arm/neon.md (neon_vrint): Likewise.
16243 (neon_vcvt): Likewise.
16244 (neon_<fmaxmin_op><mode>): Likewise.
16245 (<fmaxmin><mode>3): Likewise.
16246 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16247 * config/arm/predicates.md (arm_cond_move_operator): Check against
16248 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16250 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16252 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16253 to top of function.
16255 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16257 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16258 loop in comment with memset.
16260 2017-07-14 Martin Liska <mliska@suse.cz>
16262 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16263 * dwarf2out.c (is_java): Remove the function.
16264 (output_pubname): Remove usage of the function.
16265 (lower_bound_default): Remove usage of DW_LANG_Java.
16266 (gen_compile_unit_die): Likewise.
16267 * gcc.c: Remove compiler defaults for .java and .zip files.
16268 * gimple-expr.c (remove_suffix): Change as there's no longer
16269 extension than 4-letter one.
16270 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16271 (gimplify_save_expr): Likewise.
16272 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16273 as it's possible even for other languages than Java.
16274 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16275 * lto-opts.c (lto_write_options): Remove reference to Java.
16276 * opts.c (strip_off_ending): Update file extension handling.
16277 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16278 * tree-eh.c (lower_resx): Likewise.
16279 * tree.c (free_lang_data_in_type): Remove dead code.
16280 (find_decls_types_r): Likewise.
16281 (build_common_builtin_nodes): Remove Java from a comment.
16282 (verify_type): Remove dead code.
16283 * varasm.c (assemble_external): Remove Java from a comment.
16285 2017-07-14 Martin Liska <mliska@suse.cz>
16287 * opts.c (finish_options): Add quotes.
16288 (common_handle_option): Likewise.
16290 2017-07-14 Martin Liska <mliska@suse.cz>
16292 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16293 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16294 Remove N_SO_PASCAL.
16295 * dwarf2out.c (lower_bound_default): Do not handle
16297 (gen_compile_unit_die): Likewise.
16298 * gcc.c: Remove default extension binding for GNU Pascal.
16299 * stmt.c: Remove Pascal language from a comment.
16300 * xcoffout.c: Likewise.
16302 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16305 * diagnostic-show-locus.c (fixit_cmp): New function.
16306 (layout::layout): Sort m_fixit_hints.
16307 (column_range::column_range): Assert that the values are valid.
16308 (struct char_span): New struct.
16309 (correction::overwrite): New method.
16310 (struct source_line): New struct.
16311 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16312 calls in terms of classes source_line and char_span, and
16313 correction::overwrite.
16314 (selftest::test_overlapped_fixit_printing_2): New function.
16315 (selftest::diagnostic_show_locus_c_tests): Call it.
16317 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16319 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16320 early if there is no lhs.
16322 2017-07-13 Martin Liska <mliska@suse.cz>
16324 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16325 (gen_reference_type_die): Likewise.
16326 * stor-layout.c: Remove Pascal-related comment.
16328 2017-07-13 Martin Liska <mliska@suse.cz>
16330 * opts.c (finish_options): Add quotes to error messages.
16331 (parse_sanitizer_options): Likewise.
16333 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16335 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16337 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16339 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16341 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16343 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16345 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16347 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16350 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16351 provides the hardware capability bits, define the macro
16352 __BUILTIN_CPU_SUPPORTS__.
16353 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16354 if GLIBC does not provide the hardware capability bits. Add a
16355 gcc_unreachable call if the built-in cpu function is neither
16356 __builtin_cpu_is nor __builtin_cpu_supports.
16357 (rs6000_get_function_versions_dispatcher): Change the warning
16358 that an old GLIBC is used which does not export the capability
16359 bits to be an error.
16360 * doc/extend.texi (target_clones attribute): Document the
16361 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16362 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16363 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16364 the macros defined by GCC if the newer GLIBC is available.
16366 2017-07-12 Jeff Law <law@redhat.com>
16368 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16369 remaining includes slightly.
16370 * config/riscv/riscv-builtins.c: Include profile-count.h.
16372 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16375 * config/avr/avr.c (avr_set_current_function): In diagnostic
16376 messages: Quote keywords and (parts of) identifiers.
16377 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16380 2017-07-12 Carl Love <cel@us.ibm.com>
16382 * config/rs6000/rs6000-c.c: Add support for built-in functions
16383 vector bool char vec_revb (vector bool char);
16384 vector bool short vec_revb (vector short char);
16385 vector bool int vec_revb (vector bool int);
16386 vector bool long long vec_revb (vector bool long long);
16387 * doc/extend.texi: Update the built-in documentation file for the
16388 new built-in functions.
16390 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16392 * config/s390/s390.md: Remove movcc splitter.
16394 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16396 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16397 load/store on condition.
16399 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16402 * config/avr/avr.c (avr_encode_section_info)
16403 [progmem && !TREE_READONLY]: Error if progmem object needs
16406 2017-07-11 Michael Collison <michael.collison@arm.com>
16408 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16411 2017-07-11 Carl Love <cel@us.ibm.com>
16413 * config/rs6000/rs6000-c.c: Add support for builtins
16414 vector unsigned int vec_parity_lsbb (vector signed int);
16415 vector unsigned int vec_parity_lsbb (vector unsigned int);
16416 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16417 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16418 vector unsigned long long vec_parity_lsbb (vector signed long long);
16419 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16420 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16421 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16422 * doc/extend.texi: Update the built-in documentation file for the
16423 new built-in functions.
16425 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16427 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16428 (layout::m_primary_loc): New field.
16429 (layout::layout): Initialize new field. Move location filtering
16430 logic from here to...
16431 (layout::maybe_add_location_range): ...this new method. Add
16432 support for filtering to just the lines already specified by other
16434 (layout::will_show_line_p): New method.
16435 (gcc_rich_location::add_location_if_nearby): New method.
16436 (selftest::test_add_location_if_nearby): New test function.
16437 (selftest::diagnostic_show_locus_c_tests): Call it.
16438 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16441 2017-07-11 Tom de Vries <tom@codesourcery.com>
16443 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16444 (bb_first_real_insn): New function.
16445 (nvptx_single): Add extra initialization of broadcasted condition
16448 2017-07-11 Nathan Sidwell <nathan@acm.org>
16450 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16452 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16454 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16455 Binutils doc as TEXI will mess them up.
16456 * doc/invoke.texi (AVR Options): Same here.
16458 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16460 * config/sparc/sparc.opt (mfix-ut700): New option.
16461 (mfix-gr712rc): Likewise.
16462 (sparc_fix_b2bst): New variable.
16463 * doc/invoke.texi (SPARC options): Document them.
16464 (ARM options): Fix warnings.
16465 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16466 instructions to prevent sequences that can trigger the store-store
16467 errata for certain LEON3FT processors.
16468 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16469 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16470 * config/sparc/sparc.md (fix_b2bst): New attribute.
16471 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16473 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16476 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16478 (*rsqrtsf2_sse): Ditto.
16480 (div<mode>3): Macroize insn from divdf3 and divsf3
16481 using MODEF mode iterator.
16483 2017-07-10 Martin Sebor <msebor@redhat.com>
16485 PR tree-optimization/80397
16486 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16487 instead of testing for equality to INTEGER_TYPE.
16489 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16491 * config.gcc: Remove uclibc from arc target spec.
16493 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16495 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16497 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16500 * lto-wrapper.c (remove_option): New function.
16501 (merge_and_complain): Merge PIC/PIE options more realistically.
16503 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16505 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16509 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16510 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16511 * config.in: Regenerate.
16512 * configure: Regenerate.
16513 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16514 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16515 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16516 (TARGET_GASISR_PROLOGUES): ...target mask.
16517 * common/config/avr/avr-common.c
16518 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16519 Set -mgas-isr-prologues.
16520 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16521 INSERT_PASS_BEFORE for it.
16522 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16523 * config/avr/avr.c (avr_option_override)
16524 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16525 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16526 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16527 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16528 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16530 (make_avr_pass_pre_proep): New function.
16531 (emit_push_sfr) <treg>: Add argument to function and use it
16532 instead of TMP_REG.
16533 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16534 and set machine->gasisr.yes.
16535 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16536 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16537 __gcc_isr.n_pushed to .L__stack_usage.
16538 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16539 (avr_asm_final_postscan_insn): ...this new static function.
16540 * config/avr/avr.h (machine_function)
16541 <is_no_gccisr, use_L__stack_usage>: New fields.
16542 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16543 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16544 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16545 (gasisr, *gasisr): New expander and insn.
16546 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16547 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16548 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16550 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16552 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16555 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16557 Move jump-tables out of .text again.
16560 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16561 (ASM_OUTPUT_ADDR_VEC): New function.
16562 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16563 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16564 INSN_ADDRESSes as asm comment.
16565 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16566 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16567 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16568 * config/avr/avr.md (*tablejump): Adjust comment.
16569 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16570 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16572 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16573 (avr_output_addr_vec): New proto.
16574 (avr_log_t) <insn_addresses>: New field.
16576 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16579 * config/i386/i386.c (ix86_function_arg_advance): Set
16580 outgoing_args_on_stack to true if there are outgoing arguments
16582 (ix86_function_arg): Likewise.
16583 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16584 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16585 * config/i386/i386.h (machine_function): Add
16586 outgoing_args_on_stack.
16588 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16590 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16591 supporting pthreds.
16592 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16594 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16596 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16597 (REAL_H): Remove $(MACHMODE_H).
16598 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
16600 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
16601 $(MACHMODE_H) and double-int.h.
16602 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
16604 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
16607 2017-07-07 Andrew Pinski <apinski@cavium.com>
16609 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
16610 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
16612 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16614 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16615 Add warning if GCC was not configured to link against a GLIBC that
16616 exports the hardware capability bits.
16617 (make_resolver_func): Make resolver function private and not a
16618 COMDAT function. Create the name with clone_function_name instead
16619 of make_unique_name.
16622 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16623 correct operand in doing the split.
16625 2017-07-07 Carl Love <cel@us.ibm.com>
16627 * config/rs6000/rs6000-c: Add support for built-in function
16628 vector unsigned short vec_pack_to_short_fp32 (vector float,
16630 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16631 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16632 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16633 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16634 (convert_4f32_8i16): Add define_expand.
16635 * doc/extend.texi: Update the built-in documentation file for the
16636 new built-in function.
16638 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16640 * config/sparc/m8.md: New file.
16641 * config/sparc/sparc.md: Include m8.md.
16643 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16645 * config/sparc/sparc.opt: New option -mvis4b.
16646 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16647 (sparc_option_override): Handle VIS4B.
16648 (enum sparc_builtins): Define
16649 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16650 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16651 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16652 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16653 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16654 (check_constant_argument): New function.
16655 (sparc_vis_init_builtins): Define builtins
16656 __builtin_vis_dictunpack{8,16,32},
16657 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16658 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16659 __builtin_vis_fpcmpde{8,16,32}shl and
16660 __builtin_vis_fpcmpur{8,16,32}shl.
16661 (sparc_expand_builtin): Check that the constant operands to
16662 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16663 constant and in range.
16664 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16666 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16667 (SPARC_IMM5_P): Likewise.
16668 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16669 (enabled): Handle vis4b.
16670 (UNSPEC_DICTUNPACK): New unspec.
16671 (UNSPEC_FPCMPSHL): Likewise.
16672 (UNSPEC_FPUCMPSHL): Likewise.
16673 (UNSPEC_FPCMPDESHL): Likewise.
16674 (UNSPEC_FPCMPURSHL): Likewise.
16675 (cpu_feature): New CPU feature `vis4b'.
16676 (dictunpack{8,16,32}): New insns.
16677 (FPCSMODE): New mode iterator.
16678 (fpcscond): New code iterator.
16679 (fpcsucond): Likewise.
16680 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16681 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16682 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16683 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16684 * config/sparc/constraints.md: Define constraints `q' for unsigned
16685 2-bit integer constants and `t' for unsigned 5-bit integer
16687 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16689 (imm5_operand_dictunpack16): Likewise.
16690 (imm5_operand_dictunpack32): Likewise.
16691 (imm2_operand): Likewise.
16692 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16693 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16694 ditunpack* and fpcmp*shl builtins.
16696 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16698 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16699 * config.in: Add HAVE_AS_SPARC6 define.
16700 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16702 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16704 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16705 (CPP_CPU_SPEC): Handle m8.
16706 (ASM_CPU_SPEC): Likewise.
16707 * config/sparc/sparc-opts.h (enum processor_type): Add
16709 * config/sparc/sparc.c (m8_costs): New struct.
16710 (sparc_option_override): Handle TARGET_CPU_m8.
16711 (sparc32_initialize_trampoline): Likewise.
16712 (sparc64_initialize_trampoline): Likewise.
16713 (sparc_issue_rate): Likewise.
16714 (sparc_register_move_cost): Likewise.
16715 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16716 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16717 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16718 (CPP_CPU_SPEC): Handle M8.
16719 (ASM_CPU_SPEC): Likewise.
16720 (AS_M8_FLAG): Define.
16721 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16722 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16723 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16725 * configure: Regenerate.
16726 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16729 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16731 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16733 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16734 ("*movdi_insn_sp32"): Do not set v3pipe.
16735 ("*movsi_insn"): Likewise.
16736 ("*movdi_insn_sp64"): Likewise.
16737 ("*movsf_insn"): Likewise.
16738 ("*movdf_insn_sp32"): Likewise.
16739 ("*movdf_insn_sp64"): Likewise.
16740 ("*zero_extendsidi2_insn_sp64"): Likewise.
16741 ("*sign_extendsidi2_insn"): Likewise.
16742 ("*mov<VM32:mode>_insn"): Likewise.
16743 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16744 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16745 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16746 ("<vlop:code><VL:mode>3"): Likewise.
16747 ("*not_<vlop:code><VL:mode>3"): Likewise.
16748 ("*nand<VL:mode>_vis"): Likewise.
16749 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16750 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16751 ("one_cmpl<VL:mode>2"): Likewise.
16752 ("faligndata<VM64:mode>_vis"): Likewise.
16753 ("alignaddrsi_vis"): Likewise.
16754 ("alignaddrdi_vis"): Likweise.
16755 ("alignaddrlsi_vis"): Likewise.
16756 ("alignaddrldi_vis"): Likewise.
16757 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16758 ("bmaskdi_vis"): Likewise.
16759 ("bmasksi_vis"): Likewise.
16760 ("bshuffle<VM64:mode>_vis"): Likewise.
16761 ("cmask8<P:mode>_vis"): Likewise.
16762 ("cmask16<P:mode>_vis"): Likewise.
16763 ("cmask32<P:mode>_vis"): Likewise.
16764 ("pdistn<P:mode>_vis"): Likewise.
16765 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16767 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16769 * config/sparc/sparc.md ("subtype"): New insn attribute.
16770 ("*wrgsr_sp64"): Set insn subtype.
16771 ("*rdgsr_sp64"): Likewise.
16772 ("alignaddrsi_vis"): Likewise.
16773 ("alignaddrdi_vis"): Likewise.
16774 ("alignaddrlsi_vis"): Likewise.
16775 ("alignaddrldi_vis"): Likewise.
16776 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16777 ("fexpand_vis"): Likewise.
16778 ("fpmerge_vis"): Likewise.
16779 ("faligndata<VM64:mode>_vis"): Likewise.
16780 ("bshuffle<VM64:mode>_vis"): Likewise.
16781 ("cmask8<P:mode>_vis"): Likewise.
16782 ("cmask16<P:mode>_vis"): Likewise.
16783 ("cmask32<P:mode>_vis"): Likewise.
16784 ("fchksm16_vis"): Likewise.
16785 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16786 ("fmean16_vis"): Likewise.
16787 ("fp<plusminus_insn>64_vis"): Likewise.
16788 ("<plusminus_insn>v8qi3"): Likewise.
16789 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16790 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16791 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16792 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16793 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16794 ("*movqi_insn"): Likewise.
16795 ("*movhi_insn"): Likewise.
16796 ("*movsi_insn"): Likewise.
16797 ("movsi_pic_gotdata_op"): Likewise.
16798 ("*movdi_insn_sp32"): Likewise.
16799 ("*movdi_insn_sp64"): Likewise.
16800 ("movdi_pic_gotdata_op"): Likewise.
16801 ("*movsf_insn"): Likewise.
16802 ("*movdf_insn_sp32"): Likewise.
16803 ("*movdf_insn_sp64"): Likewise.
16804 ("*zero_extendhisi2_insn"): Likewise.
16805 ("*zero_extendqihi2_insn"): Likewise.
16806 ("*zero_extendqisi2_insn"): Likewise.
16807 ("*zero_extendqidi2_insn"): Likewise.
16808 ("*zero_extendhidi2_insn"): Likewise.
16809 ("*zero_extendsidi2_insn_sp64"): Likewise.
16810 ("ldfsr"): Likewise.
16811 ("prefetch_64"): Likewise.
16812 ("prefetch_32"): Likewise.
16813 ("tie_ld32"): Likewise.
16814 ("tie_ld64"): Likewise.
16815 ("*tldo_ldub_sp32"): Likewise.
16816 ("*tldo_ldub1_sp32"): Likewise.
16817 ("*tldo_ldub2_sp32"): Likewise.
16818 ("*tldo_ldub_sp64"): Likewise.
16819 ("*tldo_ldub1_sp64"): Likewise.
16820 ("*tldo_ldub2_sp64"): Likewise.
16821 ("*tldo_ldub3_sp64"): Likewise.
16822 ("*tldo_lduh_sp32"): Likewise.
16823 ("*tldo_lduh1_sp32"): Likewise.
16824 ("*tldo_lduh_sp64"): Likewise.
16825 ("*tldo_lduh1_sp64"): Likewise.
16826 ("*tldo_lduh2_sp64"): Likewise.
16827 ("*tldo_lduw_sp32"): Likewise.
16828 ("*tldo_lduw_sp64"): Likewise.
16829 ("*tldo_lduw1_sp64"): Likewise.
16830 ("*tldo_ldx_sp64"): Likewise.
16831 ("*mov<VM32:mode>_insn"): Likewise.
16832 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16833 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16835 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16837 * config/sparc/sparc.md ("type"): New insn type viscmp.
16838 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16840 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16841 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16842 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16843 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16845 ("n7_vis_logical_11cycle"): Likewise.
16846 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16847 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16848 * config/sparc/niagara.md ("niag_vis"): Likewise.
16849 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16850 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16852 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16854 * config/sparc/sparc.md: New instruction type `bmask'.
16855 (bmaskdi_vis): Use the `bmask' type.
16856 (bmasksi_vis): Likewise.
16857 * config/sparc/ultra3.md (us3_array): Likewise.
16858 * config/sparc/niagara7.md (n7_array): Likewise.
16859 * config/sparc/niagara4.md (n4_array): Likewise.
16860 * config/sparc/niagara2.md (niag2_vis): Likewise.
16861 (niag3_vis): Likewise.
16862 * config/sparc/niagara.md (niag_vis): Likewise.
16864 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16866 * ipa-comdats.c: Remove optimize check from gate.
16867 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16868 for functions not optimized.
16869 (ipa_fn_summary_read): Skip optimize check.
16870 (ipa_fn_summary_write): Likewise.
16871 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16873 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16875 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16876 (check_callers): Give up on caller which is not optimized.
16877 (inline_small_functions): Likewise.
16878 (ipa_inline): Do not give up when not optimizing.
16879 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16880 away unoptimizes cdtors.
16881 (whole_program_function_and_variable_visibility): Do
16882 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16883 * ipa.c (process_references): Do not check optimize.
16884 (symbol_table::remove_unreachable_nodes): Update optimize check.
16885 (set_writeonly_bit): Update optimize check.
16886 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16887 (pass_ipa_single_use::gate): Remove.
16889 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16891 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16892 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16893 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16894 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16895 permute_load, permute_store, adjust_extract, adjust_splat,
16896 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16897 replace_swap_with_copy, dump_swap_insn_table,
16898 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16899 recombine_lvx_pattern, recombine_stvx_pattern,
16900 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16901 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16902 to file rs6000-p8swap.c.
16903 * config/rs6000/rs6000-p8swap.c: New file.
16904 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16905 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16906 and rs6000*-*-* targets.
16908 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16910 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
16912 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16914 * lto-wrapper.c (merge_and_complain): Do not merge
16915 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
16916 fsigned_zeros, ftrapping_math, fwrapv.
16917 (append_compiler_options): Do not track these options.
16918 (append_linker_options): Likewie
16920 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16922 * cgraphunit.c (cgraph_node::finalize_function): When
16923 !flag_toplevel_reorde set no_reorder flag.
16924 (varpool_node::finalize_decl): Likewise.
16925 (symbol_table::compile): Drop no toplevel reorder path.
16927 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16929 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
16930 edges; zero probability is not better than uninitialized.
16932 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16934 * asan.h (asan_sanitize_allocas_p): Declare.
16935 * asan.c (asan_sanitize_allocas_p): New function.
16936 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
16937 (handle_builtin_alloca): Likewise.
16938 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
16939 if !asan_sanitize_allocas_p.
16940 * params.def (asan-instrument-allocas): Add new option.
16941 * params.h (ASAN_PROTECT_ALLOCAS): Define.
16942 * opts.c (common_handle_option): Disable allocas sanitization for
16945 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
16947 * asan.c: Include gimple-fold.h.
16948 (get_last_alloca_addr): New function.
16949 (handle_builtin_stackrestore): Likewise.
16950 (handle_builtin_alloca): Likewise.
16951 (asan_emit_allocas_unpoison): Likewise.
16952 (get_mem_refs_of_builtin_call): Add new parameter, remove const
16953 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
16954 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
16955 (instrument_builtin_call): Pass gimple iterator to
16956 get_mem_refs_of_builtin_call.
16957 (last_alloca_addr): New global.
16958 * asan.h (asan_emit_allocas_unpoison): Declare.
16959 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
16960 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
16961 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
16962 if function calls alloca.
16963 * gimple-fold.c (replace_call_with_value): Remove static keyword.
16964 * gimple-fold.h (replace_call_with_value): Declare.
16965 * internal-fn.c: Include asan.h.
16966 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
16967 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
16969 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16971 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
16972 (C_SELFTEST_FLAGS): New.
16973 (CPP_SELFTEST_FLAGS): New.
16974 (SELFTEST_DEPS): New, from deps of s-selftest.
16975 (C_SELFTEST_DEPS): New, from deps of s-selftest.
16976 (CPP_SELFTEST_DEPS): New.
16977 (selftest): Add dependency on s-selftest-c++.
16978 (s-selftest): Rename to...
16979 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
16980 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
16981 than SELFTEST_FLAGS.
16982 (selftest-gdb): Rename to...
16983 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
16985 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
16986 (selftest-valgrind): Rename to...
16987 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
16989 (selftest-valgrind): Reintroduce as an alias for
16990 selftest-c-valgrind.
16991 (s-selftest-c++): New.
16992 (selftest-c++-gdb): New.
16993 (selftest-c++-valgrind): New.
16995 2017-07-06 Olivier Hainque <hainque@adacore.com>
16997 * gcc.c (process_command): When deciding if undefined variables
16998 should be ignored when processing specs, accept "gcc -v" as well.
17000 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17002 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17003 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17005 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17007 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17008 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17009 * config/arm/arm-tables.opt: Regenerate.
17010 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17012 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17014 2017-07-06 Carl Love <cel@us.ibm.com>
17016 * ChangeLog: Clean up from mid air collision
17018 2017-07-06 Carl Love <cel@us.ibm.com>
17020 * config/rs6000/rs6000-c.c: Add support for built-in functions
17021 vector signed int vec_subc (vector signed int, vector signed int);
17022 vector signed __int128 vec_subc (vector signed __int128,
17023 vector signed __int128);
17024 vector unsigned __int128 vec_subc (vector unsigned __int128,
17025 vector unsigned __int128);
17026 vector signed int vec_sube (vector signed int, vector signed int,
17027 vector signed int);
17028 vector unsigned int vec_sube (vector unsigned int,
17029 vector unsigned int,
17030 vector unsigned int);
17031 vector signed __int128 vec_sube (vector signed __int128,
17032 vector signed __int128,
17033 vector signed__int128);
17034 vector unsigned __int128 vec_sube (vector unsigned __int128,
17035 vector unsigned __int128,
17036 vector unsigned __int128);
17037 vector signed int vec_subec (vector signed int, vector signed int,
17038 vector signed int);
17039 vector unsigned int vec_subec (vector unsigned int,
17040 vector unsigned int,
17041 vector unsigned int);
17042 vector signed __int128 vec_subec (vector signed __int128,
17043 vector signed __int128,
17044 vector signed__int128);
17045 vector unsigned __int128 vec_subec (vector unsigned __int128,
17046 vector unsigned __int128,
17047 vector unsigned __int128);
17048 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17049 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17050 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17051 BU_ALTIVEC_OVERLOAD_X definitions.
17052 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17053 * doc/extend.texi: Update the built-in documentation file for the new
17054 built-in functions.
17056 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17059 * diagnostic-show-locus.c (layout::layout): Use start and finish
17060 spelling location for the start and finish of each range.
17061 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17062 Add unused aspect param.
17063 * input.c (expand_location_1): Add "aspect" param, and use it
17064 to access the correct part of the location.
17065 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17067 (expand_location_to_spelling_point): Likewise.
17068 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17069 param, and pass it to expand_location_1.
17071 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17073 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17074 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17075 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17076 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17077 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17078 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17079 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17080 _mm_maskz_getmant_ss): New intrinsics.
17081 (__builtin_ia32_getexpss128_mask): Changed to ...
17082 __builtin_ia32_getexpss128_round ... this.
17083 (__builtin_ia32_getexpsd128_mask): Changed to ...
17084 __builtin_ia32_getexpsd128_round ... this.
17085 * config/i386/i386-builtin-types.def
17086 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17087 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17088 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17089 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17090 __builtin_ia32_getmantss_mask_round): New builtins.
17091 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17092 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17093 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17094 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17095 * config/i386/sse.md
17096 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17097 avx512f_sgetexp<mode><mask_scalar_name>
17098 <round_saeonly_scalar_name> ... this.
17099 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
17100 %0, %1, %2<round_saeonly_op3>}): Changed to ...
17101 vgetexp<ssescalarmodesuffix>
17102 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17103 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
17104 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
17105 avx512f_vgetmant<mode><mask_scalar_name>
17106 <round_saeonly_scalar_name> ... this.
17107 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17108 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
17109 vgetmant<ssescalarmodesuffix>
17110 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
17111 %0<mask_scalar_operand4>, %1, %2
17112 <round_saeonly_scalar_mask_op4>, %3} ... this.
17113 * config/i386/subst.md (mask_scalar_operand4,
17114 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
17115 round_saeonly_scalar_nimm_predicate): New subst attributes.
17117 2017-07-06 Julia Koval <julia.koval@intel.com>
17119 * config/i386/i386.c (ix86_erase_embedded_rounding):
17120 Remove code for old rounding pattern.
17122 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
17124 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
17126 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
17128 * doc/sourcebuild.texi (Test Directives, Variants of
17129 dg-require-support): Add documentation for dg-require-stack-check.
17131 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
17133 * config/i386/subst.md (mask_scalar, round_scalar,
17134 round_saeonly_scalar): New meta-templates.
17135 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
17136 round_scalar_mask_operand3, round_scalar_mask_op3,
17137 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
17138 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
17139 round_saeonly_scalar_constraint,
17140 round_saeonly_scalar_prefix): New subst attribute.
17141 * config/i386/sse.md
17142 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
17143 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
17144 <round_scalar_name> ... this.
17145 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
17146 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
17147 <round_scalar_name> ... this.
17148 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
17149 <sse>_vm<code><mode>3<mask_scalar_name>
17150 <round_saeonly_scalar_name> ... this.
17151 (v<plusminus_mnemonic><ssescalarmodesuffix>
17152 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17153 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17154 v<plusminus_mnemonic><ssescalarmodesuffix>
17155 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17156 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17157 (v<multdiv_mnemonic><ssescalarmodesuffix>
17158 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17159 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17160 v<multdiv_mnemonic><ssescalarmodesuffix>
17161 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17162 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17163 (v<maxmin_float><ssescalarmodesuffix>
17164 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17165 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17166 v<maxmin_float><ssescalarmodesuffix>
17167 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17168 %0<mask_scalar_operand3>, %1, %<iptr>2
17169 <round_saeonly_scalar_mask_op3>} ... this.
17171 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17173 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17174 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17176 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17177 Alan Hayward <alan.hayward@arm.com>
17178 David Sherwood <david.sherwood@arm.com>
17180 * combine.c (simplify_if_then_else): Remove "enum" before
17182 * compare-elim.c (can_eliminate_compare): Likewise.
17183 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17185 (aarch64_lookup_simd_builtin_type): Likewise.
17186 (aarch64_simd_builtin_type): Likewise.
17187 (aarch64_init_simd_builtin_types): Likewise.
17188 (aarch64_simd_expand_args): Likewise.
17189 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17191 (aarch64_reverse_mask): Likewise.
17192 (aarch64_simd_emit_reg_reg_move): Likewise.
17193 (aarch64_gen_adjusted_ldpstp): Likewise.
17194 (aarch64_ccmp_mode_to_code): Likewise.
17195 (aarch64_operands_ok_for_ldpstp): Likewise.
17196 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17197 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17199 (aarch64_min_divisions_for_recip_mul): Likewise.
17200 (aarch64_reassociation_width): Likewise.
17201 (aarch64_get_condition_code_1): Likewise.
17202 (aarch64_simd_emit_reg_reg_move): Likewise.
17203 (aarch64_simd_attr_length_rglist): Likewise.
17204 (aarch64_reverse_mask): Likewise.
17205 (aarch64_operands_ok_for_ldpstp): Likewise.
17206 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17207 (aarch64_gen_adjusted_ldpstp): Likewise.
17208 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17210 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17211 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17212 (arm_lookup_simd_builtin_type): Likewise.
17213 (arm_simd_builtin_type): Likewise.
17214 (arm_init_simd_builtin_types): Likewise.
17215 (arm_expand_builtin_args): Likewise.
17216 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17217 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17218 (ft32_setup_incoming_varargs): Likewise.
17219 (ft32_function_arg): Likewise.
17220 (ft32_function_arg_advance): Likewise.
17221 (ft32_pass_by_reference): Likewise.
17222 (ft32_arg_partial_bytes): Likewise.
17223 (ft32_valid_pointer_mode): Likewise.
17224 (ft32_addr_space_pointer_mode): Likewise.
17225 (ft32_addr_space_legitimate_address_p): Likewise.
17226 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17228 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17229 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17230 (iamcu_alignment): Likewise.
17231 (canonicalize_vector_int_perm): Likewise.
17232 (ix86_noce_conversion_profitable_p): Likewise.
17233 (ix86_mpx_bound_mode): Likewise.
17234 (ix86_operands_ok_for_move_multiple): Likewise.
17235 * config/microblaze/microblaze-protos.h
17236 (microblaze_expand_conditional_branch_reg): Likewise.
17237 * config/microblaze/microblaze.c
17238 (microblaze_expand_conditional_branch_reg): Likewise.
17239 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17241 (rs6000_reassociation_width): Likewise.
17242 (rs6000_invalid_binary_op): Likewise.
17243 (fusion_p9_p): Likewise.
17244 (emit_fusion_p9_load): Likewise.
17245 (emit_fusion_p9_store): Likewise.
17246 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17248 (riscv_hard_regno_mode_ok_p): Likewise.
17249 (riscv_address_insns): Likewise.
17250 (riscv_split_symbol): Likewise.
17251 (riscv_legitimize_move): Likewise.
17252 (riscv_function_value): Likewise.
17253 (riscv_hard_regno_nregs): Likewise.
17254 (riscv_expand_builtin): Likewise.
17255 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17256 (riscv_build_integer): Likewise.
17257 (riscv_split_integer): Likewise.
17258 (riscv_legitimate_constant_p): Likewise.
17259 (riscv_cannot_force_const_mem): Likewise.
17260 (riscv_regno_mode_ok_for_base_p): Likewise.
17261 (riscv_valid_base_register_p): Likewise.
17262 (riscv_valid_offset_p): Likewise.
17263 (riscv_valid_lo_sum_p): Likewise.
17264 (riscv_classify_address): Likewise.
17265 (riscv_legitimate_address_p): Likewise.
17266 (riscv_address_insns): Likewise.
17267 (riscv_load_store_insns): Likewise.
17268 (riscv_force_binary): Likewise.
17269 (riscv_split_symbol): Likewise.
17270 (riscv_force_address): Likewise.
17271 (riscv_legitimize_address): Likewise.
17272 (riscv_move_integer): Likewise.
17273 (riscv_legitimize_const_move): Likewise.
17274 (riscv_legitimize_move): Likewise.
17275 (riscv_address_cost): Likewise.
17276 (riscv_subword): Likewise.
17277 (riscv_output_move): Likewise.
17278 (riscv_canonicalize_int_order_test): Likewise.
17279 (riscv_emit_int_order_test): Likewise.
17280 (riscv_function_arg_boundary): Likewise.
17281 (riscv_pass_mode_in_fpr_p): Likewise.
17282 (riscv_pass_fpr_single): Likewise.
17283 (riscv_pass_fpr_pair): Likewise.
17284 (riscv_get_arg_info): Likewise.
17285 (riscv_function_arg): Likewise.
17286 (riscv_function_arg_advance): Likewise.
17287 (riscv_arg_partial_bytes): Likewise.
17288 (riscv_function_value): Likewise.
17289 (riscv_pass_by_reference): Likewise.
17290 (riscv_setup_incoming_varargs): Likewise.
17291 (riscv_print_operand): Likewise.
17292 (riscv_elf_select_rtx_section): Likewise.
17293 (riscv_save_restore_reg): Likewise.
17294 (riscv_for_each_saved_reg): Likewise.
17295 (riscv_register_move_cost): Likewise.
17296 (riscv_hard_regno_mode_ok_p): Likewise.
17297 (riscv_hard_regno_nregs): Likewise.
17298 (riscv_class_max_nregs): Likewise.
17299 (riscv_memory_move_cost): Likewise.
17300 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17301 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17302 (rl78_addr_space_address_mode): Likewise.
17303 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17305 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17306 (rs6000_reassociation_width): Likewise.
17307 (rs6000_invalid_binary_op): Likewise.
17308 (fusion_p9_p): Likewise.
17309 (emit_fusion_p9_load): Likewise.
17310 (emit_fusion_p9_store): Likewise.
17311 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17312 (ok_for_simple_move_operands): Likewise.
17313 (ok_for_simple_move_strict_operands): Likewise.
17314 (ok_for_simple_arith_logic_operands): Likewise.
17315 (visium_legitimize_reload_address): Likewise.
17316 (visium_select_cc_mode): Likewise.
17317 (output_cbranch): Likewise.
17318 (visium_split_double_move): Likewise.
17319 (visium_expand_copysign): Likewise.
17320 (visium_expand_int_cstore): Likewise.
17321 (visium_expand_fp_cstore): Likewise.
17322 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17323 (visium_function_arg): Likewise.
17324 (visium_function_arg_advance): Likewise.
17325 (visium_libcall_value): Likewise.
17326 (visium_setup_incoming_varargs): Likewise.
17327 (visium_legitimate_constant_p): Likewise.
17328 (visium_legitimate_address_p): Likewise.
17329 (visium_legitimize_address): Likewise.
17330 (visium_secondary_reload): Likewise.
17331 (visium_register_move_cost): Likewise.
17332 (visium_memory_move_cost): Likewise.
17333 (prepare_move_operands): Likewise.
17334 (ok_for_simple_move_operands): Likewise.
17335 (ok_for_simple_move_strict_operands): Likewise.
17336 (ok_for_simple_arith_logic_operands): Likewise.
17337 (visium_function_value_1): Likewise.
17338 (rtx_ok_for_offset_p): Likewise.
17339 (visium_legitimize_reload_address): Likewise.
17340 (visium_split_double_move): Likewise.
17341 (visium_expand_copysign): Likewise.
17342 (visium_expand_int_cstore): Likewise.
17343 (visium_expand_fp_cstore): Likewise.
17344 (visium_split_cstore): Likewise.
17345 (visium_select_cc_mode): Likewise.
17346 (visium_split_cbranch): Likewise.
17347 (output_cbranch): Likewise.
17348 (visium_print_operand_address): Likewise.
17349 * expmed.c (flip_storage_order): Likewise.
17350 * expmed.h (emit_cstore): Likewise.
17351 (flip_storage_order): Likewise.
17352 * genrecog.c (validate_pattern): Likewise.
17353 * hsa-gen.c (gen_hsa_addr): Likewise.
17354 * internal-fn.c (expand_arith_overflow): Likewise.
17355 * ira-color.c (allocno_copy_cost_saving): Likewise.
17356 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17357 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17358 (process_invariant_for_inheritance): Likewise.
17359 * lra-eliminations.c (move_plus_up): Likewise.
17360 * omp-low.c (lower_oacc_reductions): Likewise.
17361 * simplify-rtx.c (simplify_subreg): Likewise.
17362 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17363 (TARGET_CHKP_BOUND_MODE): Likewise..
17364 * targhooks.c (default_chkp_bound_mode): Likewise.
17365 (default_setup_incoming_vararg_bounds): Likewise.
17366 * targhooks.h (default_chkp_bound_mode): Likewise.
17367 (default_setup_incoming_vararg_bounds): Likewise.
17368 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17369 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17370 (have_whole_vector_shift): Likewise.
17371 * tree-vect-stmts.c (vectorizable_load): Likewise.
17372 * doc/tm.texi: Regenerate.
17374 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17376 Graceful degrade if Binutils PR21472 is not available.
17379 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17380 .rodata in flash test fails.
17381 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17382 * confgure: Regenerate.
17383 * config.in: Regenerate.
17384 * config/avr/avr.c (avr_asm_named_section)
17385 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17386 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17387 (avr_asm_init_sections): Same.
17389 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17391 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17392 (fma<VH:mode>4_intrinsic): Likewise.
17393 (*fmsub<VCVTF:mode>4): Likewise.
17394 (*fmsub<VH:mode>4_intrinsic): Likewise.
17396 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17399 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17400 Don't depend on "optimize > 0".
17401 (out_movhi_r_mr, out_movqi_mr_r): Same.
17402 (out_movhi_mr_r, out_movqi_r_mr): Same.
17403 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17404 io_address_operand on "optimize > 0".
17406 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17408 * tree-loop-distribution.c: Add general explanantion on the pass.
17409 (generate_loops_for_partition): Mark distributed loop.
17410 (pg_add_dependence_edges): New parameter. Handle alias data
17411 dependence specially and record it in the parameter if asked.
17412 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17413 (init_partition_graph_vertices, add_partition_graph_edge): New.
17414 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17415 (free_partition_graph_vdata, build_partition_graph): New.
17416 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17417 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17418 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17419 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17420 (version_for_distribution_p, finalize_partitions): New.
17421 (distribute_loop): Handle alias data dependence specially. Factor
17422 out loop fusion code as functions and call these functions.
17424 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17426 * tree-loop-distribution.c (classify_partition): New parameter and
17427 better handle reduction statement.
17428 (rdg_build_partitions): Revise comment.
17429 (distribute_loop): Compute statements in all partitions and pass it
17430 to classify_partition.
17432 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17434 * tree-loop-distribution.c (enum partition_type): New.
17435 (struct partition): New field type.
17436 (partition_merge_into): Add parameter. Update partition type.
17437 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17438 (build_rdg_partition_for_vertex): Compute partition type.
17439 (rdg_build_partitions): Dump partition type.
17440 (distribute_loop): Update calls to partition_merge_into.
17442 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17444 * tree-loop-distribution.c (struct ddr_hasher): New.
17445 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17447 (classify_partition): Call get_data_dependence.
17448 (pg_add_dependence_edges): Ditto.
17449 (distribute_loop): Release data dependence hash table.
17451 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17453 * tree-loop-distribution.c (ref_base_address): Delete.
17454 (similar_memory_accesses): Rename ...
17455 (share_memory_accesses): ... to this. Check if partitions access
17456 the same memory reference.
17457 (distribute_loop): Call share_memory_accesses.
17459 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17461 * tree-loop-distribution.c (struct partition): New field recording
17462 its data reference.
17463 (partition_alloc, partition_free): Init and release data refs.
17464 (partition_merge_into): Merge data refs.
17465 (build_rdg_partition_for_vertex): Collect data refs for partition.
17466 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17468 (distribute_loop): Remve data refs from vertice data of partition
17471 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17473 * tree-loop-distribution.c (params.h): Include header file.
17474 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17475 (datarefs_vec): New global var.
17476 (create_rdg_vertices): Use datarefs_vec directly.
17477 (free_rdg): Don't free data references.
17478 (build_rdg): Update use. Don't free data references.
17479 (distribute_loop): Compute global variable for data references.
17480 Bail out if there are too many data references.
17482 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17484 * tree-loop-distribution.c (loop_nest): New global var.
17485 (build_rdg): Use loop directly, rather than loop nest.
17486 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17488 (distribute_loop): Compute global variable loop nest. Update use.
17490 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17492 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17493 (partition_merge_into): New parameter. Dump reason for fusion.
17494 (distribute_loop): Update use of partition_merge_into.
17496 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17498 * tree-loop-distribution.c (bb_top_order_index): New.
17499 (bb_top_order_index_size, bb_top_order_cmp): New.
17500 (stmts_from_loop): Use topological order.
17501 (pass_loop_distribution::execute): Compute and release topological
17502 order for basic blocks.
17504 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17506 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17509 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17511 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17512 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17513 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17514 * internal-fn.def (LOOP_DIST_ALIAS): New.
17515 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17516 (fold_loop_internal_call): ... this.
17517 (vect_loop_dist_alias_call): New function.
17518 (set_uid_loop_bbs): Call fold_loop_internal_call.
17519 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17522 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17525 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17526 Require dead FLAGS_REG at the beginning of a peephole.
17528 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17531 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17532 arguments in the call to __builtin_ia32_sbb_u32.
17533 (_subborrow_u64): Swap _X and _Y arguments in the call to
17534 __builtin_ia32_sbb_u64.
17536 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17539 * tree-vrp.c (compare_assert_loc): Turn into a function template
17540 with stable template parameter. Only test if a->e is NULL,
17541 !a->e == !b->e has been verified already. Use e == NULL or
17542 e != NULL instead of e or ! e tests. If stable is true, don't use
17543 iterative_hash_expr, on the other side allow a or b or both NULL
17544 and sort the NULLs last.
17545 (process_assert_insertions): Sort using compare_assert_loc<false>
17546 instead of compare_assert_loc, later sort using
17547 compare_assert_loc<true> before calling process_assert_insertions_for
17548 in a loop. Use break instead of continue once seen NULL pointer.
17550 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17552 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17553 Cortex-R7 and Cortex-R8 processors.
17555 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17557 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17558 uninitialized while src is not.
17560 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17562 * common/config/arm/arm-common.c: Adjust include path for
17564 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17565 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17566 (arm-cpu-data.h): Likewise.
17567 (arm-cpu-cdata.h): Likewise.
17568 * config/arm/arm-cpu.h: Delete.
17569 * config/arm/arm-cpu-cdata.h: Delete.
17570 * config/arm/arm-cpu-data.h: Delete.
17572 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17574 * config/arm/arm-cpus.in (cortex-a55): New.
17575 (cortex-a75): Likewise.
17576 (cortex-a75.cortex-a55): Likewise.
17577 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17579 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17580 * config/arm/arm-cpu-cdata.h: Regenerate.
17581 * config/arm/arm-cpu-data.h: Regenerate.
17582 * config/arm/arm-cpu.h: Regenerate.
17583 * config/arm/arm-tables.opt: Regenerate.
17584 * config/arm/arm-tune.md: Regenerate.
17586 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17588 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17590 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17592 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17595 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17597 PR tree-optimization/81292
17598 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
17599 full_string_p, also call adjust_related_strinfos if the adjustment
17600 is simple, otherwise invalidate related strinfos.
17602 2017-07-04 Martin Liska <mliska@suse.cz>
17605 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
17606 newly created variable as DECL_IGNORED_P.
17608 2017-07-04 Martin Liska <mliska@suse.cz>
17611 * ipa-inline.c (inline_small_functions):
17612 Use xstrdup_for_dump.
17614 2017-07-04 Tom de Vries <tom@codesourcery.com>
17616 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17618 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17621 * config/darwin.c (darwin_function_switched_text_sections):
17624 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17626 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17628 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17630 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17632 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17634 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17635 min_profitable_iters, and th as inclusive lower bounds.
17636 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17637 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17638 for min_profitable_iters and min_profitable_estimate.
17639 (vect_transform_loop): Treat th as an inclusive lower bound.
17640 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17642 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17645 * config/darwin.c (darwin_function_switched_text_sections):
17646 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17647 in two pieces, and suppress the use of buf.
17649 2017-07-03 Nathan Sidwell <nathan@acm.org>
17651 * hash-table.h (hash_table_mod1): Fix indentation.
17653 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17655 PR middle-end/81290
17656 * predict.c (force_edge_cold): Be more careful about propagation
17658 * profile-count.h (profile_probability::guessed,
17659 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17661 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17663 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17665 * doc/invoke.texi (rcpc architecture extension): Document it.
17667 2017-07-03 Richard Biener <rguenther@suse.de>
17669 PR tree-optimization/60510
17670 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17671 the scalar reduction PHI and use it.
17672 (vectorizable_reduction): Properly guard the single_defuse_cycle
17673 path for non-SLP reduction chains where we cannot use it.
17674 Rework reduc_def/index and vector type deduction. Rework
17675 vector operand gathering during reduction op code-gen.
17676 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17677 chains dissolve the chain and leave it to non-SLP reduction
17680 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17682 * tree-data-ref.h (dr_alignment): Declare.
17683 * tree-data-ref.c (dr_alignment): New function.
17684 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17685 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17687 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17689 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17691 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17692 and base_misalignment fields.
17693 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17694 * tree-data-ref.c: Include builtins.h.
17695 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17696 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17697 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17698 * tree-vect-data-refs.c: Include tree-cfg.h.
17699 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17700 fields instead of calculating an alignment here.
17701 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17702 innermost_loop_behavior fields.
17704 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17706 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17708 (DR_STEP_ALIGNMENT): New macro.
17709 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17710 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17711 (create_data_ref): Print it.
17712 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17713 to tell whether the step preserves vector (mis)alignment.
17714 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17715 Move the check for an integer step and generalise to all INTEGER_CST.
17716 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17717 Print the outer step alignment.
17719 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17721 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17722 with offset_alignment.
17723 (DR_ALIGNED_TO): Delete.
17724 (DR_OFFSET_ALIGNMENT): New macro.
17725 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17726 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17727 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17728 (create_data_ref): Likewise.
17729 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17730 (vect_analyze_data_refs): Likewise.
17731 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17732 creating dummy innermost behavior.
17734 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17736 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17737 with a "innermost_loop_behavior *" and refeence tree.
17738 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17739 (create_data_ref): Update call accordingly.
17740 * tree-predcom.c (find_looparound_phi): Likewise.
17742 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17744 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17745 fields with dr_wrt_vec_loop.
17746 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17747 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17748 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17749 (vect_dr_behavior): New function.
17750 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17751 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17752 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17753 track whether the step preserves the misalignment.
17754 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17755 Use vect_dr_behavior.
17756 (vect_setup_realignment): Update call accordingly.
17757 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17758 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17759 call to vect_create_addr_base_for_vector_ref.
17760 (vect_create_cond_for_align_checks): Likewise.
17761 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17762 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17763 (vect_recog_mask_conversion_pattern): Likewise.
17764 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17765 (new_stmt_vec_info): Remove redundant zeroing.
17767 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17769 * common/config/arm/arm-common.c (arm_be8_option): New function.
17770 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17771 (ISA_ARMv6): Add isa_bit_be8.
17772 * config/arm/arm.h (arm_be8_option): Add prototype.
17773 (BE8_SPEC_FUNCTION): New define.
17774 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17775 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17776 (mlittle-endian): Similarly.
17777 (mbe8, mbe32): New options.
17778 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17779 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17781 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17783 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17785 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17787 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17788 (cleanup_tree_cfg_bb): Use it.
17789 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17791 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17793 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17796 * loop-doloop.c (add_test): Update profile.
17798 2017-07-03 Martin Liska <mliska@suse.cz>
17801 * sanopt.c (rewrite_usage_of_param): New function.
17802 (sanitize_rewrite_addressable_params): Likewise.
17803 (pass_sanopt::execute): Call rewrite_usage_of_param.
17805 2017-07-03 Richard Biener <rguenther@suse.de>
17807 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17808 back to using VIEW_CONVERT_EXPR.
17810 2017-07-03 Martin Liska <mliska@suse.cz>
17813 * doc/extend.texi: Document when a resolver function is
17814 generated for target_clones.
17816 2017-07-03 Martin Liska <mliska@suse.cz>
17818 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17819 and shadow memory of auto variables which are subject of
17820 use-after-scope sanitization.
17821 (asan_expand_mark_ifn): Add do set only when is_poison.
17823 2016-07-03 Richard Biener <rguenther@suse.de>
17825 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17827 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17828 (vectorizable_reduction): Perform reduction PHI creation when
17829 visiting a reduction PHI and adjust and simplify code generation
17830 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17831 (vect_transform_loop): Visit reduction PHIs.
17832 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17833 defs into the SLP tree.
17834 (vect_build_slp_tree): Reduction defs terminate the recursion.
17835 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17837 (vect_get_vec_defs_for_stmt_copy): Export.
17838 (vect_get_vec_defs): Likewise.
17839 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17841 (vect_get_vec_defs_for_stmt_copy): Declare.
17842 (vect_get_vec_defs): Likewise.
17844 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17846 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17847 parameter with a "loop" parameter and use it instead of the
17848 loop containing DR_STMT. Don't check simple_iv when doing
17849 BB analysis. Describe the two analysis modes in the comment.
17851 2017-07-03 Tom de Vries <tom@codesourcery.com>
17853 PR tree-optimization/69468
17854 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17855 (find_same_succ_bb): Handle ignore_edge_flags.
17857 2017-07-03 Tom de Vries <tom@codesourcery.com>
17859 PR tree-optimization/81192
17860 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17862 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17864 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17866 2017-07-03 Tom de Vries <tom@codesourcery.com>
17868 PR tree-optimization/81192
17869 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17870 BB_SAME_SUCC (bb) == NULL.
17872 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17874 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17877 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17879 * dumpfile.c: Include profile-count.h
17880 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17882 (insert_cond_bb): Update profile.
17883 * tree-cfg.h (insert_cond_bb): Update prototype.
17884 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17885 * tree-dump.c: Do not include tree-cfg.
17887 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17889 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17891 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17893 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17896 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17898 * tree-complex.c (expand_complex_div_wide): update profile.
17900 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17901 Alan Hayward <alan.hayward@arm.com>
17902 David Sherwood <david.sherwood@arm.com>
17904 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17905 (CORETYPES_H): New define.
17906 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17907 (insn-modes-inline.h, s-modes-inline-h): New rules.
17908 (generated_files): Add insn-modes-inline.h.
17909 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17910 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
17911 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
17912 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
17913 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
17914 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
17915 (build/gencodes.o, build/genconditions.o): Likewise.
17916 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
17917 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
17918 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
17919 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
17920 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
17921 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
17922 * coretypes.h: Include everything up to real.h for generators.
17923 Include insn-modes.h first. Include wide-int-print.h after
17924 wide-int.h. Include insn-modes-inline.h and then machmode.h.
17925 * machmode.h: Don't include insn-modes.h here.
17926 * function-tests.c: Remove includes of signop.h, machmode.h,
17927 double-int.h and wide-int.h.
17929 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
17931 * optc-save-gen.awk: Likewise.
17932 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
17933 * godump.c: Remove include of wide-int-print.h.
17934 * pretty-print.h: Likewise.
17935 * wide-int-print.cc: Likewise.
17936 * wide-int.cc: Likewise.
17937 * hash-map-tests.c: Remove include of signop.h.
17938 * hash-set-tests.c: Likewise.
17939 * rtl-tests.c: Likewise.
17940 * mkconfig.sh: Remove include of machmode.h.
17941 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
17943 (emit_insn_modes_inline_h): ...this new function. Emit the code
17944 into an insn-modes-inline.h header file, adding appropriate
17945 include guards and end comments.
17946 (emit_insn_modes_c_header): Remove include of machmode.h.
17947 (emit_min_insn_modes_c_header): Include coretypes.h rather than
17949 (main): Handle -i flag and call emit_insn_modes_inline_h when
17952 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17954 * tree-ssa-strlen.c (strinfo): Rename the length field to
17955 nonzero_chars. Add a full_string_p field.
17956 (compare_nonzero_chars, zero_length_string_p): New functions.
17957 (get_addr_stridx): Add an offset_out parameter.
17958 Use compare_nonzero_chars.
17959 (get_stridx): Update accordingly. Use compare_nonzero_chars.
17960 (new_strinfo): Update after above changes to strinfo.
17961 (set_endptr_and_length): Set full_string_p.
17962 (get_string_length): Update after above changes to strinfo.
17963 (unshare_strinfo): Update call to new_strinfo.
17964 (maybe_invalidate): Likewise.
17965 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
17966 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
17967 as a uhwi instead of an shwi. Update after above changes to
17968 strinfo and new_strinfo.
17969 (zero_length_string): Assert that chainsi contains full strings.
17970 Use zero_length_string_p. Update call to new_strinfo.
17971 (adjust_related_strinfos): Update after above changes to strinfo.
17972 Copy full_string_p from origsi.
17973 (adjust_last_stmt): Use zero_length_string_p.
17974 (handle_builtin_strlen): Update after above changes to strinfo and
17975 new_strinfo. Install the lhs as the string length if the previous
17976 entry didn't describe a full string.
17977 (handle_builtin_strchr): Update after above changes to strinfo
17979 (handle_builtin_strcpy): Likewise.
17980 (handle_builtin_strcat): Likewise.
17981 (handle_builtin_malloc): Likewise.
17982 (handle_pointer_plus): Likewise.
17983 (handle_builtin_memcpy): Likewise. Track nonzero characters
17984 that aren't necessarily followed by a nul terminator.
17985 (handle_char_store): Likewise.
17987 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17989 PR tree-optimization/80769
17990 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
17991 for malloc and calloc. Document the new invariant that all related
17992 strinfos have delayed lengths or none do.
17993 (verify_related_strinfos): Move earlier in file.
17994 (set_endptr_and_length): New function, split out from...
17995 (get_string_length): ...here. Also set the lengths of related
17997 (zero_length_string): Assert that chainsi has known (rather than
17999 (adjust_related_strinfos): Likewise.
18001 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18003 PR tree-optimization/81136
18004 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18005 assert that two references with the same misalignment have the same
18006 compile-time misalignment if those compile-time misalignments
18009 2017-07-01 Andi Kleen <ak@linux.intel.com>
18011 * print-tree.c (print_node): Print all attributes.
18013 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18015 * cfg.c (scale_bbs_frequencies): New function.
18016 * cfg.h (scale_bbs_frequencies): Declare it.
18017 * cfgloopanal.c (single_likely_exit): Cleanup.
18018 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18020 (scale_loop_profile): Likewise.
18021 (loop_version): Likewise.
18022 (create_empty_loop_on_edge): Update.
18023 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18024 scale_loop_frequencies, scale_loop_profile, loopify,
18025 loop_version): Update prototypes.
18026 * modulo-sched.c (sms_schedule): Update.
18027 * predict.c (unlikely_executed_edge_p): Also check probability.
18028 (probably_never_executed_edge_p): Fix typo.
18029 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18030 * tree-parloops.c (gen_parallel_loop): Update.
18031 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18032 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18033 * tree-ssa-loop-split.c (split_loop): Update.
18034 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18035 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18036 (vect_loop_versioning): Update.
18037 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18039 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18041 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18043 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18045 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18046 to keep profile consistent.
18048 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18050 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18051 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18052 * profile-count.h (max_safe_multiplier): Make unsigned.
18053 (profile_count::guessed_zero): New.
18055 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18057 * bb-reorder.c (fix_up_crossing_landing_pad,
18058 fix_crossing_conditional_branches): Use make_single_succ_edge
18059 to keep profile consistent.
18061 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18063 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18066 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18069 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18070 the right scopes, make sure cond_jump isn't preserved between multiple
18071 iterations. Search for fallthru edge whenever there are 3+ edges and
18072 use find_fallthru_edge for it.
18074 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18076 Patch by Alexander Monakov <amonakov@ispras.ru>
18077 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18078 probabilities consistently.
18080 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18082 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18083 emit_cmp_and_jump_insns.
18085 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18088 * tree-inline.c (expand_call_inline): Combine profile statuses.
18090 2017-06-30 Andrew Pinski <apinski@cavium.com>
18092 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18093 fold_stmt returned true.
18095 2017-06-30 Nathan Sidwell <nathan@acm.org>
18097 * ggc.h (empty_string): Delete.
18098 * cfgexpand.c (expand_asm_stmt): Use plain "".
18099 * optabs.c (expand_asm_memory_barrier): Likewise.
18100 * stringpool.c (empty_string): Delete.
18101 (digit_vector, digit_string): Delete.
18102 (ggc_alloc_string): Use plain "", don't optimize single digit
18103 strings. Use ggc_alloc_atomic.
18105 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
18107 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
18108 comparison set and one other set, use the cost of the non-comparison
18111 2017-06-30 Nathan Sidwell <nathan@acm.org>
18113 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
18116 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
18118 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
18119 loops. Remove now unneeded calls to gimple_switch_set_label() that
18120 just set removed labels to NULL_TREE.
18122 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
18124 * tree-ssanames.c (set_range_info_raw): Abstract from ...
18125 (set_range_info): ...here. Only call set_range_info_raw if domain
18127 (set_nonzero_bits): Call set_range_info_raw.
18128 * tree-ssanames.h (set_range_info_raw): New.
18130 2017-06-30 Jakub Jelinek <jakub@redhat.com>
18133 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
18134 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
18135 of nonimmediate_operand and <store_mask_constraint> instead of m
18136 for the input operand. For V8FI iterator, always split if input
18137 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
18138 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
18139 <store_mask_predicate> instead of register_operand and
18140 <store_mask_constraint> instead of v for the input operand. Make
18141 sure both operands aren't MEMs for if not <mask_applied>.
18143 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
18145 * lto-wrapper.c (copy_file) Close both file descriptors before
18148 2017-06-30 Martin Liska <mliska@suse.cz>
18151 * multiple_target.c (create_dispatcher_calls): Make ifunc
18152 also for function that don't have calls or are not referenced.
18154 2017-06-30 Richard Biener <rguenther@suse.de>
18156 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
18157 analyze the first scalar stmt. Move vector type computation
18158 for the BB case here from ...
18159 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
18160 live operation processing in the SLP case properly.
18162 2017-06-30 Richard Biener <rguenther@suse.de>
18164 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18166 2017-06-30 Martin Liska <mliska@suse.cz>
18169 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18170 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18172 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18174 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18177 2017-06-30 Richard Biener <rguenther@suse.de>
18179 PR tree-optimization/81249
18180 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18181 condition reduction result to original scalar type.
18183 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18185 * profile-count.h (enum profile_quality): Fix typos and whitespace
18188 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18190 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18191 type for branch probabilities.
18193 2017-06-29 Julian Brown <julian@codesourcery.com>
18194 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18196 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18197 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18198 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18199 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18201 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18203 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18204 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18205 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18206 CC usage from generic code to here.
18207 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18208 CC usage into the target macros.
18210 2017-06-29 Maya Rashish <coypu@sdf.org>
18212 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18215 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18217 * arm/arm-builtins.c: Include profile-count.h
18218 * except.c (sjlj_emit_function_enter): Use
18219 profile_probability::unlikely.
18221 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18223 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18224 and tocrel_offset be pointer args rather than implicitly using
18226 (legitimate_constant_pool_address_p, rs6000_emit_move,
18227 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18228 tocrel_offset and use in toc_relative_expr_p call.
18229 (print_operand, print_operand_address): Use static tocrel_base_oac
18230 and tocrel_offset_oac.
18231 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18234 2017-06-29 Maya Rashish <coypu@sdf.org>
18236 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18238 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18240 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18241 objects, take into account only the alignment of 'op0' and 'mode1' if
18244 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18246 * ccmp.c (ccmp_tree_comparison_p): New function.
18247 (ccmp_candidate_p): Update to use above function.
18248 (get_compare_parts): New function.
18249 (expand_ccmp_next): Update to use new functions.
18250 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18252 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18253 take mode as argument.
18254 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18255 * expr.c (expand_expr_real_1): Pass mode as argument.
18257 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18259 * combine.c (combine_instructions): Print insns to dump_file, together
18262 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18264 * asan.c (asan_emit_stack_protection): Update.
18265 (create_cond_insert_point): Update.
18266 * auto-profile.c (afdo_propagate_circuit): Update.
18267 * basic-block.h (struct edge_def): Turn probability to
18268 profile_probability.
18269 (EDGE_FREQUENCY): Update.
18270 * bb-reorder.c (find_traces_1_round): Update.
18271 (better_edge_p): Update.
18272 (sanitize_hot_paths): Update.
18273 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18274 (make_single_succ_edge): Update.
18275 (check_bb_profile): Update.
18276 (dump_edge_info): Update.
18277 (update_bb_profile_for_threading): Update.
18278 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18279 probabilitycount to 0.
18280 * cfgbuild.c (compute_outgoing_frequencies): Update.
18281 * cfgcleanup.c (try_forward_edges): Update.
18282 (outgoing_edges_match): Update.
18283 (try_crossjump_to_edge): Update.
18284 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18285 (expand_gimple_tailcall): Update.
18286 (construct_init_block): Use make_single_succ_edge.
18287 (construct_exit_block): Use make_single_succ_edge.
18288 * cfghooks.c (verify_flow_info): Update.
18289 (redirect_edge_succ_nodup): Update.
18290 (split_edge): Update.
18291 (account_profile_record): Update.
18292 * cfgloopanal.c (single_likely_exit): Update.
18293 * cfgloopmanip.c (scale_loop_profile): Update.
18294 (set_zero_probability): Remove.
18295 (duplicate_loop_to_header_edge): Update.
18296 * cfgloopmanip.h (loop_version): Update prototype.
18297 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18298 (force_nonfallthru_and_redirect): Update.
18299 (update_br_prob_note): Update.
18300 (rtl_verify_edges): Update.
18301 (purge_dead_edges): Update.
18302 (rtl_lv_add_condition_to_bb): Update.
18303 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18304 * cgraphunit.c (init_lowered_empty_function): Update.
18305 (cgraph_node::expand_thunk): Update.
18306 * cilk-common.c: Include profile-count.h
18307 * dojump.c (inv): Remove.
18308 (jumpifnot): Update.
18309 (jumpifnot_1): Update.
18310 (do_jump_1): Update.
18312 (do_jump_by_parts_greater_rtx): Update.
18313 (do_compare_rtx_and_jump): Update.
18314 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18315 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18316 * dwarf2cfi.c: Include profile-count.h
18317 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18318 (sjlj_emit_dispatch_table): Likewise.
18319 * explow.c: Include profile-count.h
18320 * expmed.c (emit_store_flag_force): Update.
18321 (do_cmp_and_jump): Update.
18322 * expr.c (compare_by_pieces_d::generate): Update.
18323 (compare_by_pieces_d::finish_mode): Update.
18324 (emit_block_move_via_loop): Update.
18325 (store_expr_with_bounds): Update.
18326 (store_constructor): Update.
18327 (expand_expr_real_2): Update.
18328 (expand_expr_real_1): Update.
18329 * expr.h (try_casesi, try_tablejump): Update prototypes.
18330 * gimple-pretty-print.c (dump_probability): Update.
18331 (dump_profile): New.
18332 (dump_gimple_label): Update.
18333 (dump_gimple_bb_header): Update.
18334 * graph.c (draw_cfg_node_succ_edges): Update.
18335 * hsa-gen.c (convert_switch_statements): Update.
18336 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18337 (find_if_case_1): Update.
18338 (find_if_case_2): Update.
18339 * internal-fn.c (expand_arith_overflow_result_store): Update.
18340 (expand_addsub_overflow): Update.
18341 (expand_neg_overflow): Update.
18342 (expand_mul_overflow): Update.
18343 (expand_vector_ubsan_overflow): Update.
18344 * ipa-cp.c (good_cloning_opportunity_p): Update.
18345 * ipa-split.c (split_function): Use make_single_succ_edge.
18346 * ipa-utils.c (ipa_merge_profiles): Update.
18347 * loop-doloop.c (add_test): Update.
18348 (doloop_modify): Update.
18349 * loop-unroll.c (compare_and_jump_seq): Update.
18350 (unroll_loop_runtime_iterations): Update.
18351 * lra-constraints.c (lra_inheritance): Update.
18352 * lto-streamer-in.c (input_cfg): Update.
18353 * lto-streamer-out.c (output_cfg): Update.
18354 * mcf.c (adjust_cfg_counts): Update.
18355 * modulo-sched.c (sms_schedule): Update.
18356 * omp-expand.c (expand_omp_for_init_counts): Update.
18357 (extract_omp_for_update_vars): Update.
18358 (expand_omp_ordered_sink): Update.
18359 (expand_omp_for_ordered_loops): Update.
18360 (expand_omp_for_generic): Update.
18361 (expand_omp_for_static_nochunk): Update.
18362 (expand_omp_for_static_chunk): Update.
18363 (expand_cilk_for): Update.
18364 (expand_omp_simd): Update.
18365 (expand_omp_taskloop_for_outer): Update.
18366 (expand_omp_taskloop_for_inner): Update.
18367 * omp-simd-clone.c (simd_clone_adjust): Update.
18368 * optabs.c (expand_doubleword_shift): Update.
18369 (expand_abs): Update.
18370 (emit_cmp_and_jump_insn_1): Update.
18371 (expand_compare_and_swap_loop): Update.
18372 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18373 * predict.c (predictable_edge_p): Update.
18374 (edge_probability_reliable_p): Update.
18375 (set_even_probabilities): Update.
18376 (combine_predictions_for_insn): Update.
18377 (combine_predictions_for_bb): Update.
18378 (propagate_freq): Update.
18379 (estimate_bb_frequencies): Update.
18380 (force_edge_cold): Update.
18381 * profile-count.c (profile_count::dump): Add missing space into dump.
18382 (profile_count::debug): Add newline.
18383 (profile_count::differs_from_p): Explicitly convert to unsigned.
18384 (profile_count::stream_in): Update.
18385 (profile_probability::dump): New member function.
18386 (profile_probability::debug): New member function.
18387 (profile_probability::differs_from_p): New member function.
18388 (profile_probability::differs_lot_from_p): New member function.
18389 (profile_probability::stream_in): New member function.
18390 (profile_probability::stream_out): New member function.
18391 * profile-count.h (profile_count_quality): Rename to ...
18392 (profile_quality): ... this one.
18393 (profile_probability): New.
18394 (profile_count): Update.
18395 * profile.c (compute_branch_probabilities): Update.
18396 * recog.c (peep2_attempt): Update.
18397 * sched-ebb.c (schedule_ebbs): Update.
18398 * sched-rgn.c (find_single_block_region): Update.
18399 (compute_dom_prob_ps): Update.
18400 (schedule_region): Update.
18401 * sel-sched-ir.c (compute_succs_info): Update.
18402 * stmt.c (struct case_node): Update.
18403 (do_jump_if_equal): Update.
18404 (get_outgoing_edge_probs): Update.
18405 (conditional_probability): Update.
18406 (emit_case_dispatch_table): Update.
18407 (expand_case): Update.
18408 (expand_sjlj_dispatch_table): Update.
18409 (emit_case_nodes): Update.
18410 * targhooks.c: Update.
18411 * tracer.c (better_p): Update.
18412 (find_best_successor): Update.
18413 * trans-mem.c (expand_transaction): Update.
18414 * tree-call-cdce.c: Update.
18415 * tree-cfg.c (gimple_split_edge): Upate.
18416 (move_sese_region_to_fn): Upate.
18417 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18418 * tree-eh.c (lower_resx): Upate.
18419 (cleanup_empty_eh_move_lp): Upate.
18420 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18421 * tree-inline.c (copy_edges_for_bb): Update.
18422 (copy_cfg_body): Update.
18423 * tree-parloops.c (gen_parallel_loop): Update.
18424 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18425 (gimple_gen_time_profiler): Update.
18426 * tree-ssa-dce.c (remove_dead_stmt): Update.
18427 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18428 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18429 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18430 (unloop_loops): Update.
18431 (try_peel_loop): Update.
18432 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18433 * tree-ssa-loop-split.c (connect_loops): Update.
18434 (split_loop): Update.
18435 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18436 (hoist_guard): Update.
18437 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18438 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18439 (value_replacement): Update.
18440 * tree-ssa-reassoc.c (branch_fixup): Update.
18441 * tree-ssa-tail-merge.c (replace_block_by): Update.
18442 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18443 (create_edge_and_update_destination_phis): Update.
18444 (compute_path_counts): Update.
18445 (recompute_probabilities): Update.
18446 (update_joiner_offpath_counts): Update.
18447 (freqs_to_counts_path): Update.
18448 (duplicate_thread_path): Update.
18449 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18450 (struct switch_conv_info): Update.
18451 (gen_inbound_check): Update.
18452 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18453 (vect_do_peeling): Update.
18454 (vect_loop_versioning): Update.
18455 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18456 (optimize_mask_stores): Update.
18457 * ubsan.c (ubsan_expand_null_ifn): Update.
18458 * value-prof.c (gimple_divmod_fixed_value): Update.
18459 (gimple_divmod_fixed_value_transform): Update.
18460 (gimple_mod_pow2): Update.
18461 (gimple_mod_pow2_value_transform): Update.
18462 (gimple_mod_subtract): Update.
18463 (gimple_mod_subtract_transform): Update.
18464 (gimple_ic): Update.
18465 (gimple_stringop_fixed_value): Update.
18466 (gimple_stringops_transform): Update.
18467 * value-prof.h: Update.
18469 2017-06-29 Carl Love <cel@us.ibm.com>
18471 * config/rs6000/rs6000-c.c: Add support for built-in functions
18472 vector signed int vec_signed (vector float);
18473 vector signed long long vec_signed (vector double);
18474 vector signed int vec_signed2 (vector double, vector double);
18475 vector signed int vec_signede (vector double);
18476 vector signed int vec_signedo (vector double);
18477 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18478 instruction generator.
18479 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18480 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18481 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18483 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18484 vunsignede_v2df): Add define_expands.
18485 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18486 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18487 VEC_UNSIGNEDO): Add definitions.
18488 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18489 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18490 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18491 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18492 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18493 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18494 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18495 vec_unsignede, vec_unsignedo): Add builtin defines.
18496 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18498 * doc/extend.texi: Update the built-in documentation file for the
18499 new built-in functions.
18501 2017-06-29 Richard Biener <rguenther@suse.de>
18503 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18504 reduction chains to LOOP_VINFO_REDUCTIONS.
18505 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18506 SLP reductions after processing reduction chains.
18508 2017-06-29 Nathan Sidwell <nathan@acm.org>
18510 * builtins.c (fold_builtin_FUNCTION): Use
18511 lang_hooks.decl_printable_name.
18513 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18515 PR middle-end/81194
18516 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18517 with only one label.
18518 * stmt.c (expand_case): Assert NCASES is greater than one.
18520 2017-06-29 Richard Biener <rguenther@suse.de>
18522 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18524 (group_case_labels): Likewise.
18525 (find_taken_edge): Push sanity checking on val to workers...
18526 (find_taken_edge_cond_expr): ... here
18527 (find_taken_edge_switch_expr): ... and here, handle cases
18528 with just a default label.
18529 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18530 (group_case_labels): Likewise.
18531 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18532 group_case_labels does anything cleanup the CFG again.
18534 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18536 PR tree-optimization/81196
18537 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18538 exit condition comparing two IVs.
18540 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18542 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18543 profile to the dummy entry at the end of the list of architectures.
18544 * config/arm/arm-cpu-cdata.h: Regenerated.
18546 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18547 Michael Collison <michael.collison@arm.com>
18550 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18552 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18553 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18554 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18555 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18556 of shift when the shift amount is masked with constant equal to
18557 the size of the mode.
18558 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18561 2017-06-29 Martin Liska <mliska@suse.cz>
18563 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18565 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18568 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18571 (edge_growth_cache_entry::edge_growth_cache_entry): New
18573 (reset_edge_growth_cache): Update to use constructor.
18575 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18577 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18578 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18579 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18581 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18583 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18584 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18586 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18588 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18589 (*-linux-uclibc*): Add t-uclibc tmake_file.
18590 * config/t-musl: New.
18591 * config/t-uclibc: New.
18593 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18595 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18597 (gen_comm_data): Emit architectural setting of arch_prof.
18598 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
18600 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
18601 (armv8-m.base, armv8-m.main): Likewise.
18602 * arm-protos.h (arm_build_target): Add profile field.
18603 (arch_option): Likewise.
18604 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
18606 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
18607 arm_active_target.profile.
18609 2017-06-28 Richard Biener <rguenther@suse.de>
18611 PR middle-end/81227
18612 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
18613 TYPE_OVERFLOW_WRAPS.
18614 * match.pd (negate_expr_p): Likewise.
18615 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18616 fold_build2, not fold_binary.
18618 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18620 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18621 Convert memory address to Pmode.
18622 (aarch64_print_operand): Assert MEM operands are always Pmode.
18624 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18627 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18628 Remove redundant if.
18629 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18630 * config/arm/aarch-common-protos.h
18631 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18632 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18634 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18637 * multiple_target.c (create_dispatcher_calls): Set the default
18638 clone to be static, not public.
18640 2017-06-28 Richard Biener <rguenther@suse.de>
18642 * tree-vect-loop.c (vectorizable_reduction): Move special
18643 cond reduction IV var creation ...
18644 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18645 parameter. Use STMT_VINFO_VECTYPE.
18646 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18649 2017-06-28 Martin Liska <mliska@suse.cz>
18652 * ipa-visibility.c (non_local_p): Handle visibility.
18654 2017-06-28 Martin Liska <mliska@suse.cz>
18657 * common.opt: Add IntegerRange to various options.
18658 * opt-functions.awk (integer_range_info): New function.
18659 * optc-gen.awk: Add integer_range_info to cl_options struct.
18660 * opts-common.c (decode_cmdline_option): Handle
18661 CL_ERR_INT_RANGE_ARG.
18662 (cmdline_handle_error): Likewise.
18663 * opts.c (print_filtered_help): Show valid interval in
18664 when --help is provided.
18665 * opts.h (struct cl_option): Add range_min and range_max fields.
18666 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18668 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18670 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18671 (x * C EQ/NE y * C): New transformation.
18673 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18675 * genmultilib (combination_space): Accept '+' in option names.
18677 2017-06-28 Martin Liska <mliska@suse.cz>
18680 * asan.c (instrument_derefs): Bail out inner references
18681 that are hard register variables.
18683 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18686 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18687 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18689 2017-06-28 Richard Biener <rguenther@suse.de>
18691 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18692 (vect_get_slp_defs): Adjust.
18693 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18695 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18697 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18698 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18699 (vectorizable_reduction): Adjust.
18700 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18702 (vect_get_slp_defs): Likewise.
18703 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18704 (vectorizable_bswap): Adjust.
18705 (vectorizable_call): Likewise.
18706 (vectorizable_conversion): Likewise.
18707 (vectorizable_assignment): Likewise.
18708 (vectorizable_shift): Likewise.
18709 (vectorizable_operation): Likewise.
18710 (vectorizable_store): Likewise.
18711 (vectorizable_condition): Likewise.
18712 (vectorizable_comparison): Likewise.
18714 2017-06-28 Michael Collison <michael.collison@arm.com>
18717 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18719 (arm_gen_movmemqi): Removed unused variable 'i'.
18720 Convert 'for' loop into 'while' loop.
18721 (arm_expand_prologue): Remove last unnecessary set of insn.
18722 (thumb_pop): Remove unused variable 'pushed_words'.
18723 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18725 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18727 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18728 * config/s390/s390-protos.h: Add prototype of
18729 s390_rel_address_ok_p.
18730 * config/s390/s390.c (s390_got_symbol): New function.
18731 (s390_rel_address_ok_p): New function.
18732 (legitimize_pic_address): Use s390_rel_address_ok_p.
18733 (s390_load_got): Use s390_got_symbol.
18734 (s390_option_override): Issue error if
18735 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18736 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18738 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18740 2017-06-27 Andrew Pinski <apinski@cavium.com>
18742 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18743 (X * copysign (1.0, X)): New pattern.
18744 (X * copysign (1.0, -X)): New pattern.
18745 (copysign (-1.0, CST)): New pattern.
18747 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18749 * genmultilib (combination_space): Remove variable.
18750 Validate reuse rules against regular expression for any sequence
18751 of multilib options in any order.
18753 2017-06-27 Michael Collison <michael.collison@arm.com>
18755 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18756 call aarch64_split_simd_combine.
18757 * (aarch64_combine_internal<mode>): Delete pattern.
18758 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18759 Allow register and subreg operands.
18761 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18763 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18764 specific need, just fallback on defaults.
18765 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18767 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18768 Olivier Hainque <hainque@adacore.com>
18770 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18772 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18773 targets. Pick a default if no particular attempt applied.
18774 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18777 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18779 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18780 (x86_64-wrs-vxworks7): Likewise.
18782 2017-06-27 Marek Polacek <polacek@redhat.com>
18785 * ubsan.c (instrument_null): Check get_base_address's result for null.
18787 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18789 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18791 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18793 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18794 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18795 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18796 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18797 New function types.
18798 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18799 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18800 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18801 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18802 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18803 BUILT_IN_FEUPDATEENV): New builtins.
18804 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18805 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18806 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18807 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18809 (builtin_structptr_types): Adjust size.
18810 * tree.c (builtin_structptr_types): Add four entries.
18812 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18813 Olivier Hainque <hainque@adacore.com>
18815 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18816 (TLS_SYM): New local macro, forcing reference to __tls__ on
18817 link command lines for VxWorks 7 RTPs, triggering initialization
18819 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18820 OS features TLS support, true for RTPs on VxWorks 7.
18821 * config/vxworks.c (vxworks_override_options): Setup emutls
18824 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18826 * predict.c (test_prediction_value_range): Use -1U instead of -1
18827 to avoid narrowing conversion warning.
18828 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18829 to avoid narrowing conversion warning.
18830 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18832 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18834 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18836 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18837 64bit configurations.
18838 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18839 (SIZE_TYPE): Likewise.
18840 * config/vxworks.c (vxworks_emutls_var_fields): Use
18841 long_unsigned_type_node instead of unsigned_type_node as the offset
18842 field type, which is "pointer" mode in emutls.c.
18844 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18847 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18849 PR middle-end/81207
18850 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18851 gimple_vuse copying separately from gimple_vdef copying.
18853 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18855 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18858 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18859 Olivier Hainque <hainque@adacore.com>
18861 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18862 for all vxworks7 targets.
18863 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18864 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18865 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18866 variations for VX6/VX7 and 32/64bits later on in ...
18867 (VXWORKS_LIB_SPEC): Leverage new macros.
18868 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18869 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18871 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18873 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18874 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18876 2017-06-26 Carl Love <cel@us.ibm.com>
18878 * config/rs6000/rs6000-c.c: Add support for built-in functions
18879 vector bool char vec_reve (vector bool char);
18880 vector signed char vec_reve (vector signed char);
18881 vector unsigned char vec_reve (vector unsigned char);
18882 vector bool int vec_reve (vector bool int);
18883 vector signed int vec_reve (vector signed int);
18884 vector unsigned int vec_reve (vector unsigned int);
18885 vector bool long long vec_reve (vector bool long long);
18886 vector signed long long vec_reve (vector signed long long);
18887 vector unsigned long long vec_reve (vector unsigned long long);
18888 vector bool short vec_reve (vector bool short);
18889 vector signed short vec_reve (vector signed short);
18890 vector double vec_reve (vector double);
18891 vector float vec_reve (vector float);
18892 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18893 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18894 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18895 (altivec_vreve): New pattern.
18896 * config/rs6000/altivec.h (vec_reve): New define.
18897 * doc/extend.texi (vec_rev): Update the built-in documentation file
18898 for the new built-in functions.
18900 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18902 PR tree-optimization/71815
18903 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18905 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18907 (slsr_process_phi): Likewise.
18908 (replace_uncond_cands_and_profitable_phis): Don't replace a
18909 multiply candidate with a stride of 1 (copy or cast).
18910 (phi_incr_cost): Call uses_consumed_by_stmt rather than
18912 (lowest_cost_path): Likewise.
18913 (total_savings): Likewise.
18915 2017-06-26 Richard Biener <rguenther@suse.de>
18918 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
18919 Use def_builtin_pure for all gather builtins.
18921 2017-06-26 Richard Biener <rguenther@suse.de>
18923 PR tree-optimization/81203
18924 * tree-tailcall.c (find_tail_calls): Do not move stmts into
18925 non-dominating BBs.
18927 2017-06-26 Marek Polacek <polacek@redhat.com>
18930 * doc/invoke.texi: Document -Wmultistatement-macros.
18932 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
18934 * doc/sourcebuild.texi (ARM-specific attributes): Document new
18935 arm_neon_ok_no_float_abi effective target.
18937 2017-06-26 Richard Biener <rguenther@suse.de>
18939 PR tree-optimization/80928
18940 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
18941 (copy_bbs): Set BB_DUPLICATED flag early.
18942 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
18944 (execute_on_shrinking_pred): Likewise.
18945 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
18946 BB_DUPLICATED blocks.
18947 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
18948 iterate over all PHIs considering removal of *gsi.
18950 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
18952 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
18955 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18957 * config/rs6000/rs6000-string.c: (expand_block_clear,
18958 do_load_for_compare, select_block_compare_mode,
18959 compute_current_alignment, expand_block_compare,
18960 expand_strncmp_align_check, expand_strn_compare,
18961 expand_block_move, rs6000_output_load_multiple)
18962 Move functions related to string/block move/compare
18963 to a separate file.
18964 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
18965 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
18966 for this function which is now used in two files.
18967 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
18968 * config.gcc: Add rs6000-string.o to extra_objs for
18969 targets powerpc*-*-* and rs6000*-*-*.
18971 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
18974 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
18975 32-bit, since indexed is not valid for DImode.
18976 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
18977 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
18978 (define_peephole2 for Altivec d-form load): Add 32-bit support.
18979 (define_peephole2 for Altivec d-form store): Likewise.
18982 * multiple_target.c (create_dispatcher_calls): Only create the
18983 dispatcher call if the function is the default clone of a
18984 versioned function.
18986 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
18988 PR middle-end/80902
18989 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
18990 a call, force the call to not be a tail call.
18992 2017-06-23 Jeff Law <law@redhat.com>
18994 * doc/contrib.texi: Add entry for Steven Pemberton's work on
18997 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
18999 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19000 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19001 handling for early expansion of vector shifts (sl,sr,sra,rl).
19002 (builtin_function_type): Add vector shift right instructions
19003 to the unsigned argument list.
19005 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19007 rtl-optimizatoin/79286
19008 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19009 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19010 trap. PIC register plus a const unspec without offset can never trap.
19012 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19014 * tree.h (builtin_structptr_type): New type.
19015 (builtin_structptr_types): Declare new array.
19016 * tree.c (builtin_structptr_types): New array.
19017 (free_lang_data, build_common_tree_nodes): Use it.
19019 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19022 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19025 2017-06-22 Matt Turner <mattst88@gmail.com>
19027 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19028 Lake models to skylake case. Assume skylake for unknown
19029 models with clflushopt.
19031 2017-06-22 Jeff Law <law@redhat.com>
19033 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19034 frame sizes that do not satisfy aarch64_uimm12_shift.
19036 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19038 * profile-count.h (apply_probability,
19039 apply_scale, probability_in): Fix checks for zero.
19041 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19043 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19044 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19046 2016-06-22 Richard Biener <rguenther@suse.de>
19048 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19049 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19050 REDUC_MAX_EXPR support.
19051 (vectorizable_reduction): Likewise.
19052 (vect_create_epilog_for_reduction): Likewise.
19054 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19056 * match.pd (A / (1 << B) -> A >> B): New.
19057 * generic-match-head.c: Include optabs-tree.h.
19058 * gimple-match-head.c: Likewise.
19059 * optabs-tree.h (target_supports_op_p): New.
19060 * optabs-tree.c (target_supports_op_p): New.
19062 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19064 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19065 $gcc_cv_ld --help output.
19066 (gcc_cv_ld_demangle): Likewise.
19067 (gcc_cv_ld_eh_frame_hdr): Likewise.
19068 (gcc_cv_ld_pie): Likewise.
19069 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19070 (gcc_cv_ld_buildid): Likewise.
19071 (gcc_cv_ld_sysroot): Likewise.
19072 (ld_bndplt_support): Likewise.
19073 (ld_pushpopstate_support): Likewise.
19074 * configure: Regenerate.
19075 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19077 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19080 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19081 operands indexes to avoid gap between operands and match_dups.
19083 2017-06-21 Andrew Pinski <apinski@cavium.com>
19085 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19086 Increment Arith_shift and Arith_shift_reg by 1.
19087 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19089 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19090 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19091 (aarch64_strip_extend): Add new argument and test for it.
19092 (aarch64_cheap_mult_shift_p): New function.
19093 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19094 add a cost if it is true.
19095 Update calls to aarch64_strip_extend.
19096 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19098 2017-06-21 Andrew Pinski <apinski@cavium.com>
19100 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
19102 (thunderxt88): Likewise.
19103 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
19104 (thunderx_prefetch_tune): New variable.
19105 (thunderx2t99_prefetch_tune): Update for the correct values.
19106 (thunderxt88_tunings): New variable.
19107 (thunderx_tunings): Use thunderx_prefetch_tune instead of
19108 generic_prefetch_tune.
19109 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
19111 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19113 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
19114 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
19115 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
19116 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
19117 (aarch64_atomic_cas<mode>, GPI): Likewise.
19119 2017-06-21 Martin Liska <mliska@suse.cz>
19121 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
19122 statements on cold and hot labels.
19123 * predict.c (tree_estimate_probability_bb): Remove the
19124 prediction from this place.
19126 2017-06-21 Martin Liska <mliska@suse.cz>
19128 PR tree-optimization/79489
19129 * gimplify.c (maybe_add_early_return_predict_stmt): New
19131 (gimplify_return_expr): Call the function.
19132 * predict.c (tree_estimate_probability_bb): Remove handling
19134 * predict.def: Update comment about early return predictor.
19135 * gimple-predict.h (is_gimple_predict): New function.
19136 * predict.def: Change default value of early return to 66.
19137 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
19139 * passes.def: Put pass_strip_predict_hints to the beginning of
19142 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
19144 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
19145 FUNCTION_DECL declarations.
19146 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
19148 (dwaf2out_decl): Likewise.
19149 * godump.c (go_early_global_decl): Skip call to the real debug hook
19150 for FUNCTION_DECL declarations.
19151 * passes.c (rest_of_decl_compilation): Skip call to the
19152 early_global_decl debug hook for FUNCTION_DECL declarations, unless
19153 -fdump-go-spec is passed.
19155 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19157 * config/i386/i386.c (struct builtin_isa): New field pure_p.
19158 Reorder for compactness.
19159 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
19160 (def_builtin_pure, def_builtin_pure2): New functions.
19161 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19163 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19165 * match.pd (nop_convert): New predicate.
19166 ((A +- CST1) +- CST2): Allow some NOP conversions.
19168 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19171 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19172 with ctors/dtors if GOVD_SHARED is set.
19174 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19176 * config/aarch64/aarch64.md (movti_aarch64):
19177 Emit mov rather than orr.
19178 (movtf_aarch64): Likewise.
19179 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19180 Emit mov rather than orr.
19182 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19184 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19185 Swap alternatives, make integer dup more expensive.
19187 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19189 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19190 Return true for non-tls symbols.
19192 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19194 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19195 (cortex-a75): Likewise.
19196 (cortex-a75.cortex-a55): Likewise.
19197 * config/aarch64/aarch64-tune.md: Regenerate.
19198 * doc/invoke.texi (-mtune): Document new values for -mtune.
19200 2017-06-21 Tom de Vries <tom@codesourcery.com>
19202 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19203 stack_size feature.
19204 (Effective-Target Keywords, Other attributes): Suggest using
19205 dg-add-options stack_size feature to get stack limit in stack_size
19206 effective target documentation.
19208 2017-06-21 Julian Brown <julian@codesourcery.com>
19209 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19211 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19212 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19213 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19215 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19216 attribute type list for neon_multiply.
19217 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19218 attribute type list for neon_multiply.
19219 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19220 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19221 attribute type list for neon_multiply.
19222 * config/arm/types.md (crypto_pmull): Add.
19223 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19224 attribute type list.
19226 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19228 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19231 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19233 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19234 to make sure not to dereference a NULL cost_classes_ptr pointer.
19236 2017-06-20 Carl Love <cel@us.ibm.com>
19238 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19239 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19240 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19241 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19242 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19243 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19244 VMULOSW): New enum "unspec" values.
19245 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19246 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19247 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19248 altivec_vmulosw): New patterns.
19249 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19250 VMULOSW): Add definitions.
19252 2017-06-20 Julia Koval <julia.koval@intel.com>
19254 * config/i386/i386.c: Fix rounding expand for new pattern.
19255 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19257 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19259 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19260 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19262 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19264 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19267 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19269 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19270 architecture, then by implementer ID.
19271 * config/aarch64/aarch64-tune.md: Regenerate.
19273 2017-06-20 Richard Biener <rguenther@suse.de>
19275 PR middle-end/81097
19276 * fold-const.c (split_tree): Fold to type before negating.
19278 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19280 * diagnostic-show-locus.c
19281 (selftest::test_fixit_deletion_affecting_newline): New function.
19282 (selftest::diagnostic_show_locus_c_tests): Call it.
19284 2017-06-20 Andreas Schwab <schwab@suse.de>
19287 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19290 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19292 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19293 __ARM_FEATURE_COPROC according to support.
19295 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19297 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19298 Rewritten to avoid overflow for > 32-bit pointers.
19301 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19302 by removing enum keyword.
19303 (ubsan_type_descriptor): Likewise. Formatting fix.
19306 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19307 splitter): Require TARGET_SSE2 in the condition.
19309 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19312 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19313 for doing vector set of SFmode on ISA 3.0.
19314 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19315 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19317 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19318 SFmode value into a V4SF variable that was extracted from another
19319 V4SF variable without converting the element to double precision
19320 and back to single precision vector format.
19321 (vsx_insert_extract_v4sf_p9_2): Likewise.
19323 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19325 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19326 in UWHI to avoid undefined overflow.
19329 * ubsan.h (enum ubsan_encode_value_phase): New.
19330 (ubsan_encode_value): Change second argument to
19331 enum ubsan_encode_value_phase with default value of
19332 UBSAN_ENCODE_VALUE_GENERIC.
19333 * ubsan.c (ubsan_encode_value): Change second argument to
19334 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19335 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19336 create_tmp_var_raw instead of create_tmp_var and use a
19338 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19339 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19340 ubsan_encode_value callers.
19343 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19344 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19345 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19347 2017-06-19 Richard Biener <rguenther@suse.de>
19349 PR middle-end/81118
19350 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19351 estimates if we changed anything.
19353 2017-06-19 Richard Biener <rguenther@suse.de>
19355 PR tree-optimization/80887
19356 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19357 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19358 simplified lookups, then reset mprts_hook.
19359 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19361 (try_to_simplify): Likewise.
19363 2017-06-19 Martin Liska <mliska@suse.cz>
19366 * gimplify.c (gimplify_switch_expr):
19367 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19369 2017-06-19 Martin Liska <mliska@suse.cz>
19371 * doc/install.texi: Document that PGO runs in 4 stages.
19373 2017-06-19 Martin Liska <mliska@suse.cz>
19376 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19377 to dispatcher function name.
19378 * multiple_target.c (replace_function_decl): New function.
19379 (create_dispatcher_calls): Redirect both edges and references.
19381 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19383 * profile-count.c (profile_count::dump): Dump quality.
19384 (profile_count::differs_from_p): Update for unsigned val.
19385 * profile-count.h (profile_count_quality): New enum.
19386 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19388 2017-06-19 Richard Biener <rguenther@suse.de>
19390 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19391 struct function as arg.
19392 (estimate_numbers_of_iterations): Export overload with loop arg.
19393 (free_numbers_of_iterations_estimates_loop): Use an overload of
19394 free_numbers_of_iterations_estimates instead.
19395 * tree-cfg.c (remove_bb): Adjust.
19396 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19397 * tree-parloops.c (gen_parallel_loop): Likewise.
19398 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19400 (tree_unroll_loops_completely): Likewise.
19401 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19402 Use an overload instead and export.
19403 (estimated_loop_iterations): Adjust.
19404 (max_loop_iterations): Likewise.
19405 (likely_max_loop_iterations): Likewise.
19406 (estimate_numbers_of_iterations): Take struct function as arg
19408 (loop_exits_before_overflow): Adjust.
19409 (free_numbers_of_iterations_estimates_loop): Use an overload.
19410 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19411 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19413 2017-06-19 Richard Biener <rguenther@suse.de>
19416 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19417 RANGE_EXPR conservatively.
19419 2017-06-16 Carl Love <cel@us.ibm.com>
19421 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19422 definitions for vec_float, vec_float2, vec_floato,
19423 vec_floate built-ins.
19424 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19425 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19427 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19428 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19429 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19430 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19431 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19432 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19433 vec_floato): Add builtin defines.
19434 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19435 Update the built-in documentation file for the new built-in
19438 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19440 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19441 (mthumb): Mark as the negative of -marm.
19443 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19445 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19447 (ARM Options, -mtune): Document that this accepts the same
19448 extension options as -mcpu.
19449 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19451 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19453 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19454 permitted extensions.
19456 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19458 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19459 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19460 (armv8-m.main): Add option +nodsp.
19461 * config/arm/arm-cpu-cdata.h: Regenerated.
19463 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19465 * config/arm/t-fuchsia: New file.
19466 * config.gcc (arm*-*-fuchsia*): Use it.
19468 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19470 * config/arm/t-symbian: Rewrite for new option infrastructure.
19472 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19474 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19475 (MULTILIB_REQUIRED): Likewise.
19477 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19479 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19480 (MULTILIB_RESUE): Likewise.
19481 (MULTILIB_MATCHES): Likewise.
19482 (MULTLIB_REQUIRED): Likewise.
19484 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19486 * config/arm/t-rtems: Rewrite for new option framework.
19488 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19490 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19491 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19492 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19493 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19494 * config/arm/t-multilib: ... here.
19495 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19496 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19497 armv7-a and armv8*-a when A-profile libraries have not been built.
19498 * config/arm/t-rmprofile: Rewrite.
19500 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19502 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19503 with a backslash. Remove the backslash after substituting unescaped
19505 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19507 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19509 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19510 the driver to pass through the default setting of -mfloat-abi.
19511 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19513 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19514 (all_feat_combs): New rule.
19515 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19517 * config/arm/t-aprofile: Rewrite.
19519 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19521 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19522 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19523 fpu is not specified by the user/command-line.
19524 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19525 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19526 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19527 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19528 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19529 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19531 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19533 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19534 * config/arm/t-arm-elf: Rewritten.
19536 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19538 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19539 have some floating-point instructions.
19540 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19541 (TARGET_MAYBE_HARD_FLOAT): New macro.
19542 * config/arm/arm-builtins.c (arm_init_builtins): Use
19543 TARGET_MAYBE_HARD_FLOAT.
19544 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19546 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19548 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19549 (configargs.h): Include it.
19550 (arm_print_hint_for_fpu_option): New function.
19551 (arm_parse_fpu_option): New function.
19552 (candidate_extension): New class.
19553 (arm_canon_for_multilib): New function.
19554 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19555 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19556 (ARCH_CANONICAL_SPECS): New macro.
19557 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19559 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19561 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19562 are set after handling multilib fragments. Set target_cpu_default2
19565 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19567 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19569 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19570 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19572 (check_arch): Likewise.
19573 * config/arm/arm.c (arm_configure_build_target): Handle
19574 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19575 options in the default.
19577 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19579 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19580 when an option is an alias of another.
19581 * config/arm/parsecpu.awk (optalias): New parser token.
19582 (gen_comm_data): Mark non-alias options as such. Emit entries
19583 for extension aliases.
19584 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19585 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19586 (armv6kz, armv6zk, armv6t2): Likewise.
19587 (armv7): Make vfpv3-d16 an alias.
19588 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19590 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19591 Sort in canonical order.
19592 (armv8-a): Sort in canonical order.
19593 (armv8.1-a, armv8.2-a): Likewise.
19594 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19596 (cortex-a9): Sort in canonical order.
19597 * config/arm/arm.c (selftests.h): Include it.
19598 (arm_test_cpu_arch_data): New function.
19599 (arm_run_self_tests): New function.
19600 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
19601 (targetm): Move declaration to the end of the file.
19602 * arm-cpu-cdata.h: Regenerated.
19604 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19606 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
19607 call to target_mode_check describing the type of option passed.
19608 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
19609 (arm_target_thumb_only): Use arm_parse_arch_option_name or
19610 arm_parse_cpu_option_name to match parameters against list of
19612 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
19613 arm_arch_core_flags data structure.
19614 * config/arm/arm-cpu_cdata.h: Regenerated.
19616 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19618 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19620 (arm_print_hint_for_cpu_option): Likewise.
19621 (arm_print_hint_for_arch_option): Likewise.
19622 (arm_parse_cpu_option_name): Likewise.
19623 (arm_parse_arch_option_name): Likewise.
19624 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19625 of entries in the all_fpus list.
19626 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19627 (arm_parse_cpu_option_name): Declare.
19628 (arm_parse_arch_option_name): Declare.
19629 (arm_parse_option_features): Declare.
19630 (arm_intialize_isa): Declare.
19631 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19633 (gen_comm_data): ... here. Make definitions non-static.
19634 * config/arm/arm-cpu-data.h: Regenerated.
19635 * config/arm/arm-cpu-cdata.h: Regenerated.
19637 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19639 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19640 (cpu_arch_extension): New structure.
19641 (cpu_arch_option, arch_option, cpu_option): New structures.
19642 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19643 architecture types.
19644 (gen_data): Generate new format data tables.
19645 * config/arm/arm.c (cpu_tune): New structure.
19646 (cpu_option, processors): Delete.
19647 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19648 (arm_print_hint_for_cpu_option): ... this and ...
19649 (arm_print_hint_for_arch_option): ... this.
19650 (arm_parse_arch_cpu_name): Delete. Replace with ...
19651 (arm_parse_cpu_option_name): ... this and ...
19652 (arm_parse_arch_option_name): ... this.
19653 (arm_unrecognized_feature): Change type of target parameter to
19655 (arm_parse_arch_cpu_features): Delete. Replace with ...
19656 (arm_parse_option_features): ... this.
19657 (arm_configure_build_target): Rework to use new configuration data
19659 (arm_print_tune_info): Rework for new configuration data tables.
19660 * config/arm/arm-cpu-data.h: Regenerated.
19661 * config/arm/arm-cpu.h: Regenerated.
19663 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19665 * Makefile.in (OBJS): Move sbitmap.o from here ...
19666 (OBJS-libcommon): ... to here.
19668 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19670 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19671 (ISA_ALL_CRYPTO): New macro.
19672 (ISA_ALL_SIMD): New macro
19673 (ISA_ALL_FP): New macro.
19674 * config/arm/arm.c (fpu_bitlist): Update initializer.
19675 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19677 (arm9e): Add fpu. Add option for nofp
19678 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19679 (arm926ej-s, arm1026ej-s): Likewise.
19680 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19681 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19682 neon-fp16, neon-vfpv4, nofp and nosimd.
19683 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19684 (cortex-a8): Add fpu. Add option for nofp.
19685 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19686 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19687 (cortex-r4f): Add fpu.
19688 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19689 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19691 (cortex-r8): Likewise.
19692 (cortex-m4): Add fpu. Add option for nofp.
19693 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19694 (cortex-a17.cortex-a7): Likewise.
19695 (cortex-a32): Add fpu. Add options for crypto and nofp.
19696 (cortex-a35, cortex-a53): Likewise.
19697 (cortex-a57): Add fpu. Add option for crypto.
19698 (cortex-a72, cortex-a73): Likewise.
19699 (exynos-m1): Likewise.
19700 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19701 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19702 (cortex-m33): Add fpu. Add option for nofp.
19703 * config/arm/arm-cpu-cdata.h: Regenerated
19704 * config/arm/arm-cpu-data.h: Regenerated.
19706 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19708 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19709 (armv5te, armv5tej): Likewise.
19710 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19711 (armv7): Add options fp and vfpv3-d16.
19712 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19713 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19715 (armv7ve): Likewise.
19716 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19717 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19718 (armv8-a): Add nocrypto option.
19719 (armv8.1-a, armv8.2-a): Likewise.
19720 (armv8-m.main): add options fp, fp.dp and nofp.
19722 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19724 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19726 (armv8-a+crc): Delete.
19727 (armv8.1-a): Add options simd, crypto and nofp.
19728 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19729 (armv8.2-a+fp16): Delete.
19730 (armv8-m.main): Add option dsp.
19731 (armv8-m.main+dsp): Delete.
19732 (cortex-a8): Add fpu. Add nofp option.
19733 (cortex-a9): Add fpu. Add nofp and nosimd options.
19734 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19735 link to main cpu and architecture data structures.
19736 (gen_comm_data): Only put isa attributes from the main architecture
19738 (option): New statement for architecture and CPU entries.
19739 * arm.c (struct cpu_option): New structure.
19740 (struct processors): Add entry for options.
19741 (arm_unrecognized_feature): New function.
19742 (arm_parse_arch_cpu_name): Ignore any characters after the first
19744 (arm_parse_arch_cpu_feature): New function.
19745 (arm_configure_build_target): Separate out any CPU and architecture
19746 features and parse separately. Don't error out if -mfpu=auto is
19747 used with only an architecture string.
19748 (arm_print_asm_arch_directives): New function.
19749 (arm_file_start): Call it.
19750 * config/arm/arm-cpu-cdata.h: Regenerated.
19751 * config/arm/arm-cpu-data.h: Likewise.
19752 * config/arm/arm-tables.opt: Likewise.
19754 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19756 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19757 assembler when it is not -mfpu=auto.
19759 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19761 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19762 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19763 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19764 (ASM_CPU_SPEC): Rewrite.
19765 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19766 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19767 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19768 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19769 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19770 copied string is NUL-terminated. Also strip any characters prefixed
19772 (arm_rewrite_selected_arch): New function.
19773 (arm_rewrite_march): New function.
19775 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19777 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19778 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19779 (march, mcpu, mtune): Convert to string-based options.
19780 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19781 (arm_parse_arch_cpu_name): New function.
19782 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19783 identify selected architecture or CPU.
19784 (arm_option_save): New function.
19785 (TARGET_OPTION_SAVE): Redefine.
19786 (arm_option_restore): Restore string options.
19787 (arm_option_print): Print string options.
19789 2017-06-16 Martin Sebor <msebor@redhat.com>
19791 PR tree-optimization/80933
19792 PR tree-optimization/80934
19793 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19794 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19795 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19796 (gimple_fold_builtin): Call them.
19798 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19800 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19801 as unlikely; update profile.
19803 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19805 * predict.c (force_edge_cold): Handle declaring edges impossible
19808 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19810 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19812 (try_unroll_loop_completely): Fix reporting.
19814 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19816 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19818 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19821 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19822 if given a non-constant argument for an intrinsic which requires a
19825 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19827 * profile.c (compare_freqs): New function.
19828 (branch_prob): Sort edge list.
19829 (find_spanning_tree): Assume that the list is priority sorted.
19831 2017-06-16 Richard Biener <rguenther@suse.de>
19833 PR tree-optimization/81090
19834 * passes.def (pass_record_bounds): Remove.
19835 * tree-pass.h (make_pass_record_bounds): Likewise.
19836 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19837 make_pass_record_bounds): Likewise.
19838 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19839 not free niter estimates at the beginning but at the end.
19840 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19842 2017-06-16 Richard Biener <rguenther@suse.de>
19844 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19845 initializer to workaround ICE in host GCC 4.8.
19847 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19849 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19851 (clone_inlined_nodes): Update.
19853 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19855 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19856 prefetch settings, and enable prefetching by default at -O3.
19858 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19860 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19861 Set flag_prefetch_loop_arrays according to tuning data.
19863 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19865 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19866 New tune structure.
19867 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19868 [Unrelated to main purpose of the patch] Place the pointer field last
19869 to enable type checking errors when tune structure are wrongly merged.
19870 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19871 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19872 (thunderx2t99_prefetch_tune): New tune constants.
19873 (tune_params *_tunings): Update all tunings (no functional change).
19874 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19875 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19876 from tunings structures.
19878 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19881 * ubsan.c (instrument_null): Add T argument, use it instead
19882 of computing it based on IS_LHS.
19883 (instrument_object_size): Likewise.
19884 (pass_ubsan::execute): Adjust instrument_null and
19885 instrument_object_size callers to pass gimple_get_lhs or
19886 gimple_assign_rhs1 result to it. Use instrument_null instead of
19887 calling get_base_address and instrument_mem_ref. Handle
19888 aggregate call arguments for object-size sanitization.
19890 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19892 PR tree-optimization/81089
19893 * tree-vrp.c (is_masked_range_test): Validate operands of
19896 2017-06-15 Martin Sebor <msebor@redhat.com>
19899 * dumpfile.c (dump_register): Avoid calling memset to initialize
19900 a class with a default ctor.
19901 * gcc.c (struct compiler): Remove const qualification.
19902 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19903 * hash-table.h: Ditto.
19904 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19906 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19907 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19909 * params.h (struct param_info): Make struct members non-const.
19910 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
19911 with default initialization.
19912 * vec.h (vec_copy_construct, vec_default_construct): New helper
19914 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
19915 with vec_copy_construct.
19916 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
19917 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
19918 * doc/invoke.texi (-Wclass-memaccess): Document.
19920 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19922 * emit-rtl.h (is_leaf): Update comment about local
19923 register allocator.
19925 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19928 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
19929 for a variable to have a section before checking if the section has a
19931 Set section to.persistent if persistent attribute is set.
19932 Warn if .persistent attribute is used on an automatic variable.
19934 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
19936 PR rtl-optimization/80474
19937 * reorg.c (update_block): Do not ignore instructions in a delay slot.
19939 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
19941 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
19944 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
19946 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
19947 (casesi): Emit bounds checking as RTL.
19948 (casesi_internal_mips16_<mode>): Remove bounds checking.
19950 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
19952 * config/xtensa/xtensa.c (xtensa_option_override): Append
19953 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
19954 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
19955 xtensa_doloop_hooks): Define unconditionally.
19956 (xtensa_reorg_loops): Only call reorg_loops in the presence of
19958 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
19959 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
19960 for it in xtensa_option_override.
19961 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
19962 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
19964 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
19966 * doc/cppopts.texi: Document '-' special value to -MF.
19968 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
19970 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
19971 (cortex_a53_fconst): Likewise.
19972 (cortex_a53_fpmul): Likewise.
19973 (cortex_a53_f_load_64): Likewise.
19974 (cortex_a53_f_load_many): Likewise.
19975 (cortex_a53_advsimd_alu): Likewise.
19976 (cortex_a53_advsimd_alu_q): Likewise.
19977 (cortex_a53_advsimd_mul): Likewise.
19978 (cortex_a53_advsimd_mul_q): Likewise.
19979 (fpmac bypass): Add new bypass for fpmac-fpmac case.
19980 Add missing fmul, r2f_cvt and fconst cases.
19982 2017-06-14 Richard Biener <rguenther@suse.de>
19984 PR middle-end/81088
19985 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
19987 (fold_binary_loc): When associating do not treat pre-existing
19988 TREE_OVERFLOW on literal constants as a reason to allow
19989 TREE_OVERFLOW on associated literal constants.
19991 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
19993 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
19994 (MASK_FEATURES): New macro.
19995 * config/sparc/sparc.c (sparc_option_override): Remove the special
19996 handling of -mfpu and generalize it to all MASK_FEATURES switches.
19998 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20000 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20001 a division of 0 if non-call exceptions are enabled.
20003 2017-06-14 Andrew Pinski <apinski@cavium.com>
20004 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20007 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20008 Improve vector initialization code gen for only variable case.
20010 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20012 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20014 2017-06-14 Richard Biener <rguenther@suse.de>
20016 PR tree-optimization/81083
20017 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20020 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20022 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20023 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20024 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20025 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20026 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20027 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20029 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20031 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20032 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20034 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20036 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20038 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20040 * config/rs6000/t-rtems: Don't handle SPE.
20042 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20044 * config/rs6000/t-linux: Don't handle SPE.
20046 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20048 * config/rs6000/eabispe.h: Delete file.
20050 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20052 * config/rs6000/t-spe: Delete file.
20054 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20056 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20057 (rs6000_legitimate_offset_address_p): Return false for anything in
20058 V2SImode or V2SFmode.
20060 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20062 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20063 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20064 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20066 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20067 (rs6000_legitimate_offset_address_p): Ditto.
20068 (rs6000_emit_move): Ditto.
20069 (rs6000_init_builtins): Remove V4HI_type_node.
20071 2017-06-13 Martin Liska <mliska@suse.cz>
20074 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20075 (gate_asan): Likewise.
20076 * asan.h (asan_no_sanitize_address_p): Remove the function.
20077 (sanitize_flags_p): New function.
20078 * builtins.def: Fix coding style.
20079 * common.opt: Use renamed enum value.
20080 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20081 * doc/extend.texi: Document no_sanitize attribute.
20082 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20083 to SANITIZE_UNDEFINED_NONDEFAULT.
20084 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20085 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20086 Use sanitize_flags_p.
20087 * gimplify.c (gimplify_function_tree): Likewise.
20088 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20089 * opts.c (parse_no_sanitize_attribute): New function.
20090 (common_handle_option): Use renamed enum value.
20091 * opts.h (parse_no_sanitize_attribute): Declare.
20092 * tree.c (sanitize_flags_p): New function.
20093 * tree.h: Declared here.
20094 * tsan.c: Use sanitize_flags_p.
20095 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20096 (instrument_mem_ref): Likewise.
20097 (instrument_bool_enum_load): Likewise.
20098 (do_ubsan_in_current_function): Remove the function.
20099 (pass_ubsan::execute): Use sanitize_flags_p.
20100 * ubsan.h: Remove do_ubsan_in_current_function
20101 * tree-cfg.c (print_no_sanitize_attr_value): New function.
20102 (dump_function_to_file): Use it here.
20104 2017-06-13 Martin Jambor <mjambor@suse.cz>
20106 PR tree-optimization/80803
20107 PR tree-optimization/81063
20108 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
20109 (propagate_subaccesses_across_link): Enqueue subtree whenever
20110 necessary instead of relying on the caller.
20112 2017-06-13 Martin Jambor <mjambor@suse.cz>
20114 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
20115 that have a first_link.
20116 (sort_and_splice_var_accesses): Do not check first_link before
20118 (subtree_mark_written_and_enqueue): Likewise.
20119 (propagate_all_subaccesses): Likewise and do not stop at first
20120 parent with a first_link.
20122 2017-06-13 Martin Jambor <mjambor@suse.cz>
20124 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
20127 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20129 * match.pd: New pattern.
20131 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20133 * tree-vrp.c (is_masked_range_test): New function.
20134 (register_edge_assert_for): Determine ranges for
20137 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20139 PR tree-optimization/67328
20140 * fold-const.c (maskable_range_p): New function.
20141 (build_range_check): Generate bittests if possible.
20143 2017-06-13 Martin Liska <mliska@suse.cz>
20145 * gimple-pretty-print.c (dump_probability): Add new argument.
20146 (dump_edge_probability): Dump both probability and count.
20147 (dump_gimple_label): Likewise.
20148 (dump_gimple_bb_header): Likewise.
20150 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
20153 * config/avr/avr-devices.c: Fix indentation.
20154 * config/avr/gen-avr-mmcu-specs.c: Dito.
20156 2017-06-13 Richard Biener <rguenther@suse.de>
20158 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
20159 instead get vector type from stmt_info.
20160 (vectorizable_reduction): Adjust. Remove dead code.
20162 2017-06-13 Richard Biener <rguenther@suse.de>
20164 PR middle-end/81065
20165 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20166 case of C * (x * C2 + C3).
20167 (fold_addr_of_array_ref_difference): Properly fold index difference.
20169 2017-06-12 David S. Miller <davem@davemloft.net>
20172 * config/sparc/sparc.md (return expander): Emit frame blockage if
20173 function uses alloca.
20175 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20177 * combine.c (make_field_assignment): Check len rather than the mode
20178 precision when calling force_to_mode.
20180 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20182 Support multilibs and devices that see flash in RAM address range.
20185 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20186 (avr_mcu_t) <flash_pm_offset>: New field.
20187 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20188 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20189 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20190 (AVR_TINY_PM_OFFSET): Remove macro.
20191 * config/avr/avr.opt (-mshort-calls): New option.
20192 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20193 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20194 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20195 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20196 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20197 instead of avr_arch->have_jmp_call.
20198 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20199 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20200 avr_arch->flash_pm_offset to define.
20201 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20202 new field flash_pm_offset. Add entry for avrxmega3.
20203 (avr_texinfo): Add entry for avrxmega3.
20204 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20205 attiny212, attiny214,
20206 attiny412, attiny414, attiny416, attiny417,
20207 attiny814, attiny816, attiny817,
20208 attiny1614, attiny1616, attiny1617,
20209 attiny3214, attiny3216, attiny3217.
20210 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20211 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20212 (avr_print_operand_address) [AVR_TINY]: Same.
20213 (avr_asm_init_sections) <readonly_data_section>: Only patch
20214 callback if avr_arch->flash_pm_offset = 0.
20215 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20216 for rodata if avr_arch->flash_pm_offset != 0.
20217 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20218 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20219 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20220 (m_options): Append opt_rcall.
20221 (m_dirnames): Append dir_rcall.
20222 * config/avr/t-multilib: Regenerate.
20224 * configure.ac [target=avr]: Check whether avrxmega3 default
20225 linker description file works as needed.
20226 * configure: Regenerate.
20227 * doc/avr-mmcu.texi: Regenerate.
20228 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20229 <__AVR_ARCH__>: Document avrxmega3 and 103.
20230 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20231 <__AVR_SHORT_CALLS__>: Document it.
20232 <__AVR_PM_BASE_ADDRESS__>: Document it.
20233 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20234 (AVR Variable Attributes) <progmem>: Document this is
20235 not needed for avrxmega3.
20236 (AVR Named Address Spaces) <__flash>: Dito.
20238 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20240 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20242 2017-06-12 Doug Rupp <rupp@adacore.com>
20244 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20245 Append vxworks-stdint.h to the tm_file list.
20246 * config/vxworks-stdint.h: New file.
20248 2017-06-12 Martin Liska <mliska@suse.cz>
20250 PR tree-optimization/81041
20251 * tree-profile.c (gimple_gen_ic_func_profiler):
20252 Create an extra BB in profile-generate
20253 (gimple_gen_time_profiler): Likewise.
20255 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20257 PR tree-optimization/81003
20258 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20259 (update_range_test): Use it instead of force_gimple_operand_gsi.
20261 2017-06-12 Richard Biener <rguenther@suse.de>
20263 PR tree-optimization/81053
20264 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20265 with backedge value not defined in loop. Simplify def stmt
20268 2017-06-11 Tom de Vries <tom@codesourcery.com>
20271 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20273 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20274 nvptx_cannot_force_const_mem.
20276 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20278 * opts.c (finish_options): Move test for flag_split_stack after
20279 it has been initialized.
20281 2017-06-11 Jason Merrill <jason@redhat.com>
20283 * tree.h (id_equal): New.
20284 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20285 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20286 instead of strcmp of IDENTIFIER_POINTER.
20288 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20290 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20291 (mark_all_inlined_calls_cdtor): Fix formating.
20292 (inline_transform): Rescale profile before inlining.
20294 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20296 * cgraph.h (cgraph_edge::clone): Update prototype.
20297 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20298 (cgraph_node::create_clone): Update.
20299 (cgraph_node::create_version_clone): Update.
20300 * tree-inline.c (copy_bb): Update.
20301 (expand_call_inline): Update.
20303 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20305 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20306 factored out from ...
20307 (rs6000_emit_prologue): ... here.
20309 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20311 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20312 factored out from ...
20313 (rs6000_emit_prologue): ... here.
20315 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20317 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20319 (handle_missing_profiles): Fix computation of tp_first_run.
20320 (counts_to_freqs): Do not touch freqs when count is 0.
20322 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20324 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20327 2017-06-10 Tom de Vries <tom@codesourcery.com>
20329 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20330 attributes): Document signal effective target.
20332 2017-06-10 Tom de Vries <tom@codesourcery.com>
20334 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20335 Document effective target stack_size.
20337 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20339 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20340 to the edit_context if they can be auto-applied.
20342 2017-06-9 Ian Lance Taylor <iant@golang.org>
20344 * opts.c (finish_options): If -fsplit-stack, disable implicit
20345 -forder-blocks-and-partition.
20346 * doc/invoke.texi (Optimize Options): Document that when using
20347 -fsplit-stack -forder-blocks-and-partition is not implicitly
20350 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20352 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20353 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20354 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20355 * builtins.def (abort, trap, unreachable): Declare cold.
20356 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20357 * tree-core.h (ECF_COLD): New.
20358 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20359 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20361 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20363 * predict.c (unlikely_executed_stmt_p): Cleanup.
20365 2017-06-09 Richard Biener <rguenther@suse.de>
20367 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20368 model if the ref is always written to.
20370 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20372 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20374 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20376 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20379 2017-06-09 Tom de Vries <tom@codesourcery.com>
20382 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20383 "target cannot support label values" when encountering LABEL_REF.
20385 2017-06-09 Martin Liska <mliska@suse.cz>
20387 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20388 (gimple_gen_ic_func_profiler): Emit direct comparison
20389 of __gcov_indirect_call_callee with NULL.
20390 (gimple_gen_time_profiler): Change probability from
20391 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20393 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20395 * profile.c (edge_gcov_counts): Turn to pointer.
20396 (compute_branch_probabilities, compute_branch_probabilities): Update.
20397 (branch_prob): Do not clear edge_gcov_count.
20398 * profile.h (edge_gcov_counts): Turn to pointer.
20399 (edge_gcov_count): Update.
20401 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20403 * gimple.h (gimple_check_failed): Mark cold.
20405 2017-06-09 Richard Biener <rguenther@suse.de>
20407 PR tree-optimization/66623
20408 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20409 refactor check_reduction into two parts, properly computing
20410 whether we have to check reduction validity for outer loop
20413 2017-06-09 Richard Biener <rguenther@suse.de>
20415 PR tree-optimization/79483
20416 * graphite-scop-detection.c (order): New global.
20417 (get_order): Compute bb to order mapping that satisfies code
20418 generation constraints.
20419 (cmp_pbbs): New helper.
20420 (build_scops): Start domwalk at entry block, sort generated
20423 2017-06-09 Richard Biener <rguenther@suse.de>
20425 PR middle-end/81007
20426 * ipa-polymorphic-call.c
20427 (ipa_polymorphic_call_context::restrict_to_inner_class):
20428 Skip FIELD_DECLs with error_mark_node type.
20429 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20432 2017-06-09 Martin Liska <mliska@suse.cz>
20434 * predict.c (struct branch_predictor): New struct.
20435 (test_prediction_value_range): New test.
20436 (predict_c_tests): New function.
20437 * selftest-run-tests.c (selftest::run_tests): Run the function.
20438 * selftest.h: Declare new tests.
20440 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20443 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20444 gen_add3_insn did not fail.
20445 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20446 r0, construct that number in a temporary reg and add that reg to r0.
20447 If asked to put the result in r0 as well, fail.
20449 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20451 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20452 for early expansion of vec_eqv.
20454 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20456 PR middle-end/81005
20457 * ubsan.c (instrument_null): Avoid pointless code temporary.
20458 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20460 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20464 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20467 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20470 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20472 * predict.c (unlikely_executed_edge_p): Move ahead.
20473 (probably_never_executed_edge_p): Use it.
20475 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20477 PR middle-end/79988
20478 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20479 gimple_call_builtin_p call.
20481 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20483 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20484 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20485 rtl_check_failed_type2, rtl_check_failed_code1,
20486 rtl_check_failed_code2, rtl_check_failed_code_mode,
20487 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20488 rtvec_check_failed_bounds, rtl_check_failed_flag,
20489 _fatal_insn_not_found, _fatal_insn): Likewise.
20490 * tree.h (tree_contains_struct_check_failed,
20491 tree_check_failed, tree_not_check_failed,
20492 tree_class_check_failed, tree_range_check_failed,
20493 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20494 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20495 tree_operand_check_failed, omp_clause_check_failed,
20496 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20499 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20501 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20502 flag_branch_probabilities.
20503 * ipa-inline.c (edge_badness): Likewise.
20504 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20505 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20506 * predict.c (maybe_hot_frequency_p): Likewise.
20507 (probably_never_executed): Likewise.
20508 * sched-ebb.c (schedule_ebbs): Likewise.
20509 * sched-rgn.c (find_single_block_region): Likewise.
20510 * tracer.c (tail_duplicate): Likewise.
20512 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20514 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20515 longer requires x_flag_profile_use.
20517 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20519 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20520 instead of flag_reorder_blocks_and_partition.
20521 * dbxout.c (dbxout_function_end): Likewise.
20522 * dwarf2out.c (gen_subprogram_die): Likewise.
20523 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20524 * hw-doloop.c (reorg_loops): Likewise.
20525 * varasm.c (assemble_start_function,
20526 assemble_end_function): Likewise.
20527 (decide_function_section): Do not check for
20528 flag_reorder_blocks_and_partition.
20530 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20532 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20534 (chkp_get_hard_register_fake_addr_expr): Ditto.
20535 (chkp_build_addr_expr): Add check for hard reg case.
20536 (chkp_parse_array_and_component_ref): Ditto.
20537 (chkp_find_bounds_1): Ditto.
20538 (chkp_process_stmt): Don't generate bounds store for
20541 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20543 * predict.c (maybe_hot_bb_p): Do not check profile status.
20544 (maybe_hot_edge_p): Likewise.
20545 (probably_never_executed): Check for zero counts even if profile
20547 (unlikely_executed_edge_p): New function.
20548 (unlikely_executed_stmt_p): New function.
20549 (unlikely_executed_bb_p): New function.
20550 (set_even_probabilities): Use unlikely predicates.
20551 (combine_predictions_for_bb): Likewise.
20552 (predict_paths_for_bb): Likewise.
20553 (predict_paths_leading_to_edge): Likewise.
20554 (determine_unlikely_bbs): New function.
20555 (estimate_bb_frequencies): Use it.
20556 (compute_function_frequency): Use zero counts even if profile is
20558 * profile-count.h: Fix typo.
20560 2017-08-08 Julia Koval <julia.koval@intel.com>
20562 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20563 _mm512_mask_cvtsepi16_storeu_epi8,
20564 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20565 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20566 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20567 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20568 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20569 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20570 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20571 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20572 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20573 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20574 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20575 __builtin_ia32_pmovuswb256mem_mask,
20576 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20577 __builtin_ia32_pmovwb512mem_mask): New builtins.
20579 2017-08-08 Julia Koval <julia.koval@intel.com>
20581 PR target/73350,80862
20582 * config/i386/subst.md (round): Fix round pattern.
20583 * config/i386/i386.c (ix86_erase_embedded_rounding):
20584 Fix erasing rounding for the fixed pattern.
20586 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20588 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20590 2017-06-08 Martin Liska <mliska@suse.cz>
20592 PR gcov-profile/80911
20593 * gcov.c (block_info::block_info): New constructor.
20595 2017-06-07 Carl Love <cel@us.ibm.com>
20597 * config/rs6000/rs6000-c: The return type of the following
20598 built-in functions was implemented as int not long long. Fix sign
20599 of return value for the unsigned version of vec_mulo and vec_mule.
20600 vector unsigned long long vec_bperm (vector unsigned long long,
20601 vector unsigned char)
20602 vector signed long long vec_mule (vector signed int,
20604 vector unsigned long long vec_mule (vector unsigned int,
20605 vector unsigned int)
20606 vector signed long long vec_mulo (vector signed int,
20608 vector unsigned long long vec_mulo (vector unsigned int,
20609 vector unsigned int)
20610 * doc/extend.texi: Fix the documentation for the built-in
20613 2017-06-07 Carl Love <cel@us.ibm.com>
20616 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20619 2017-06-07 Carl Love <cel@us.ibm.com>
20621 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20622 support, Generate doublehv for signed int/float for BE case only.
20624 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20626 * doc/invoke.texi (mcx16): Rewrite.
20628 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20630 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20631 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20632 *mov<mode>_softfloat, and an anonymous splitter): Use
20633 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20635 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20637 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20639 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20640 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20641 (rs6000_debug_reg_global): Adjust.
20642 (rs6000_init_hard_regno_mode_ok): Adjust.
20643 (rs6000_dbx_register_number): Adjust.
20644 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20645 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20646 Remove SPE_ACC and SPEFSCR.
20647 (REG_ALLOC_ORDER): Ditto.
20648 (FRAME_POINTER_REGNUM): Change to 111.
20649 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20650 (REG_CLASS_NAMES): Ditto.
20651 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20652 (REGISTER_NAMES): Ditto.
20653 (ADDITIONAL_REG_NAMES): Ditto.
20654 (rs6000_reg_names): Ditto.
20655 * config/rs6000/rs6000.md: Renumber some register number
20658 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20660 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20662 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20663 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20665 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20666 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20667 Delete the SPE high registers.
20668 (REG_ALLOC_ORDER): Ditto.
20669 (enum reg_class): Remove SPE_HIGH_REGS.
20670 (REG_CLASS_NAMES): Ditto.
20671 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20672 (REGISTER_NAMES): Ditto.
20673 (rs6000_reg_names): Ditto.
20674 * doc/tm.texi.in: Remove SPE as example.
20675 * doc/tm.texi: Regenerate.
20677 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20679 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20680 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20681 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20682 * config/rs6000/rs6000.md (type): Remove "brinc".
20684 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20686 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20687 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20688 * config/rs6000/linuxspe.h: Delete file.
20689 * config/rs6000/rs6000.md: Don't include spe.md.
20690 * config/rs6000/spe.h: Delete file.
20691 * config/rs6000/spe.md: Delete file.
20692 * config/rs6000/t-rs6000: Remove spe.md.
20694 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20696 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20697 (reg_or_none500mem_operand): Delete.
20698 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20699 instead of reg_or_none500mem_operand.
20701 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20703 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20704 handling of SPE flags.
20705 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20707 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20709 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20711 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20712 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20713 paired_divv2sf3): Similar.
20714 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20715 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20716 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20718 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20719 Rename the paired_* instruction patterns.
20720 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20722 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20723 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20724 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20725 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20726 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20727 PAIRED_VECTOR_MODE.
20728 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20729 (spe_func_has_64bit_regs_p): Delete.
20730 (spe_expand_predicate_builtin): Delete.
20731 (spe_expand_evsel_builtin): Delete.
20732 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20733 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20734 (invalid_e500_subreg): Delete.
20735 (rs6000_legitimize_address): Always force_reg op2 as well, for
20736 paired single memory accesses.
20737 (rs6000_member_type_forces_blk): Delete.
20738 (rs6000_spe_function_arg): Delete.
20739 (rs6000_expand_unop_builtin): Delete SPE handling.
20740 (rs6000_expand_binop_builtin): Ditto.
20741 (spe_expand_stv_builtin): Delete.
20742 (bdesc_2arg_spe): Delete.
20743 (spe_expand_builtin): Delete.
20744 (spe_expand_predicate_builtin): Delete.
20745 (spe_expand_evsel_builtin): Delete.
20746 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20747 (spe_init_builtins): Delete.
20748 (spe_func_has_64bit_regs_p): Delete.
20749 (savres_routine_name): Delete "info" parameter. Adjust callers.
20750 (rs6000_emit_stack_reset): Ditto.
20751 (rs6000_dwarf_register_span): Delete.
20752 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20753 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20754 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20756 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20758 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20759 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20760 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20761 mulv2sf3, divv2sf3): Delete expanders.
20763 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20765 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20767 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20769 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20770 * config/rs6000/rs6000.c: Ditto.
20772 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20774 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20775 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20776 comparison_operator.
20778 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20780 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20781 * config/rs6000/rs6000.opt: Ditto.
20782 * config/rs6000/t-rtems: Ditto.
20784 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20786 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20787 TARGET_E500_SINGLE by 0, simplify.
20788 * config/rs6000/rs6000.c: Ditto.
20789 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20790 (spe_build_register_parallel): Delete.
20791 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20792 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20793 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20794 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20795 (E500_CONVERT): Delete.
20796 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20798 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20800 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20801 * config/rs6000/dfp.md: Ditto.
20802 (negdd2, *negdd2_fpr): Merge.
20803 (absdd2, *absdd2_fpr): Merge.
20804 (negtd2, *negtd2_fpr): Merge.
20805 (abstd2, *abstd2_fpr): Merge.
20806 * config/rs6000/e500.h: Delete file.
20807 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20808 TARGET_FPRS by 1 and simplify.
20809 * config/rs6000/rs6000-c.c: Ditto.
20810 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20811 TARGET_DF_SPE by 0.
20812 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20814 * config/rs6000/rs6000.md: Ditto.
20815 (floatdidf2, *floatdidf2_fpr): Merge.
20816 (move_from_CR_gt_bit): Delete.
20817 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20818 (E500_CR_IOR_COMPARE): Delete.
20819 (All patterns that require !TARGET_FPRS): Delete.
20820 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20822 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20824 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20826 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20828 * graphds.c (add_edge): Intitialize edge's attached data.
20829 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20830 pointer parameter. Call pointed function on each edge during
20831 graph traversing. Skip traversing the edge when the function
20833 (graphds_dfs, graphds_scc): Ditto.
20834 (for_each_edge): New parameter. Pass the new parameter to callback
20836 * graphds.h (skip_edge_callback): New function pointer type.
20837 (graphds_dfs, graphds_scc): New function pointer parameter.
20838 (graphds_edge_callback, for_each_edge): New parameter.
20840 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20842 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20843 out code checking if runtime alias check is possible to below ...
20844 Call the new function.
20845 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20846 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20848 2017-06-07 Marek Polacek <polacek@redhat.com>
20851 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20852 TYPE_OVERFLOW_WRAPS check.
20854 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20856 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20857 if versioning is required.
20858 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20859 peeling with the check for versioning.
20861 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20863 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20864 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20865 Set true to new parameter if new ssa variable is defined.
20866 (vect_gen_vector_loop_niters): Refactor. Set range information
20867 for the new vector loop bound variable.
20868 (vect_do_peeling): Ditto.
20870 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20872 * tree-affine.c (ssa.h): Include header file.
20873 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20874 has wrapping overflow behavior.
20876 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20878 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20880 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20882 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20883 (tree_to_aff_combination): ... here.
20885 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20887 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20888 reg_pressure model function.
20889 (ivopts_global_cost_for_size): Delete.
20890 (determine_set_costs, iv_ca_recount_cost): Call new model function
20891 ivopts_estimate_reg_pressure.
20893 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20895 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20896 expensive than udiv. Remove floating point cases from mod.
20898 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20900 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20901 Increase idiv cost.
20903 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20905 * config/aarch64/aarch64.md
20906 (copysignsf3): Fix mask generation.
20908 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20910 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
20912 (class dump_manager): Add register_dumps method.
20913 * dumpfile.c: Include langhooks.h.
20914 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
20915 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
20916 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
20917 (dump_manager::dump_register): Start with 512 entries instead of 32.
20918 (dump_manager::register_dumps): New method.
20919 * toplev.c (general_init): Instead of invoking register_dumps
20920 langhook, invoke register_dumps method on the dump manager.
20921 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
20924 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
20926 * doc/md.texi: Clarify the restrictions on a define_insn condition.
20927 Say that # requires an associated define_split to exist, and that
20928 the define_split must be suitable for use after register allocation.
20930 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
20932 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
20933 (compute_outgoing_frequencies): Also initialize zero counts.
20934 (find_many_sub_basic_blocks): Do not produce uninitialized profile
20935 around loops; preserve more of profile when nothing changes.
20937 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
20939 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
20941 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
20942 * config/arm/arm-cpu-cdata.h: Regenerate.
20943 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
20944 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
20945 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
20946 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
20947 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
20949 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
20951 * config/arm/t-rmprofile: Likewise.
20952 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
20954 2017-06-06 David S. Miller <davem@davemloft.net>
20957 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
20958 blockage if function uses alloca.
20960 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20962 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
20963 New "uid" fields to hold pretty-print IDs of group and ref.
20964 Memory references are now identified as <group_id>:<ref_id>
20965 instead of using [random] addresses.
20966 (dump_mem_details): Simplify, no functional change.
20967 (dump_mem_ref): Simplify and make output more concise.
20968 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
20969 (find_or_create_group): Initialize group uid.
20970 (record_ref): Initialize ref uid. Improve debug output.
20971 (prune_group_by_reuse, should_issue_prefetch_p,)
20972 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
20973 (mark_nontemporal_store, determine_loop_nest_reuse):
20974 Improve debug output.
20976 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20978 * dbgcnt.def (prefetch): New debug counter.
20979 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
20980 (schedule_prefetches): Stop issueing prefetches if debug counter
20983 2017-06-06 Tom de Vries <tom@codesourcery.com>
20985 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
20986 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
20988 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20990 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
20991 Use aarch64_reg_or_zero predicate for operand 4.
20992 (aarch64_compare_and_swap<mode> define_insn_and_split):
20993 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
20994 (aarch64_store_exclusive<mode>): Likewise for operand 2.
20996 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
20998 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
20999 (arm_compute_save_core_reg_mask): This.
21000 (thumb1_compute_save_reg_mask): Rename into ...
21001 (thumb1_compute_save_core_reg_mask): This.
21002 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21003 (arm_compute_frame_layout): Likewise.
21005 2017-06-06 Richard Biener <rguenther@suse.de>
21007 PR tree-optimization/80974
21008 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21009 keep or clear leaders SSA info.
21011 2017-06-06 Tom de Vries <tom@codesourcery.com>
21013 * config/nvptx/nvptx.c (split_mode_p): New function.
21014 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21016 2017-06-06 Tom de Vries <tom@codesourcery.com>
21018 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21020 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21023 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21026 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21028 * shrink-wrap.c (handle_simple_exit): Update profile.
21029 (try_shrink_wrapping): Upate profile.
21031 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21033 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21034 (tree_guess_outgoing_edge_probabilities): New.
21035 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21036 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21038 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21040 * ipa-split.c (split_function): Initialize return bb profile.
21042 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21044 * profile.c (compute_branch_probabilities): Also initialize
21045 EXIT_BLOCK profile.
21047 2017-06-06 Richard Biener <rguenther@suse.de>
21049 PR tree-optimization/80928
21050 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21051 (vect_analyze_loop_operations): Properly guard analysis for
21053 (vect_transform_loop): Likewise.
21054 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21055 (vect_model_induction_cost): Do not cost for pure SLP.
21056 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21057 of induction in inner loop vectorization.
21058 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21059 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21060 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21062 (vect_analyze_slp_cost_1): Cost induction.
21063 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21064 (vect_get_slp_vect_defs): Likewise.
21065 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21066 (vect_transform_stmt): Handle SLP reductions.
21067 * tree-vectorizer.h (vectorizable_induction): Adjust.
21069 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21071 * config/rs6000/rs6000.c (make_resolver_func): Update
21072 init_lowered_empty_function call.
21074 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21076 * doc/invoke.texi: Document the -fprofile-abs-path option.
21077 * common.opt (fprofile-abs-path): New option.
21078 * gcov-io.h (gcov_write_filename): Declare.
21079 * gcov-io.c (gcov_write_filename): New function.
21080 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21081 * profile.c (output_location): Likewise.
21083 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21085 * shring-wrap.c: Revert accidental commit.
21087 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21089 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21091 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21093 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21095 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21096 profile in callgraph edge.
21097 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21098 (apply_scale): Likewise.
21099 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
21100 Also scale profile when inlining function with zero profile.
21101 (initialize_cfun): Update exit block profile even when it is zero.
21102 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
21103 when profile is read.
21105 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21107 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
21108 (CLONE_*): New constants to define the processors we can generate
21109 code for with the target_clone attribute.
21110 (rs6000_clone_map): New array to identify which clone processors
21111 the current program is running on.
21112 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
21113 target_clone attribute.
21114 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
21115 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
21116 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
21117 (cpu_expand_builtin): Add support for target_clone attribute.
21118 (rs6000_valid_attribute_p): Allow "default" attribute.
21119 (get_decl_name): New debug function to simplify printing the
21120 current function name in debugging statements.
21121 (rs6000_clone_priority): New functions to support the target_clone
21122 attribute, and be able to generate code to switch between ISA 2.05
21123 through ISA 3.0 (power6 through power9).
21124 (rs6000_compare_version_priority): Likewise.
21125 (rs6000_get_function_versions_dispatcher): Likewise.
21126 (make_resolver_func): Likewise.
21127 (add_condition_to_bb): Likewise.
21128 (dispatch_function_versions): Likewise.
21129 (rs6000_generate_version_dispatcher_body): Likewise.
21130 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
21131 (fusion_gpr_load_p): Fix a spacing issue.
21132 * doc/extend.texi (Common Function Attributes): Document that the
21133 PowerPC supports the target_clone attribute.
21135 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21137 * config/arm/arm.h: explain F symbol found in description of ARM
21138 register allocation in its legend.
21140 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21142 * config/mips/frame-header-opt.c: Include profile-count.h.
21143 * config/riscv/riscv.c: Include profile-count.h
21145 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21147 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
21149 (sm_set_flag_if_changed): Add bbs field.
21150 (execute_sm_if_changed_flag_set): Pass BBS.
21151 (execute_sm): Update.
21153 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21155 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
21158 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21160 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
21161 (peephole2): New peephole2 to emit the above.
21162 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21164 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21166 * config/aarch64/aarch64.c (define_peephole2 above
21167 *sub_<shift>_<mode>): New peephole.
21169 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21171 * config/i386/i386.c (make_resolver_func): Update.
21172 * Makefile.in: Add profile-count.h and profile-count.o
21173 * auto-profile.c (afdo_indirect_call): Update to new API.
21174 (afdo_set_bb_count): Update.
21175 (afdo_propagate_edge): Update.
21176 (afdo_propagate_circuit): Update.
21177 (afdo_calculate_branch_prob): Update.
21178 (afdo_annotate_cfg): Update.
21179 * basic-block.h: Include profile-count.h
21180 (struct edge_def): Turn count to profile_count.
21181 (struct basic_block_def): Likewie.
21182 (REG_BR_PROB_BASE): Move to profile-count.h
21183 (RDIV): Move to profile-count.h
21184 * bb-reorder.c (max_entry_count): Turn to profile_count.
21185 (find_traces): Update.
21186 (rotate_loop):Update.
21187 (connect_traces):Update.
21188 (sanitize_hot_paths):Update.
21189 * bt-load.c (migrate_btr_defs): Update.
21190 * cfg.c (RDIV): Remove.
21191 (init_flow): Use alloc_block.
21192 (alloc_block): Uninitialize count.
21193 (unchecked_make_edge): Uninitialize count.
21194 (check_bb_profile): Update.
21195 (dump_edge_info): Update.
21196 (dump_bb_info): Update.
21197 (update_bb_profile_for_threading): Update.
21198 (scale_bbs_frequencies_int): Update.
21199 (scale_bbs_frequencies_gcov_type): Update.
21200 (scale_bbs_frequencies_profile_count): New.
21201 * cfg.h (update_bb_profile_for_threading): Update.
21202 (scale_bbs_frequencies_profile_count): Declare.
21203 * cfgbuild.c (compute_outgoing_frequencies): Update.
21204 (find_many_sub_basic_blocks): Update.
21205 * cfgcleanup.c (try_forward_edges): Update.
21206 (try_crossjump_to_edge): Update.
21207 * cfgexpand.c (expand_gimple_tailcall): Update.
21208 (construct_exit_block): Update.
21209 * cfghooks.c (verify_flow_info): Update.
21210 (dump_bb_for_graph): Update.
21211 (split_edge): Update.
21212 (make_forwarder_block): Update.
21213 (duplicate_block): Update.
21214 (account_profile_record): Update.
21215 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21216 (get_estimated_loop_iterations): Update.
21217 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21218 (single_likely_exit): Update.
21219 * cfgloopmanip.c (scale_loop_profile): Update.
21221 (set_zero_probability): Update.
21222 (lv_adjust_loop_entry_edge): Update.
21223 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21224 (purge_dead_edges): Update.
21225 (rtl_account_profile_record): Update.
21226 * cgraph.c (cgraph_node::create): Uninitialize count.
21227 (symbol_table::create_edge): Uninitialize count.
21228 (cgraph_update_edges_for_call_stmt_node): Update.
21229 (cgraph_edge::dump_edge_flags): Update.
21230 (cgraph_node::dump): Update.
21231 (cgraph_edge::maybe_hot_p): Update.
21232 * cgraph.h: Include profile-count.h
21233 (create_clone), create_edge, create_indirect_edge): Update.
21234 (cgraph_node): Turn count to profile_count.
21235 (cgraph_edge0: Likewise.
21236 (make_speculative, clone): Update.
21237 (create_edge): Update.
21238 (init_lowered_empty_function): Update.
21239 * cgraphclones.c (cgraph_edge::clone): Update.
21240 (duplicate_thunk_for_node): Update.
21241 (cgraph_node::create_clone): Update.
21242 * cgraphunit.c (cgraph_node::analyze): Update.
21243 (cgraph_node::expand_thunk): Update.
21244 * final.c (dump_basic_block_info): Update.
21245 * gimple-streamer-in.c (input_bb): Update.
21246 * gimple-streamer-out.c (output_bb): Update.
21247 * graphite.c (print_global_statistics): Update.
21248 (print_graphite_scop_statistics): Update.
21249 * hsa-brig.c: Include basic-block.h.
21250 * hsa-dump.c: Include basic-block.h.
21251 * hsa-gen.c (T sum_slice): Update.
21252 (convert_switch_statements):Update.
21253 * hsa-regalloc.c: Include basic-block.h.
21254 * ipa-chkp.c (chkp_produce_thunks): Update.
21255 * ipa-cp.c (struct caller_statistics): Update.
21256 (init_caller_stats): Update.
21257 (gather_caller_stats): Update.
21258 (ipcp_cloning_candidate_p): Update.
21259 (good_cloning_opportunity_p): Update.
21260 (get_info_about_necessary_edges): Update.
21261 (dump_profile_updates): Update.
21262 (update_profiling_info): Update.
21263 (update_specialized_profile): Update.
21264 (perhaps_add_new_callers): Update.
21265 (decide_about_value): Update.
21266 (ipa_cp_c_finalize): Update.
21267 * ipa-devirt.c (struct odr_type_warn_count): Update.
21268 (struct decl_warn_count): Update.
21269 (struct final_warning_record): Update.
21270 (possible_polymorphic_call_targets): Update.
21271 (ipa_devirt): Update.
21272 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21273 * ipa-icf.c (sem_function::merge): Update.
21274 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21275 * ipa-inline.c (compute_uninlined_call_time): Update.
21276 (compute_inlined_call_time): Update.
21277 (want_inline_small_function_p): Update.
21278 (want_inline_self_recursive_call_p): Update.
21279 (edge_badness): Update.
21280 (lookup_recursive_calls): Update.
21281 (recursive_inlining): Update.
21282 (inline_small_functions): Update.
21283 (dump_overall_stats): Update.
21284 (dump_inline_stats): Update.
21285 * ipa-profile.c (ipa_profile_generate_summary): Update.
21286 (ipa_propagate_frequency): Update.
21287 (ipa_profile): Update.
21288 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21289 * ipa-utils.c (ipa_merge_profiles): Update.
21290 * loop-doloop.c (doloop_modify): Update.
21291 * loop-unroll.c (report_unroll): Update.
21292 (unroll_loop_runtime_iterations): Update.
21293 * lto-cgraph.c (lto_output_edge): Update.
21294 (lto_output_node): Update.
21295 (input_node): Update.
21296 (input_edge): Update.
21297 (merge_profile_summaries): Update.
21298 * lto-streamer-in.c (input_cfg): Update.
21299 * lto-streamer-out.c (output_cfg): Update.
21300 * mcf.c (create_fixup_graph): Update.
21301 (adjust_cfg_counts): Update.
21302 (sum_edge_counts): Update.
21303 * modulo-sched.c (sms_schedule): Update.
21304 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21305 * predict.c (maybe_hot_count_p): Update.
21306 (probably_never_executed): Update.
21307 (dump_prediction): Update.
21308 (combine_predictions_for_bb): Update.
21309 (propagate_freq): Update.
21310 (handle_missing_profiles): Update.
21311 (counts_to_freqs): Update.
21312 (rebuild_frequencies): Update.
21313 (force_edge_cold): Update.
21314 * predict.h: Include profile-count.h
21315 (maybe_hot_count_p, counts_to_freqs): UPdate.
21316 * print-rtl-function.c: Do not include cfg.h
21317 * print-rtl.c: Include basic-block.h
21318 * profile-count.c: New file.
21319 * profile-count.h: New file.
21320 * profile.c (is_edge_inconsistent): Update.
21321 (correct_negative_edge_counts): Update.
21322 (is_inconsistent): Update.
21323 (set_bb_counts): Update.
21324 (read_profile_edge_counts): Update.
21325 (compute_frequency_overlap): Update.
21326 (compute_branch_probabilities): Update; Initialize and deinitialize
21328 (branch_prob): Update.
21329 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21330 (edge_gcov_count): New.
21331 (bb_gcov_count): New.
21332 * shrink-wrap.c (try_shrink_wrapping): Update.
21333 * tracer.c (better_p): Update.
21334 * trans-mem.c (expand_transaction): Update.
21335 (ipa_tm_insert_irr_call): Update.
21336 (ipa_tm_insert_gettmclone_call): Update.
21337 * tree-call-cdce.c: Update.
21338 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21339 (gimple_duplicate_sese_tail): Update.
21340 (gimple_account_profile_record): Update.
21341 (execute_fixup_cfg): Update.
21342 * tree-inline.c (copy_bb): Update.
21343 (copy_edges_for_bb): Update.
21344 (initialize_cfun): Update.
21345 (freqs_to_counts): Update.
21346 (copy_cfg_body): Update.
21347 (expand_call_inline): Update.
21348 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21349 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21350 (try_unroll_loop_completely): Update.
21351 (try_peel_loop): Update.
21352 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21353 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21354 * tree-ssa-loop-split.c (connect_loops): Update.
21355 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21356 * tree-ssa-reassoc.c (branch_fixup): Update.
21357 * tree-ssa-tail-merge.c (replace_block_by): Update.
21358 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21359 (compute_path_counts): Update.
21360 (update_profile): Update.
21361 (recompute_probabilities): Update.
21362 (update_joiner_offpath_counts): Update.
21363 (estimated_freqs_path): Update.
21364 (freqs_to_counts_path): Update.
21365 (clear_counts_path): Update.
21366 (ssa_fix_duplicate_block_edges): Update.
21367 (duplicate_thread_path): Update.
21368 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21369 (struct switch_conv_info): Update.
21370 * tree-tailcall.c (decrease_profile): Update.
21371 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21372 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21373 * value-prof.c (check_counter): Update.
21374 (gimple_divmod_fixed_value): Update.
21375 (gimple_mod_pow2): Update.
21376 (gimple_mod_subtract): Update.
21377 (gimple_ic_transform): Update.
21378 (gimple_stringop_fixed_value): Update.
21379 * value-prof.h (gimple_ic): Update.
21381 2017-06-02 Carl Love <cel@us.ibm.com>
21383 * config/rs6000/rs6000-c: Add support for built-in functions
21384 vector double vec_doublee (vector signed int);
21385 vector double vec_doublee (vector unsigned int);
21386 vector double vec_doublee (vector float);
21387 vector double vec_doubleh (vector signed int);
21388 vector double vec_doubleh (vector unsigned int);
21389 vector double vec_doubleh (vector float);
21390 vector double vec_doublel (vector signed int);
21391 vector double vec_doublel (vector unsigned int);
21392 vector double vec_doublel (vector float);
21393 vector double vec_doubleo (vector signed int);
21394 vector double vec_doubleo (vector unsigned int);
21395 vector double vec_doubleo (vector float);.
21396 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21397 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21399 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21400 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21401 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21403 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21404 vec_doublel, vec_doubleh.
21405 * doc/extend.texi: Update the built-in documentation file for the
21406 new built-in functions.
21408 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21411 * ipa-inline-analysis.c (free_growth_caches): Set
21412 edge_removal_hook_holder to NULL after removing it.
21414 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21416 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21417 comparision with zero.
21419 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21420 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21421 for early expansion of vec_min and vec_max builtins.
21422 (builtin_function_type): Add min/max unsigned variants to those
21423 identified as having unsigned arguments.
21425 2017-06-02 Olivier Hainque <hainque@adacore.com>
21427 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21429 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21431 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21432 Use VALL_F16 iterator rather than VALL.
21434 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21436 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21437 Emit CBNZ inside loop when doing a strong exchange and comparing
21438 against zero. Generate the CC flags after the loop.
21440 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21442 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21443 (dl_section_ref): New.
21444 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21445 On AIX, append an expression to subtract the size of the
21446 section length to dl_section_ref.
21448 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21450 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21451 for early expansion of vector absolute builtins.
21453 2017-06-02 Richard Biener <rguenther@suse.de>
21455 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21456 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21458 2017-06-02 Richard Biener <rguenther@suse.de>
21460 PR tree-optimization/80948
21461 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21462 stmt order as well.
21464 2017-06-02 Richard Biener <rguenther@suse.de>
21466 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21468 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21469 for inductions as relevant.
21471 2017-06-02 Richard Biener <rguenther@suse.de>
21473 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21474 (vectorizable_induction): ... this. Remove dead code.
21476 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21478 * builtins. (expand_builtin_alloca): Remove second parameter and
21479 infer its value from the first parameter instead.
21480 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21482 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21484 PR rtl-optimization/80903
21485 * loop-doloop.c (add_test): Unshare sequence.
21487 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21489 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21491 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21493 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21495 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21496 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21497 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21498 (xlogue_layout::get_stub_rtx): Make static.
21499 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21500 (xlogue_layout::compute_stub_managed_regs): Rename to...
21501 (xlogue_layout::count_stub_managed_regs): ...this.
21502 (xlogue_layout::is_stub_managed_reg): New function.
21503 (xlogue_layout::m_stub_names): Rename to...
21504 (xlogue_layout::s_stub_names): ...this, make static.
21505 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21506 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21507 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21508 xlogue_layout::s_stub_names): Instantiate statics.
21509 (stub_managed_regs): Remove.
21510 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21511 (disable_call_ms2sysv_xlogues): Rename to...
21512 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21513 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21515 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21516 change after reload_completed.
21517 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21519 (ix86_expand_prologue): Likewise.
21520 (ix86_expand_epilogue): Likewise.
21521 (ix86_expand_split_stack_prologue): Likewise.
21522 (ix86_compute_frame_layout): Remove frame parameter ...
21523 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21524 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21526 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21527 (ix86_frame): Move from here ...
21528 * config/i386/i386.h (ix86_frame): ... to here.
21529 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21530 complete ix86_frame data structure instead. Remove some_ld_name.
21532 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21534 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21535 symbols that hold a DECL_VALUE_EXPR.
21537 2017-06-01 Martin Jambor <mjambor@suse.cz>
21539 PR tree-optimization/80898
21540 * tree-sra.c (process_subtree_disqualification): Removed.
21541 (disqualify_candidate): Do not acll
21542 process_subtree_disqualification.
21543 (subtree_mark_written_and_enqueue): New function.
21544 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21545 RHS has been disqualified and re-queue LHS if necessary. Apart
21546 from that, ignore disqualified RHS.
21548 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21550 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21551 address fetch for z10 or later.
21553 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21555 * config/arc/arc.md (tst_movb): Add guard when splitting.
21557 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21559 * config/arc/arc.c (arc_can_eliminate): Test against
21560 arc_frame_pointer_needed.
21562 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21564 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21565 to prevent store reordering.
21566 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21567 (type): Add block type.
21568 (stack_tie): Define special instruction to be used in
21571 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21573 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21574 constraint. It is not valid for the pattern.
21575 (noncommutative_binary_comparison): Likewise.
21577 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21579 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21582 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21584 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21585 be used by the reg-alloc.
21587 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21589 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21590 reg-alloc when having mul64 or mul32x16 instructions.
21591 (mulsidi3): Likewise.
21592 (umulsidi3): Likewise.
21593 (mulsi32x16): New pattern.
21594 (mulsi64): Likewise.
21595 (mulsidi64): Likewise.
21596 (umulsidi64): Likewise.
21597 (MUL32x16_REG): Define.
21598 (mul64_600): Use MUL32x16_REG.
21599 (mac64_600): Likewise.
21600 (umul64_600): Likewise.
21601 (umac64_600): Likewise.
21603 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21605 * config/arc/arc.md (mulsi3_700): Make it commutative.
21607 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
21609 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
21611 (*sign_extendsidi2_insn): Likewise for movstosw.
21613 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21615 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21616 the type of the input discriminant value. Convert the
21617 discriminant value of signedness vary.
21619 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21621 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21622 Add to -Wall section.
21624 2017-06-01 Richard Biener <rguenther@suse.de>
21626 PR middle-end/66313
21627 * fold-const.c (fold_plusminus_mult_expr): If the factored
21628 factor may be zero use a wrapping type for the inner operation.
21629 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21630 and handle moved defs.
21631 (process_assignment): Properly guard the unary op case. Return a
21632 tri-state indicating that moving the stmt before the call may allow
21633 to continue. Pass through to_move.
21634 (find_tail_calls): Handle moving unrelated defs before
21637 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21640 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21641 splitter result in the canonical way.
21643 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21645 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21646 also for 32bit target. Update insn attributes.
21647 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21649 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21651 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21652 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21653 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21654 (_mm_maskz_min_ss): New intrinsics.
21656 2017-05-31 Martin Liska <mliska@suse.cz>
21658 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21659 Change comment style to one we normally use.
21660 (vectorizable_reduction): Likewise.
21661 (vectorizable_induction): Likewise.
21662 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21663 (vectorizable_call): Likewise.
21664 (vectorizable_simd_clone_call): Likewise.
21665 (vectorizable_conversion): Likewise.
21666 (vectorizable_assignment): Likewise.
21667 (vectorizable_shift): Likewise.
21668 (vectorizable_operation): Likewise.
21669 (vectorizable_store): Likewise.
21670 (vectorizable_load): Likewise.
21671 * tree-vectorizer.h: Likewise.
21673 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21675 * passes.c (emergency_dump_function): New.
21676 * tree-pass.h (emergency_dump_function): Declare.
21677 * plugin.c (plugins_internal_error_function): Remove.
21678 * plugin.h (plugins_internal_error_function): Remove declaration.
21679 * toplev.c (internal_error_function): New static function. Use it...
21680 (general_init): ...here.
21682 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21684 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21685 (arc_rtx_costs): Add costs for new patterns.
21686 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21687 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21689 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21691 * tree-ssa-strlen.c (get_next_strinfo): New function.
21692 (get_stridx_plus_constant): Use it.
21693 (zero_length_string): Likewise.
21694 (adjust_related_strinfos): Likewise.
21695 (adjust_last_stmt): Likewise.
21697 2017-05-31 Richard Biener <rguenther@suse.de>
21700 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21701 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21703 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21705 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21706 loop_vinfo argument and use of dependence distance vectors.
21707 Check instead whether the two references differ only in their
21708 initial value and assume that they have the same alignment if the
21709 difference is a multiple of the vector alignment.
21710 (vect_analyze_data_refs_alignment): Update call accordingly.
21712 2017-05-31 Martin Liska <mliska@suse.cz>
21715 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21717 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21719 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21720 (create_intersect_range_checks): Move from ...
21721 * tree-data-ref.c (create_intersect_range_checks_index)
21722 (create_intersect_range_checks): ... to here.
21723 (create_runtime_alias_checks): New function factored from ...
21724 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21725 here. Call above function.
21726 * tree-data-ref.h (create_runtime_alias_checks): New function.
21728 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21730 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21731 segment length for dr_b and compute it in wide_int.
21733 2017-05-31 Richard Biener <rguenther@suse.de>
21735 PR tree-optimization/80906
21736 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21737 and pass through iv_map.
21738 (copy_bb_and_scalar_dependences): Adjust.
21739 (translate_pending_phi_nodes): Likewise.
21740 (copy_loop_close_phi_args): Handle code-generating IVs instead
21743 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21745 * diagnostic-color.c (color_dict): Add "type-diff".
21746 (parse_gcc_colors): Update comment.
21747 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21748 -fdiagnostics-show-template-tree and -fno-elide-type.
21749 (GCC_COLORS): Add type-diff to example.
21751 (-fdiagnostics-show-template-tree): New.
21752 (-fno-elide-type): New.
21753 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21754 the pp_format_decoder callback. Call any m_format_postprocessor's
21756 (pretty_printer::pretty_printer): Initialize
21757 m_format_postprocessor.
21758 (pretty_printer::~pretty_printer): Delete any
21759 m_format_postprocessor.
21760 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21761 (class format_postprocessor): New class.
21762 (struct pretty_printer::format_decoder): Document the new parameters.
21763 (struct pretty_printer::m_format_postprocessor): New field.
21764 * tree-diagnostic.c (default_tree_printer): Update for new
21765 bool and const char ** params.
21766 * tree-diagnostic.h (default_tree_printer): Likewise.
21768 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21770 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21771 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21772 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21773 rs6000_gen_cell_microcode code.
21774 (rs6000_final_prescan_insn): Delete.
21775 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21776 "warn-cell-microcode" entries.
21777 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21778 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21779 throughout. Change cc_reg_not_micro_cr0_operand to
21780 cc_reg_not_cr0_operand throughout.
21781 (*extendhi<mode>2_noload): Delete.
21782 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21783 (mwarn-cell-microcode): Delete.
21784 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21785 -mgen-cell-microcode and -mwarn-cell-microcode.
21787 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21790 * config/i386/constraints.md (Yd): New constraint.
21792 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21793 and (?Yd, r) alternatives. Update insn attributes.
21794 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21795 and (?*Yd, r) alternatives. Update insn attributes.
21796 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21798 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21800 * gimplify.c (gimplify_modify_expr): Don't create a
21801 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21804 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21806 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21808 2017-05-30 Richard Biener <rguenther@suse.de>
21810 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21811 and reduc_def fields.
21812 (STMT_VINFO_REDUC_TYPE): New define.
21813 (STMT_VINFO_REDUC_DEF): Likewise.
21814 (vect_force_simple_reduction): Adjust prototype.
21815 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21816 (vect_is_simple_reduction): Remove check_reduction argument.
21817 (vect_force_simple_reduction): Adjust and set
21818 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21819 (vectorizable_reduction): Do not re-do reduction analysis
21820 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21821 * tree-parloops.c (gather_scalar_reductions): Adjust.
21823 2017-05-30 Richard Biener <rguenther@suse.de>
21825 PR middle-end/80901
21826 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21829 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21831 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21832 Introduce unknown_misalignment parameter and remove vf.
21833 (vect_peeling_hash_get_lowest_cost):
21834 Pass unknown_misalignment parameter.
21835 (vect_enhance_data_refs_alignment):
21836 Fix unsupportable data ref treatment.
21838 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21840 * tree-vect-data-refs.c (vect_get_data_access_cost):
21841 Workaround for SLP handling.
21842 (vect_enhance_data_refs_alignment):
21843 Compute costs for doing no peeling at all, compare to the best
21844 peeling costs so far and avoid peeling if cheaper.
21846 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21848 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21849 Return peeling info and set costs to zero for unlimited cost
21851 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21852 with unknown misalignment. Compute and costs for unknown
21853 misalignment, compare them to the costs for known misalignment
21854 and choose the cheapest for peeling.
21856 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21858 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21859 (vect_get_peeling_costs_all_drs): Create function.
21860 (vect_peeling_hash_get_lowest_cost):
21861 Use vect_get_peeling_costs_all_drs.
21862 (vect_peeling_supportable): Create function.
21863 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21865 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21867 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21868 DR_HAS_NEGATIVE_STEP.
21869 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21870 (vect_enhance_data_refs_alignment): Use.
21871 (vect_duplicate_ssa_name_ptr_info): Use.
21872 * tree-vectorizer.h (dr_misalignment): Use.
21873 (known_alignment_for_access_p): Use.
21875 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21878 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21880 (has_section_name): New function.
21882 2017-05-30 Martin Liska <mliska@suse.cz>
21885 * auto-profile.c (get_function_decl_from_block): Fix
21888 2017-05-30 Richard Biener <rguenther@suse.de>
21890 PR middle-end/80876
21891 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21893 2017-05-30 Martin Liska <mliska@suse.cz>
21895 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21896 * dumpfile.h (struct dump_file_info): Remove ctors.
21898 2017-05-30 Martin Liska <mliska@suse.cz>
21900 * predict.def: Fix GNU coding style.
21902 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21904 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21905 Mark 'to' argument with ATTRIBUTE_UNUSED.
21907 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21909 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21910 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
21911 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
21914 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
21916 * doc/install.texi (Options specification): Restore entry of
21917 --enable-sjlj-exceptions.
21919 2017-05-27 Michael Eager <eager@eagercon.com>
21922 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
21924 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
21926 * config/microblaze/microblaze.h
21927 (FIXED_REGISTERS): Update in macro.
21928 (CALL_USED_REGISTERS): Update in macro.
21930 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
21932 * doc/install.texi: Add links to macOS binary distributions.
21934 2017-05-27 Jakub Jelinek <jakub@redhat.com>
21938 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
21940 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
21942 2017-05-26 Martin Liska <mliska@suse.cz>
21944 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
21946 2017-05-26 Martin Liska <mliska@suse.cz>
21948 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
21949 always leading ';; '.
21950 (dump_bb_info): Likewise.
21951 (brief_dump_cfg): Likewise.
21952 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
21953 * dumpfile.c: Remove usage of TDF_VERBOSE.
21954 * dumpfile.h (enum dump_kind): Likewise.
21955 (dump_gimple_bb_header): Do not use TDF_COMMENT.
21956 * print-tree.c (debug_verbose): Remove.
21957 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
21958 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
21959 * tree-diagnostic.c (default_tree_printer): Replace
21960 TDF_DIAGNOSTIC with TDF_SLIM.
21962 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21964 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
21965 in parameter loop, rather than loop_vinfo.
21966 (create_intersect_range_checks): Ditto.
21967 (vect_create_cond_for_alias_checks): Update call to above functions.
21969 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21971 PR tree-optimization/80815
21972 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
21973 for merging runtime alias checks. Handle negative DR_STEPs.
21975 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21977 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
21979 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
21980 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
21981 out code pruning runtime alias checks.
21982 * tree-data-ref.c (prune_runtime_alias_test_list): New function
21983 factored out from above.
21984 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21986 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
21988 (prune_runtime_alias_test_list): New decalaration.
21990 2017-05-26 Bin Cheng <bin.cheng@arm.com>
21992 * tree-vect-data-refs.c (compare_tree): Rename and move ...
21993 * tree-data-ref.c (data_ref_compare_tree): ... to here.
21994 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
21995 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
21996 (operator==, comp_dr_with_seg_len_pair): Ditto.
21997 (vect_prune_runtime_alias_test_list): Ditto.
21999 2017-05-26 Martin Liska <mliska@suse.cz>
22002 * params.def: Bound partial-inlining-entry-probability param.
22004 2017-05-26 Marek Polacek <polacek@redhat.com>
22007 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22010 2017-05-26 Richard Biener <rguenther@suse.de>
22012 PR tree-optimization/80842
22013 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22016 2017-05-26 Richard Biener <rguenther@suse.de>
22018 PR tree-optimization/80844
22019 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22021 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22023 * doc/md.texi (Machine Constraints): Update x86 family
22024 machine constraints section to match 'config/i386/constraints.md'.
22026 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22028 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22030 2017-05-25 Nathan Sidwell <nathan@acm.org>
22032 * doc/invoke.texi (--enable-languages): Update documentation.
22034 2017-05-25 Martin Liska <mliska@suse.cz>
22036 * dumpfile.c: Add TDF_FOLDING.
22037 * dumpfile.h (enum dump_kind): Likewise.
22038 * genmatch.c (dt_simplify::gen_1): Use it.
22040 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22042 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22044 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22046 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22047 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22049 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22051 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22052 * match.pd (X == C): Rewrite it here.
22053 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22054 with_certain_nonzero_bits2): New predicates.
22055 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22057 2017-05-24 Nathan Sidwell <nathan@acm.org>
22059 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22062 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22065 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22067 * config/powerpcspe: New port. Files are copied from the rs6000
22068 port, with "rs6000" in filenames replaced by "powerpcspe".
22070 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22072 PR rtl-optimization/80754
22073 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22075 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22077 * config/sparc/sparc.md (length): Return the correct value for -mflat
22078 sibcalls to match output_sibcall.
22080 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22084 * config/rs6000/rs6000.c (struct machine_function): Add new field
22086 (rs6000_get_separate_components): Init that field, use it.
22087 (rs6000_components_for_bb): Use the field.
22089 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22091 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22093 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22095 PR middle-end/80823
22096 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22098 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22101 * config/s390/s390.c (s390_check_qrst_address): Check incoming
22102 address against address_operand predicate.
22103 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
22105 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
22107 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
22108 parameters passed indirectly.
22110 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
22112 * config/i386/i386.md (*movdi_internal): Remove SSE4
22113 alternative 18 (?r, *v). Update insn attributes.
22114 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
22115 Update insn attributes.
22116 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
22117 Update insn attributes.
22118 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
22119 alternative 1 (r, v). Remove isa attribute.
22120 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
22121 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
22122 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
22124 2017-05-23 Tom de Vries <tom@codesourcery.com>
22126 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
22129 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22131 * cgraphunit.c (symbol_table::process_new_functions): Update.
22132 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
22133 (inline_generate_summary): Rename to ...
22134 (ipa_fn_summary_generate): ... this one.
22135 (inline_read_summary): Rename to ...
22136 (ipa_fn_summary_read): ... this one.
22137 (inline_write_summary): Rename to ...
22138 (ipa_fn_summary_write): ... this one.
22139 (inline_free_summary): Rename to ...
22140 (ipa_free_fn_summary): ... this one.
22141 (pass_data_local_fn_summary, pass_local_fn_summary,
22142 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
22143 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
22144 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
22145 make_pass_ipa_fn_summary): New.
22146 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
22147 inline_write_summary, inline_free_summary): Remove.
22148 (ipa_free_fn_summary) : New.
22149 * ipa-inline.c (ipa_inline): Update.
22150 (pass_ipa_inline): Do not generate summaries.
22151 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
22153 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
22154 and add pass_ipa_fn_summary.
22155 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
22157 (make_pass_inline_parameters): Remove.
22159 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
22161 * omp-low.c (struct omp_context): Remove "default_kind" member.
22164 * omp-offload.c (execute_oacc_device_lower): Remove the
22165 parallelism dimensions function attributes for unparallelized
22166 OpenACC kernels constructs.
22168 2017-05-23 Martin Liska <mliska@suse.cz>
22170 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22172 (cgraph_edge::make_speculative): Likewise.
22173 (cgraph_edge::resolve_speculation): Likewise.
22174 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22175 (cgraph_node::dump): Likewise.
22176 * cgraph.h: Likewise.
22177 * cgraphunit.c (analyze_functions): Likewise.
22178 (symbol_table::compile): Likewise.
22179 * ipa-cp.c (print_all_lattices): Likewise.
22180 (determine_versionability): Likewise.
22181 (initialize_node_lattices): Likewise.
22182 (ipcp_verify_propagated_values): Likewise.
22183 (estimate_local_effects): Likewise.
22184 (update_profiling_info): Likewise.
22185 (create_specialized_node): Likewise.
22186 (perhaps_add_new_callers): Likewise.
22187 (decide_about_value): Likewise.
22188 (decide_whether_version_node): Likewise.
22189 (identify_dead_nodes): Likewise.
22190 (ipcp_store_bits_results): Likewise.
22191 * ipa-devirt.c (dump_targets): Likewise.
22192 (ipa_devirt): Likewise.
22193 * ipa-icf.c (sem_item::dump): Likewise.
22194 (sem_function::equals): Likewise.
22195 (sem_variable::equals): Likewise.
22196 (sem_item_optimizer::read_section): Likewise.
22197 (sem_item_optimizer::execute): Likewise.
22198 (congruence_class::dump): Likewise.
22199 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22200 (dump_inline_summary): Likewise.
22201 (estimate_node_size_and_time): Likewise.
22202 (inline_analyze_function): Likewise.
22203 * ipa-inline-transform.c (inline_call): Likewise.
22204 * ipa-inline.c (report_inline_failed_reason): Likewise.
22205 (want_early_inline_function_p): Likewise.
22206 (edge_badness): Likewise.
22207 (update_edge_key): Likewise.
22208 (inline_small_functions): Likewise.
22209 * ipa-profile.c (ipa_profile): Likewise.
22210 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22211 (ipa_make_edge_direct_to_target): Likewise.
22212 (remove_described_reference): Likewise.
22213 (ipa_impossible_devirt_target): Likewise.
22214 (propagate_controlled_uses): Likewise.
22215 (ipa_print_node_params): Likewise.
22216 (ipcp_transform_function): Likewise.
22217 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22218 (propagate_pure_const): Likewise.
22219 * ipa-reference.c (generate_summary): Likewise.
22220 (read_write_all_from_decl): Likewise.
22221 (propagate): Likewise.
22222 (ipa_reference_read_optimization_summary): Likewise.
22223 * ipa-utils.c (ipa_merge_profiles): Likewise.
22224 * ipa.c (walk_polymorphic_call_targets): Likewise.
22225 (symbol_table::remove_unreachable_nodes): Likewise.
22226 (ipa_single_use): Likewise.
22227 * passes.c (execute_todo): Likewise.
22228 * predict.c (drop_profile): Likewise.
22229 * symtab.c (symtab_node::get_dump_name): New function.
22230 (symtab_node::dump_name): Likewise.
22231 (symtab_node::dump_asm_name): Likewise.
22232 (symtab_node::dump_references): Likewise.
22233 (symtab_node::dump_referring): Likewise.
22234 (symtab_node::dump_base): Likewise.
22235 (symtab_node::debug_symtab): Likewise.
22236 * tree-sra.c (convert_callers_for_node): Likewise.
22237 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22238 * value-prof.c (init_node_map): Likewise.
22240 2017-05-23 Martin Liska <mliska@suse.cz>
22242 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22243 and symtab_node::debug_symtab to symbol_table::debug.
22244 * cgraphunit.c (analyze_functions): Use the renamed function.
22245 (symbol_table::compile): Likewise.
22246 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22247 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22248 * passes.c (execute_todo): Likewise.
22249 * symtab.c (symbol_table::dump): New function.
22250 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22252 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22254 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22255 that nonconst implies exec.
22257 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22259 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22260 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22261 (inline_edge_summary_vec): Turn into ...
22262 (ipa_call_summaries): ... this one.
22263 (redirect_to_unreachable, edge_set_predicate,
22264 evaluate_properties_for_edge, inline_summary_alloc,
22265 reset_ipa_call_summary, reset_inline_summary,
22266 inline_summary_t::duplicate): Update.
22267 (inline_edge_duplication_hook): Turn to ...
22268 (ipa_call_summary_t::duplicate): ... this one.
22269 (inline_edge_removal_hook): Turn to ...
22270 (ipa_call_summary_t::remove): ... this one.
22271 (dump_inline_edge_summary): Turn to ...
22272 (dump_ipa_call_summary): ... this one.
22273 (estimate_function_body_sizes): Update.
22274 (inline_update_callee_summaries): Update.
22275 (remap_edge_change_prob): Update.
22276 (remap_edge_summaries): Update.
22277 (inline_merge_summary): Update.
22278 (do_estimate_edge_time): Update.
22279 (inline_generate_summary): Update.
22280 (inline_read_section): Update.
22281 (inline_read_summary): Update.
22282 (inline_free_summary): Update.
22283 * ipa-inline.c (can_inline_edge_p): Update.
22284 (compute_inlined_call_time): Update.
22285 (want_inline_small_function_p): Update.
22286 (edge_badness): Update.
22287 (early_inliner): Update.
22288 * ipa-inline.h (inline_edge_summary): Turn to ...
22289 (ipa_call_summary): ... this one.
22290 (ipa_call_summary_t): New class.
22291 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22292 (ipa_call_summaries): New.
22293 (inline_edge_summary): Remove.
22294 (estimate_edge_growth): Update.
22295 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22296 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22297 * ipa-split.c (execute_split_functions): Update.
22298 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22300 2017-05-23 Tom de Vries <tom@codesourcery.com>
22302 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22303 attributes): Document rdrand effective target.
22305 2017-05-23 Tom de Vries <tom@codesourcery.com>
22307 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22308 attributes): Sort alphabetically.
22310 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22312 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22314 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22317 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22318 V2DF/V2DI splat into two separate patterns, one that handles
22319 registers, and the other that only handles memory. Drop support
22320 for splatting from a GPR on ISA 2.07 and then splitting the
22321 splat into direct move and splat.
22322 (vsx_splat_<mode>_reg): Likewise.
22323 (vsx_splat_<mode>_mem): Likewise.
22325 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22327 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22329 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22331 PR middle-end/80809
22332 * omp-low.c (finish_taskreg_remap): New function.
22333 (finish_taskreg_scan): If unit size of ctx->record_type
22334 is non-constant, unshare the size expression and replace
22335 decls in it with possible outer var refs.
22337 PR middle-end/80809
22338 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22339 GOVD_SHARED rather than GOVD_PRIVATE with it.
22340 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22341 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22343 PR middle-end/80853
22344 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22345 as last argument to build_outer_var_ref for pointer bases of array
22346 section reductions.
22348 2017-05-19 Martin Sebor <msebor@redhat.com>
22350 * print-tree.c (print_node): Print DECL_READ_P flag.
22352 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22354 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22355 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22356 * cgraph.c: Likewise.
22357 * cgraphunit.c: Likewise.
22358 * gengtype.c: Likewise.
22359 * ipa-cp.c: Likewise.
22360 * ipa-devirt.c: Likewise.
22361 * ipa-icf.c: Likewise.
22362 * ipa-predicate.c: Likewise.
22363 * ipa-profile.c: Likewise.
22364 * ipa-prop.c: Likewise.
22365 * ipa-split.c: Likewise.
22367 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22368 edge_predicate_pool, dump_inline_hints,
22369 inline_summary::account_size_time, redirect_to_unreachable,
22370 edge_set_predicate, set_hint_predicate,
22371 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22372 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22373 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22374 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22375 ipa_call_summary_t::remove, initialize_growth_caches,
22376 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22377 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22378 mark_modified, unmodified_parm_1, unmodified_parm,
22379 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22380 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22381 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22382 will_be_nonconstant_predicate, record_modified_bb_info,
22383 get_minimal_bb, record_modified, param_change_prob,
22384 phi_result_unknown_predicate, predicate_for_phi_result,
22385 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22386 estimate_function_body_sizes, compute_inline_parameters,
22387 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22388 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22389 inline_update_callee_summaries, remap_edge_change_prob,
22390 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22391 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22392 inline_analyze_function, inline_summary_t::insert,
22393 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22394 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22395 inline_free_summary): Move to ipa-fnsummary.h
22396 (predicate_t): Remove.
22397 * ipa-fnsummary.c: New file.
22398 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22399 (enum inline_hints_vals, inline_hints, agg_position_info,
22400 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22401 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22402 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22403 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22404 inline_read_summary, inline_write_summary, inline_free_summary,
22405 inline_analyze_function, initialize_inline_failed,
22406 inline_merge_summary, inline_update_overall_summary,
22407 compute_inline_parameters): Move to ipa-fnsummary.h
22408 * ipa-fnsummary.h: New file.
22409 * ipa-inline-transform.h: Include ipa-inline.h.
22410 * ipa-inline.c: LIkewise.
22412 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22414 * ipa-inline.c (edge_badness): Use inlined_time instead of
22415 inline_summaries->get.
22417 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22419 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22421 2017-05-22 Nathan Sidwell <nathan@acm.org>
22423 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22424 (fdump-lang): Document 'raw' option.
22425 * dumpfile.h (TDI_tu): Delete.
22426 * dumpfile.c (dump_files): Remove translation-unit.
22427 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22429 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22431 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22432 command option from $(AWK) call.
22433 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22434 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22435 [FORMAT]: Remove handling of variable.
22436 * config/avr/t-multilib: Regenerate.
22438 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22440 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22442 (dump_inline_summary): Do not print self_time.
22443 (estimate_function_body_sizes): Do not set self_time.
22444 (compute_inline_parameters): Likewise.
22445 (inline_read_section, inline_write_summary): Do not stream self_time.
22446 * ipa-inline.h (inline_summary): Drop self_time.
22448 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22450 * ipa-inline-analysis.c (account_size_time): Rename to ...
22451 (inline_summary::account_size_time): ... this one.
22452 (reset_ipa_call_summary): Turn to ...
22453 (ipa_call_summary::reset): ... this one.
22454 (reset_inline_summary): Turn to ...
22455 (inline_summary::reset): ... this one.
22456 (inline_summary_t::remove): Update.
22457 (inline_summary_t::duplicate): Update.
22458 (ipa_call_summary_t::remove): Update.
22459 (dump_inline_summary): Update.
22460 (estimate_function_body_sizes): Update.
22461 (compute_inline_parameters): Update.
22462 (estimate_node_size_and_time): Update.
22463 (inline_merge_summary): Update.
22464 (inline_update_overall_summary): Update.
22465 (inline_read_section): Update.
22466 (inline_write_summary): Update.
22467 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22468 add account_size_time and reset member functions.
22469 (ipa_call_summary): Add reset function.
22470 * ipa-predicate.h (predicate::operator &): Constify.
22472 2017-05-22 Richard Biener <rguenther@suse.de>
22474 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22476 2017-05-19 Jason Merrill <jason@redhat.com>
22478 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22480 2017-05-19 Marek Polacek <polacek@redhat.com>
22483 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22484 TYPE_OVERFLOW_WRAPS checks.
22486 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22488 * tree-core.h (enum omp_clause_default_kind): Add
22489 "OMP_CLAUSE_DEFAULT_PRESENT".
22490 * tree-pretty-print.c (dump_omp_clause): Handle it.
22491 * gimplify.c (enum gimplify_omp_var_data): Add
22492 "GOVD_MAP_FORCE_PRESENT".
22493 (gimplify_adjust_omp_clauses_1): Map it to
22494 "GOMP_MAP_FORCE_PRESENT".
22495 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22497 * gimplify.c (oacc_default_clause): Clarify.
22499 2017-05-19 Nathan Sidwell <nathan@acm.org>
22501 LANG_HOOK_REGISTER_DUMPS
22502 * toplev.c (general_init): Call register dump lang hook.
22503 * doc/invoke.texi: Document -fdump-lang option family.
22504 * dumpfile.c (dump_files): Remove class dump here.
22505 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22506 * dumpfile.h (tree_dump_index): Remove TDI_class.
22507 * langhooks-def.h (lhd_register_dumps): Declare.
22508 (LANG_HOOKS_REGISTER_DUMPS): Define.
22509 (LANG_HOOKS_INITIALIZER): Add it.
22510 * langhooks.c (lhd_register_dumps): Define.
22511 * langhooks.h (struct lang_hooks): Add register_dumps.
22513 2017-05-19 Nathan Sidwell <nathan@acm.org>
22515 * context.h (context::set_passes): New.
22516 * context.c (context::context): Do not create pass manager.
22517 * toplev.c (general_init): Create pass manager here.
22519 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22521 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22522 use this splitter if two add or or instructions would also work for
22523 the constant we want to generate.
22525 2017-05-19 Richard Biener <rguenther@suse.de>
22528 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22529 predicate evaluation.
22531 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22533 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22535 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22538 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22540 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22541 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22542 (inline_edge_summary_vec): Turn into ...
22543 (ipa_call_summaries): ... this one.
22544 (redirect_to_unreachable, edge_set_predicate,
22545 evaluate_properties_for_edge, inline_summary_alloc,
22546 reset_ipa_call_summary, reset_inline_summary,
22547 inline_summary_t::duplicate): Update.
22548 (inline_edge_duplication_hook): Turn to ...
22549 (ipa_call_summary_t::duplicate): ... this one.
22550 (inline_edge_removal_hook): Turn to ...
22551 (ipa_call_summary_t::remove): ... this one.
22552 (dump_inline_edge_summary): Turn to ...
22553 (dump_ipa_call_summary): ... this one.
22554 (estimate_function_body_sizes): Update.
22555 (inline_update_callee_summaries): Update.
22556 (remap_edge_change_prob): Update.
22557 (remap_edge_summaries): Update.
22558 (inline_merge_summary): Update.
22559 (do_estimate_edge_time): Update.
22560 (inline_generate_summary): Update.
22561 (inline_read_section): Update.
22562 (inline_read_summary): Update.
22563 (inline_free_summary): Update.
22564 * ipa-inline.c (can_inline_edge_p): Update.
22565 (compute_inlined_call_time): Update.
22566 (want_inline_small_function_p): Update.
22567 (edge_badness): Update.
22568 (early_inliner): Update.
22569 * ipa-inline.h (inline_edge_summary): Turn to ...
22570 (ipa_call_summary): ... this one.
22571 (ipa_call_summary_t): New class.
22572 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22573 (ipa_call_summaries): New.
22574 (inline_edge_summary): Remove.
22575 (estimate_edge_growth): Update.
22576 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22577 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22578 * ipa-split.c (execute_split_functions): Update.
22579 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22581 2017-05-19 Richard Biener <rguenther@suse.de>
22583 PR middle-end/80764
22584 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22586 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22588 * config/rs6000/rs6000.c (struct machine_function): Add field
22589 fpr_is_wrapped_separately.
22590 (rs6000_get_separate_components): Use 64 components. Handle the
22591 new FPR components.
22592 (rs6000_components_for_bb): Handle the FPR components.
22593 (rs6000_emit_prologue_components): Handle the FPR components.
22594 (rs6000_emit_epilogue_components): Handle the FPR components.
22595 (rs6000_set_handled_components): Handle the FPR components.
22596 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22597 that are already separately shrink-wrapped.
22598 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
22599 that are already separately shrink-wrapped.
22601 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22604 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
22607 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
22608 (define_peephole2 for Altivec d-form load): Add peepholes to catch
22609 cases where the register allocator uses a move and an offsettable
22610 memory operation to/from a FPR register on ISA 2.06/2.07.
22611 (define_peephole2 for Altivec d-form store): Likewise.
22613 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22616 * config/i386/mmx.md (*mov<mode>_internal): Enable
22617 alternatives 11, 12, 13 and 14 also for 32bit targets.
22618 Remove alternatives 15, 16, 17 and 18.
22619 * config/i386/sse.md (vec_concatv2di): Change
22620 alternative (!x, *y) to (x, ?!*Yn).
22622 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22624 * dumpfile.h (enum dump_kind): Remove stray comma.
22626 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22628 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22629 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22630 predicate::num_conditions
22631 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22632 (CHANGED): turn into predicate::changed.
22633 (agg_position_info): Move to ipa-predicate.h
22634 (add_condition, predicate::add_clause, predicate::operator &=,
22635 predicate::or_with, predicate::evaluate, predicate::probability,
22636 dump_condition, dump_clause, predicate::dump,
22637 predicate::remap_after_duplication, predicate::remap_after_inlining,
22638 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22639 (evaluate_conditions_for_known_args): Update.
22640 (set_cond_stmt_execution_predicate): Update.
22641 * ipa-inline.h: Include ipa-predicate.h
22642 (condition, inline_param_summary, conditions, agg_position_info,
22643 predicate): Move to ipa-predicate.h
22644 * ipa-predicate.c: New file.
22645 * ipa-predicate.h: New file.
22647 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22649 * final.c (leaf_function_p): Check we are not in a sequence.
22651 2017-05-18 Martin Liska <mliska@suse.cz>
22653 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22654 * dumpfile.c (dump_register): Use new enum dump_kind.
22655 (get_dump_file_name): Likewise.
22656 (dump_enable_all): Likewise.
22657 (dump_switch_p_1): Likewise.
22658 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22659 * dumpfile.h (enum dump_kind): New enum type.
22660 (struct dump_file_info): Create constructor and
22661 format fields and comments.
22662 * passes.c (pass_manager::register_one_dump_file):
22664 * statistics.c (statistics_early_init): Likewise.
22665 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22666 TDF_TREE with TDF_SLIM.
22667 (gather_memory_references_ref): Likewise.
22669 2017-05-18 Martin Liska <mliska@suse.cz>
22671 * vec.h (struct vnull): Use it.
22673 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22675 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22676 (true_predicate, false_predicate, true_predicate_p,
22677 false_predicate_p): Remove.
22678 (single_cond_predicate, not_inlined_predicate): Turn to member function
22680 (add_condition): Update.
22681 (add_clause): Turn to...
22682 (predicate::add_clause): ... this one; update; allow passing NULL
22684 (and_predicates): Turn to ...
22685 (predicate::operator &=): ... this one.
22686 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22687 (or_predicates): Turn to ...
22688 (predicate::or_with): ... this one.
22689 (evaluate_predicate): Turn to ...
22690 (predicate::evaluate): ... this one.
22691 (predicate_probability): Turn to ...
22692 (predicate::probability): ... this one.
22693 (dump_condition): Update.
22694 (dump_predicate): Turn to ...
22695 (predicate::dump): ... this one.
22696 (account_size_time): Update.
22697 (edge_set_predicate): Update.
22698 (set_hint_predicate): UPdate.
22699 (evaluate_conditions_for_known_args): Update.
22700 (evaluate_properties_for_edge): Update.
22701 (remap_predicate_after_duplication): Turn to...
22702 (predicate::remap_after_duplication): ... this one.
22703 (remap_hint_predicate_after_duplication): Update.
22704 (inline_summary_t::duplicate): UPdate.
22705 (dump_inline_edge_summary): Update.
22706 (dump_inline_summary): Update.
22707 (set_cond_stmt_execution_predicate): Update.
22708 (set_switch_stmt_execution_predicate): Update.
22709 (compute_bb_predicates): Update.
22710 (will_be_nonconstant_expr_predicate): Update.
22711 (will_be_nonconstant_predicate): Update.
22712 (phi_result_unknown_predicate): Update.
22713 (predicate_for_phi_result): Update.
22714 (array_index_predicate): Update.
22715 (estimate_function_body_sizes): Update.
22716 (estimate_node_size_and_time): Update.
22717 (estimate_ipcp_clone_size_and_time): Update.
22718 (remap_predicate): Rename to ...
22719 (predicate::remap_after_inlining): ... this one.
22720 (remap_hint_predicate): Update.
22721 (inline_merge_summary): Update.
22722 (inline_update_overall_summary): Update.
22723 (estimate_size_after_inlining): Update.
22724 (read_predicate): Rename to ...
22725 (predicate::stream_in): ... this one.
22726 (read_inline_edge_summary): Update.
22727 (write_predicate): Rename to ...
22728 (predicate::stream_out): ... this one.
22729 (write_inline_edge_summary): Update.
22730 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22731 (clause_t): Turn to uint32_t
22732 (predicate): Turn to class; implement constructor and operators
22734 (size_time_entry): Update.
22735 (inline_summary): Update.
22736 (inline_edge_summary): Update.
22738 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22740 * fold-const.c (fold_binary_loc): Move transformation...
22741 * match.pd (C - X CMP X): ... here.
22743 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22745 * config/sparc/sparc.c (sparc_option_override): Set function
22746 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22747 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22749 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22751 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22753 2017-05-18 Marek Polacek <polacek@redhat.com>
22756 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22757 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22758 gimple_assign_load_p.
22760 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22762 PR middle-end/80692
22763 * real.c (do_compare): Give decimal_do_compare preference over
22764 comparing just the signs.
22766 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22768 * doc/md.texi (Canonicalization of Instructions): Describe the
22769 canonical form of instructions that inherently set a condition
22772 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22774 PR middle-end/80775
22775 * tree-cfg.c: Move deletion of unreachable case statements to after
22776 the merging of consecutive case labels.
22778 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22780 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22781 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22782 restoring of callee-saved registers.
22784 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22786 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22787 * config/visium/visium.c (single_set_and_flags): Likewise.
22788 * config/visium/visium.md (Substitutions): Likewise.
22790 2017-05-17 Martin Liska <mliska@suse.cz>
22792 * cfg.c: Introduce dump_flags_t type and
22793 use it instead of int type.
22795 * cfghooks.c: Likewise.
22796 * cfghooks.h (struct cfg_hooks): Likewise.
22797 * cfgrtl.c: Likewise.
22798 * cfgrtl.h: Likewise.
22799 * cgraph.c (cgraph_node::get_body): Likewise.
22800 * coretypes.h: Likewise.
22801 * domwalk.c: Likewise.
22802 * domwalk.h: Likewise.
22803 * dumpfile.c (struct dump_option_value_info): Likewise.
22804 (dump_enable_all): Likewise.
22805 (dump_switch_p_1): Likewise.
22806 (opt_info_switch_p): Likewise.
22807 * dumpfile.h (enum tree_dump_index): Likewise.
22808 (struct dump_file_info): Likewise.
22809 * genemit.c: Likewise.
22810 * generic-match-head.c: Likewise.
22811 * gengtype.c (open_base_files): Likewise.
22812 * gimple-pretty-print.c: Likewise.
22813 * gimple-pretty-print.h: Likewise.
22814 * graph.c (print_graph_cfg): Likewise.
22815 * graphite-scop-detection.c (dot_all_sese): Likewise.
22816 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22817 * loop-unroll.c (report_unroll): Likewise.
22818 * passes.c (pass_manager::register_one_dump_file): Likewise.
22819 * print-tree.c: Likewise.
22820 * statistics.c: Likewise.
22821 * tree-cfg.c: Likewise.
22822 * tree-cfg.h: Likewise.
22823 * tree-dfa.c: Likewise.
22824 * tree-dfa.h: Likewise.
22825 * tree-dump.c (dump_function): Likewise.
22826 * tree-dump.h (struct dump_info): Likewise.
22827 * tree-pretty-print.c: Likewise.
22828 * tree-pretty-print.h: Likewise.
22829 * tree-ssa-live.c: Likewise.
22830 * tree-ssa-live.h: Likewise.
22831 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22832 * tree-vect-loop.c: Likewise.
22833 * tree-vect-slp.c: Likewise.
22835 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22836 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22838 PR tree-optimization/80457
22839 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22840 of all arguments to a statement as scalar_to_vec operations.
22841 (vectorizable_call): Adjust call to vect_model_simple_cost for
22843 (vectorizable_conversion): Likewise.
22844 (vectorizable_assignment): Likewise.
22845 (vectorizable_shift): Likewise.
22846 (vectorizable_operation): Likewise.
22847 (vectorizable_comparison): Likewise.
22848 (vect_is_simple_cond): Record the def types for operands.
22849 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22850 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22851 for statement argument count.
22853 2017-05-16 Carl Love <cel@us.ibm.com>
22855 * config/rs6000/rs6000-c: Add support for built-in functions
22856 vector unsigned long long vec_bperm (vector unsigned long long,
22857 vector unsigned char)
22858 vector signed long long vec_mule (vector signed int,
22860 vector unsigned long long vec_mule (vector unsigned int,
22861 vector unsigned int)
22862 vector signed long long vec_mulo (vector signed int,
22864 vector unsigned long long vec_mulo (vector unsigned int,
22865 vector unsigned int)
22866 vector signed char vec_sldw (vector signed char,
22867 vector signed char,
22869 vector unsigned char vec_sldw (vector unsigned char,
22870 vector unsigned char,
22872 vector signed short vec_sldw (vector signed short,
22873 vector signed short,
22875 vector unsigned short vec_sldw (vector unsigned short,
22876 vector unsigned short,
22878 vector signed int vec_sldw (vector signed int,
22881 vector unsigned int vec_sldw (vector unsigned int,
22882 vector unsigned int,
22884 vector signed long long vec_sldw (vector signed long long,
22885 vector signed long long,
22887 vector unsigned long long vec_sldw (vector unsigned long long,
22888 vector unsigned long long,
22890 * config/rs6000/rs6000-c: Add support for built-in functions
22891 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22892 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22893 * doc/extend.texi: Update the built-in documentation for the
22894 new built-in functions.
22896 2017-05-16 Marek Polacek <polacek@redhat.com>
22900 * tree.c (save_expr): Don't fold the expression.
22902 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22904 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22905 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22906 and (?*y,m). Update insn attributes.
22908 2017-05-16 Martin Liska <mliska@suse.cz>
22910 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
22911 flags argument of print_gimple_stmt, print_gimple_expr,
22912 print_generic_stmt and print_generic_expr.
22913 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
22914 * coretypes.h: Likewise.
22915 * except.c (dump_eh_tree): Likewise.
22916 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
22917 * gimple-pretty-print.h: Likewise.
22918 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
22919 (backprop::push_to_worklist): Likewise.
22920 (backprop::pop_from_worklist): Likewise.
22921 (backprop::process_use): Likewise.
22922 (backprop::intersect_uses): Likewise.
22923 (note_replacement): Likewise.
22924 * gimple-ssa-store-merging.c
22925 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
22926 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
22927 (pass_store_merging::execute): Likewise.
22928 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
22929 (ssa_base_cand_dump_callback): Likewise.
22930 (dump_incr_vec): Likewise.
22931 (replace_refs): Likewise.
22932 (replace_mult_candidate): Likewise.
22933 (create_add_on_incoming_edge): Likewise.
22934 (create_phi_basis): Likewise.
22935 (insert_initializers): Likewise.
22936 (all_phi_incrs_profitable): Likewise.
22937 (introduce_cast_before_cand): Likewise.
22938 (replace_one_candidate): Likewise.
22939 * gimplify.c (gimplify_expr): Likewise.
22940 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
22941 (set_rename): Likewise.
22942 (rename_uses): Likewise.
22943 (copy_loop_phi_nodes): Likewise.
22944 (add_close_phis_to_merge_points): Likewise.
22945 (copy_loop_close_phi_args): Likewise.
22946 (copy_cond_phi_args): Likewise.
22947 (graphite_copy_stmts_from_block): Likewise.
22948 (translate_pending_phi_nodes): Likewise.
22949 * graphite-poly.c (print_pdr): Likewise.
22950 (dump_gbb_cases): Likewise.
22951 (dump_gbb_conditions): Likewise.
22952 (print_scop_params): Likewise.
22953 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
22954 (build_cross_bb_scalars_use): Likewise.
22955 (gather_bbs::before_dom_children): Likewise.
22956 * hsa-dump.c (dump_hsa_immed): Likewise.
22957 * ipa-cp.c (print_ipcp_constant_value): Likewise.
22958 (get_replacement_map): Likewise.
22959 * ipa-inline-analysis.c (dump_condition): Likewise.
22960 (estimate_function_body_sizes): Likewise.
22961 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
22962 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
22963 * ipa-prop.c (ipa_dump_param): Likewise.
22964 (ipa_print_node_jump_functions_for_edge): Likewise.
22965 (ipa_modify_call_arguments): Likewise.
22966 (ipa_modify_expr): Likewise.
22967 (ipa_dump_param_adjustments): Likewise.
22968 (ipa_dump_agg_replacement_values): Likewise.
22969 (ipcp_modif_dom_walker::before_dom_children): Likewise.
22970 * ipa-pure-const.c (check_stmt): Likewise.
22971 (pass_nothrow::execute): Likewise.
22972 * ipa-split.c (execute_split_functions): Likewise.
22973 * omp-offload.c (dump_oacc_loop_part): Likewise.
22974 (dump_oacc_loop): Likewise.
22975 * trans-mem.c (tm_log_emit): Likewise.
22976 (tm_memopt_accumulate_memops): Likewise.
22977 (dump_tm_memopt_set): Likewise.
22978 (dump_tm_memopt_transform): Likewise.
22979 * tree-cfg.c (gimple_verify_flow_info): Likewise.
22980 (print_loop): Likewise.
22981 * tree-chkp-opt.c (chkp_print_addr): Likewise.
22982 (chkp_gather_checks_info): Likewise.
22983 (chkp_get_check_result): Likewise.
22984 (chkp_remove_check_if_pass): Likewise.
22985 (chkp_use_outer_bounds_if_possible): Likewise.
22986 (chkp_reduce_bounds_lifetime): Likewise.
22987 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
22988 (chkp_mark_completed_bounds): Likewise.
22989 (chkp_register_incomplete_bounds): Likewise.
22990 (chkp_mark_invalid_bounds): Likewise.
22991 (chkp_maybe_copy_and_register_bounds): Likewise.
22992 (chkp_build_returned_bound): Likewise.
22993 (chkp_get_bound_for_parm): Likewise.
22994 (chkp_build_bndldx): Likewise.
22995 (chkp_get_bounds_by_definition): Likewise.
22996 (chkp_generate_extern_var_bounds): Likewise.
22997 (chkp_get_bounds_for_decl_addr): Likewise.
22998 * tree-chrec.c (chrec_apply): Likewise.
22999 * tree-data-ref.c (dump_data_reference): Likewise.
23000 (dump_subscript): Likewise.
23001 (dump_data_dependence_relation): Likewise.
23002 (analyze_overlapping_iterations): Likewise.
23003 * tree-inline.c (expand_call_inline): Likewise.
23004 (tree_function_versioning): Likewise.
23005 * tree-into-ssa.c (dump_defs_stack): Likewise.
23006 (dump_currdefs): Likewise.
23007 (dump_names_replaced_by): Likewise.
23008 (dump_update_ssa): Likewise.
23009 (update_ssa): Likewise.
23010 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23011 * tree-parloops.c (build_new_reduction): Likewise.
23012 (try_create_reduction_list): Likewise.
23013 (ref_conflicts_with_region): Likewise.
23014 (oacc_entry_exit_ok_1): Likewise.
23015 (oacc_entry_exit_single_gang): Likewise.
23016 * tree-pretty-print.h: Likewise.
23017 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23018 (get_scalar_evolution): Likewise.
23019 (add_to_evolution): Likewise.
23020 (get_loop_exit_condition): Likewise.
23021 (analyze_evolution_in_loop): Likewise.
23022 (analyze_initial_condition): Likewise.
23023 (analyze_scalar_evolution): Likewise.
23024 (instantiate_scev): Likewise.
23025 (number_of_latch_executions): Likewise.
23026 (gather_chrec_stats): Likewise.
23027 (final_value_replacement_loop): Likewise.
23028 (scev_const_prop): Likewise.
23029 * tree-sra.c (dump_access): Likewise.
23030 (disqualify_candidate): Likewise.
23031 (create_access): Likewise.
23032 (reject): Likewise.
23033 (maybe_add_sra_candidate): Likewise.
23034 (create_access_replacement): Likewise.
23035 (analyze_access_subtree): Likewise.
23036 (analyze_all_variable_accesses): Likewise.
23037 (sra_modify_assign): Likewise.
23038 (initialize_constant_pool_replacements): Likewise.
23039 (find_param_candidates): Likewise.
23040 (decide_one_param_reduction): Likewise.
23041 (replace_removed_params_ssa_names): Likewise.
23042 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23043 * tree-ssa-copy.c (dump_copy_of): Likewise.
23044 (copy_prop_visit_cond_stmt): Likewise.
23045 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23046 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23047 (record_equivalences_from_stmt): Likewise.
23048 * tree-ssa-dse.c (compute_trims): Likewise.
23049 (delete_dead_call): Likewise.
23050 (delete_dead_assignment): Likewise.
23051 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23052 (forward_propagate_into_cond): Likewise.
23053 (pass_forwprop::execute): Likewise.
23054 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23055 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23057 (move_computations_worker): Likewise.
23058 (execute_sm): Likewise.
23059 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23060 (remove_exits_and_undefined_stmts): Likewise.
23061 (remove_redundant_iv_tests): Likewise.
23062 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23063 (adjust_iv_update_pos): Likewise.
23064 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23065 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23066 (value_replacement): Likewise.
23067 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23068 * tree-ssa-pre.c (print_pre_expr): Likewise.
23069 (get_representative_for): Likewise.
23070 (create_expression_by_pieces): Likewise.
23071 (insert_into_preds_of_block): Likewise.
23072 (eliminate_insert): Likewise.
23073 (eliminate_dom_walker::before_dom_children): Likewise.
23074 (eliminate): Likewise.
23075 (remove_dead_inserted_code): Likewise.
23076 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23077 * tree-ssa-reassoc.c (get_rank): Likewise.
23078 (eliminate_duplicate_pair): Likewise.
23079 (eliminate_plus_minus_pair): Likewise.
23080 (eliminate_not_pairs): Likewise.
23081 (undistribute_ops_list): Likewise.
23082 (eliminate_redundant_comparison): Likewise.
23083 (update_range_test): Likewise.
23084 (optimize_range_tests_var_bound): Likewise.
23085 (optimize_vec_cond_expr): Likewise.
23086 (rewrite_expr_tree): Likewise.
23087 (rewrite_expr_tree_parallel): Likewise.
23088 (linearize_expr): Likewise.
23089 (break_up_subtract): Likewise.
23090 (linearize_expr_tree): Likewise.
23091 (attempt_builtin_powi): Likewise.
23092 (attempt_builtin_copysign): Likewise.
23093 (transform_stmt_to_copy): Likewise.
23094 (transform_stmt_to_multiply): Likewise.
23095 (dump_ops_vector): Likewise.
23096 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23097 (print_scc): Likewise.
23098 (set_ssa_val_to): Likewise.
23099 (visit_reference_op_store): Likewise.
23100 (visit_use): Likewise.
23101 (sccvn_dom_walker::before_dom_children): Likewise.
23102 (run_scc_vn): Likewise.
23103 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
23105 (expr_hash_elt::print): Likewise.
23106 (const_and_copies::pop_to_marker): Likewise.
23107 (const_and_copies::record_const_or_copy_raw): Likewise.
23108 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
23109 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
23110 (dump_predicates): Likewise.
23111 (find_uninit_use): Likewise.
23112 (warn_uninitialized_phi): Likewise.
23113 (pass_late_warn_uninitialized::execute): Likewise.
23114 * tree-ssa.c (verify_vssa): Likewise.
23115 (verify_ssa): Likewise.
23116 (maybe_optimize_var): Likewise.
23117 * tree-vrp.c (dump_value_range): Likewise.
23118 (dump_all_value_ranges): Likewise.
23119 (dump_asserts_for): Likewise.
23120 (register_edge_assert_for_2): Likewise.
23121 (vrp_visit_cond_stmt): Likewise.
23122 (vrp_visit_switch_stmt): Likewise.
23123 (vrp_visit_stmt): Likewise.
23124 (vrp_visit_phi_node): Likewise.
23125 (simplify_cond_using_ranges_1): Likewise.
23126 (fold_predicate_in): Likewise.
23127 (evrp_dom_walker::before_dom_children): Likewise.
23128 (evrp_dom_walker::push_value_range): Likewise.
23129 (evrp_dom_walker::pop_value_range): Likewise.
23130 (execute_early_vrp): Likewise.
23132 2017-05-16 Richard Biener <rguenther@suse.de>
23134 * dwarf2out.c (loc_list_from_tree_1): Do not create
23135 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
23137 2017-05-16 Richard Biener <rguenther@suse.de>
23139 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
23141 (note_variable_value_in_expr): If we resolved the decl ref
23142 do not push to the stack.
23144 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
23146 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
23147 operations in fast-math mode.
23148 (vaddq_f16): Likewise.
23149 (vmul_f16): Likewise.
23150 (vmulq_f16): Likewise.
23151 (vsub_f16): Likewise.
23152 (vsubq_f16): Likewise.
23153 * config/arm/neon.md (add<mode>3): New.
23155 (fma:<VH:mode>3): New. Also remove outdated comment.
23158 2017-05-16 Martin Liska <mliska@suse.cz>
23162 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23163 (odr_types_equivalent_p): Likewise.
23165 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23167 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23169 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23172 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23173 non-interunit SSE move alternatives with '?'.
23174 (zero-extendsidi peephole2): New peephole to skip intermediate
23175 general register in SSE zero-extend sequence.
23177 2017-05-15 Jeff Law <law@redhat.com>
23179 * reorg.c (relax_delay_slots): Create a new variable to hold
23180 the temporary target rather than clobbering TARGET_LABEL.
23182 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23183 missing argument to extract_bit_field call.
23184 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23186 2017-05-15 Martin Liska <mliska@suse.cz>
23189 * gcc.c (process_command): Do not allow empty argument of -o option.
23191 2017-05-15 Renlin Li <renlin.li@arm.com>
23193 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23194 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23195 * config/aarch64/constraints.md (Usf): Add long call check.
23196 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23197 (call_value): Likewise.
23198 (sibcall): Likewise.
23199 (sibcall_value): Likewise.
23201 (call_value_insn): New.
23202 (sibcall_insn): Update rtx pattern.
23203 (sibcall_value_insn): Likewise.
23204 (call_internal): Remove.
23205 (call_value_internal): Likewise.
23206 (sibcall_internal): Likewise.
23207 (sibcall_value_internal): Likewise.
23208 (call_reg): Likewise.
23209 (call_symbol): Likewise.
23210 (call_value_reg): Likewise.
23211 (call_value_symbol): Likewise.
23213 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23216 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23218 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23220 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23221 compatible with CCGOCmode and with CCZmode.
23223 2017-05-14 Martin Sebor <msebor@redhat.com>
23225 PR middle-end/77671
23226 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23227 (gimple_fold_builtin_snprintf): Same.
23228 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23229 (gimple_fold_builtin_snprintf): Same.
23230 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23231 of character types.
23232 (is_call_safe): New function.
23233 (try_substitute_return_value): Call it.
23234 (try_simplify_call): New function.
23235 (pass_sprintf_length::handle_gimple_call): Call it.
23237 2017-05-14 Martin Sebor <msebor@redhat.com>
23239 PR middle-end/80669
23240 * builtins.c (expand_builtin_stpncpy): Simplify.
23242 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23244 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23245 * config/i386/i386.h
23246 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23247 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23248 (struct machine_function): Add new members call_ms2sysv,
23249 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23250 (struct machine_frame_state): New fields sp_realigned and
23251 sp_realigned_offset.
23252 * config/i386/i386.c
23253 (enum xlogue_stub): New enum.
23254 (enum xlogue_stub_sets): New enum.
23255 (class xlogue_layout): New class.
23256 (struct ix86_frame): New fields stack_realign_allocate_offset,
23257 stack_realign_offset and outlined_save_offset. Modify comments to
23258 detail stack layout when using out-of-line stubs.
23259 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23260 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23261 -mcall-ms2sysv-xlogues.
23262 (stub_managed_regs): New static variable.
23263 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23264 registers managed by out-of-line stub.
23265 (disable_call_ms2sysv_xlogues): New function.
23266 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23267 m->call_ms2sysv when appropriate and compute frame layout for
23269 (sp_valid_at, fp_valid_at): New inline functions.
23270 (choose_basereg): New function.
23271 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23273 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23274 Use align parameter of choose_baseaddr to generated aligned SSE movs
23276 (pro_epilogue_adjust_stack): Modify to track
23277 machine_frame_state::sp_realigned.
23278 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23279 (ix86_nsaved_sseregs): Likewise.
23280 (ix86_emit_save_regs): Likewise.
23281 (ix86_emit_save_regs_using_mov): Likewise.
23282 (ix86_emit_save_sse_regs_using_mov): Likewise.
23283 (get_scratch_register_on_entry): Likewise.
23284 (gen_frame_set): New function.
23285 (gen_frame_load): Likewise.
23286 (gen_frame_store): Likewise.
23287 (emit_outlined_ms2sysv_save): Likewise.
23288 (emit_outlined_ms2sysv_restore): Likewise.
23289 (ix86_expand_prologue): Modify stack re-alignment code and call
23290 emit_outlined_ms2sysv_save when appropriate.
23291 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23292 parameter rtx_insn *insn, which allows the function to be used to only
23293 generate the notes.
23294 (ix86_expand_epilogue): Modify validity checks of frame and stack
23295 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23296 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23297 * config/i386/predicates.md
23298 (save_multiple): New predicate.
23299 (restore_multiple): Likewise.
23300 * config/i386/sse.md
23301 (save_multiple<mode>): New pattern.
23302 (save_multiple_realign<mode>): Likewise.
23303 (restore_multiple<mode>): Likewise.
23304 (restore_multiple_and_return<mode>): Likewise.
23305 (restore_multiple_leave_return<mode>): Likewise.
23306 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23308 2017-05-14 Julia Koval <julia.koval@intel.com>
23310 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23311 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23312 (__builtin_ia32_xsetbv): New builtins.
23313 * config/i386/i386.c (ix86_expand_special_args_builtin):
23315 (ix86_expand_builtin): Special expand for new intrinsics.
23316 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23317 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23318 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23320 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23322 * cfganal.c (inverted_post_order_compute): Change argument type
23324 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23325 * df-core.c (rest_of_handle_df_initialize): Adjust.
23326 (rest_of_handle_df_finish): Likewise.
23327 (df_analyze_1): Likewise.
23328 (df_analyze): Likewise.
23329 (loop_inverted_post_order_compute): Change argument to be a vec *.
23330 (df_analyze_loop): Adjust.
23331 (df_get_n_blocks): Likewise.
23332 (df_get_postorder): Likewise.
23333 * df.h (struct df_d): Change field to be a vec.
23334 * lcm.c (compute_laterin): Adjust.
23335 (compute_available): Likewise.
23336 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23337 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23338 * tree-ssa-pre.c (compute_antic): Likewise.
23340 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23342 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23343 (depth_first_search::depth_first_search): Change structure init
23344 function to this constructor.
23345 (depth_first_search::add_bb): Rename function to this member.
23346 (depth_first_search::execute): Likewise.
23347 (flow_dfs_compute_reverse_finish): Adjust.
23349 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23351 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23352 (longest_simple_path): Likewise.
23353 * shrink-wrap.c (spread_components): Likewise.
23354 (disqualify_problematic_components): Likewise.
23355 (emit_common_heads_for_components): Likewise.
23356 (emit_common_tails_for_components): Likewise.
23357 (insert_prologue_epilogue_for_components): Likewise.
23359 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23361 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23364 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23366 * df-core.c (df_set_blocks): Start using auto_bitmap.
23367 (df_compact_blocks): Likewise.
23368 * df-problems.c (df_rd_confluence_n): Likewise.
23369 * df-scan.c (df_insn_rescan_all): Likewise.
23370 (df_process_deferred_rescans): Likewise.
23371 (df_update_entry_block_defs): Likewise.
23372 (df_update_exit_block_uses): Likewise.
23373 (df_entry_block_bitmap_verify): Likewise.
23374 (df_exit_block_bitmap_verify): Likewise.
23375 (df_scan_verify): Likewise.
23376 * lra-constraints.c (lra_constraints): Likewise.
23377 (undo_optional_reloads): Likewise.
23378 (lra_undo_inheritance): Likewise.
23379 * lra-remat.c (calculate_gen_cands): Likewise.
23380 (do_remat): Likewise.
23381 * lra-spills.c (assign_spill_hard_regs): Likewise.
23382 (spill_pseudos): Likewise.
23383 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23384 (bitmap_set_subtract_values): Likewise.
23386 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23388 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23389 management with auto_bitmap.
23390 (fix_inter_tick): Likewise.
23391 (fix_recovery_deps): Likewise.
23392 * ira.c (add_store_equivs): Likewise.
23393 (find_moveable_pseudos): Likewise.
23394 (split_live_ranges_for_shrink_wrap): Likewise.
23395 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23396 (rtx_reuse_manager::seen_def_p): Likewise.
23397 (rtx_reuse_manager::set_seen_def): Likewise.
23398 * print-rtl.h (class rtx_reuse_manager): Likewise.
23400 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23402 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23404 (migrate_btr_def): Likewise.
23405 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23406 * df-core.c (loop_post_order_compute): Likewise.
23407 (loop_inverted_post_order_compute): Likewise.
23408 * hsa-common.h: Likewise.
23409 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23410 * init-regs.c (initialize_uninitialized_regs): Likewise.
23411 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23412 (inline_small_functions): Likewise.
23413 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23414 * ira.c (combine_and_move_insns): Likewise.
23415 (build_insn_chain): Likewise.
23416 * loop-invariant.c (find_invariants): Likewise.
23417 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23418 * predict.c (tree_predict_by_opcode): Likewise.
23419 (predict_paths_leading_to): Likewise.
23420 (predict_paths_leading_to_edge): Likewise.
23421 (estimate_loops_at_level): Likewise.
23422 (estimate_loops): Likewise.
23423 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23424 (spread_components): Likewise.
23425 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23426 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23427 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23428 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23429 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23430 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23431 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23432 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23433 (mark_threaded_blocks): Likewise.
23434 (thread_through_all_blocks): Likewise.
23435 * tree-ssa.c (verify_ssa): Likewise.
23436 (execute_update_addresses_taken): Likewise.
23437 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23439 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23441 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23443 (post_order_compute): Likewise.
23444 (inverted_post_order_compute): Likewise.
23445 (pre_and_rev_post_order_compute_fn): Likewise.
23447 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23449 * genrecog.c (int_set::int_set): Explicitly construct our
23450 auto_vec base class.
23451 * vec.h (auto_vec::auto_vec): New constructor.
23453 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23455 * bitmap.h (class auto_bitmap): New constructor taking
23456 bitmap_obstack * argument.
23458 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23460 * bitmap.h (class auto_bitmap): Change type of m_bits to
23461 bitmap_head, and adjust ctor / dtor and member operators.
23463 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23465 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23466 when returned register mode doesn't match original mode.
23468 2017-05-12 Jeff Law <law@redhat.com>
23469 Jakub Jelinek <jakub@redhat.com>
23471 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23472 we look for cc setter after the compare-elim changes.
23473 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23474 within the vector to match what compare-elim now expects.
23475 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23476 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23478 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23479 after the compare-elim changes.
23480 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23481 the vector to match what compare-elim now expects.
23482 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23483 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23484 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23485 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23486 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23488 * config/visium/visium.c (single_set_and_flags): Fix where
23489 we look for cc setter after the compare-elim changes.
23490 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23491 with the vector to match what compare-elim now expects.
23492 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23493 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23494 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23495 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23496 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23497 (neg<mode>2_insn_set_overflow): Likewise.
23499 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23501 PR middle-end/79794
23502 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23503 maybe_expand_insn call, set ops[0].target. If still set after call,
23504 set alt_rtl. Add extra arg to recursive calls.
23505 (extract_bit_field): Add alt_rtl argument. Pass to
23507 * expmed.h (extract_bit_field): Fix prototype.
23508 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23509 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23510 to extract_bit_field_calls.
23511 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23512 Pass alt_rtl to extract_bit_field calls.
23513 * calls.c (store_unaligned_arguments_into_psuedos)
23514 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23515 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23517 * optabs.h (struct expand_operand): Add target bitfield.
23519 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23521 * compare-elim.c (try_eliminate_compare): Canonicalize
23522 operation with embedded compare to
23523 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23524 (set (reg) (operation)].
23526 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23528 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23531 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23532 cost of adding a carry flag for ADC instruction.
23533 [case MINUS]: Ignore the cost of subtracting a carry flag
23534 for SBB instruction.
23536 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23538 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23540 * config/rs6000/bmiintrin.h: New file.
23541 * config/rs6000/bmi2intrin.h: New file.
23542 * config/rs6000/x86intrin.h: New file.
23544 2017-05-12 Jeff Law <law@redhat.com>
23546 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23549 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23551 PR middle-end/80707
23552 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23554 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23556 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23557 early expansion of vector divide builtins.
23558 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23559 builtins identified as having unsigned arguments.
23561 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23563 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23564 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23565 expansion of vector logical operations (and, andc, or, xor,
23568 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23570 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23571 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23573 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23575 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23576 early GIMPLE expansion of vector multiplies.
23578 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23580 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23581 TARGET_HAVE_MOVT conditional.
23582 (movt splitter): Likewise.
23584 2017-05-12 Richard Biener <rguenther@suse.de>
23586 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23587 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23588 Fold all stmts not inplace.
23590 2017-05-12 Richard Biener <rguenther@suse.de>
23592 PR tree-optimization/80713
23593 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23594 inserted_exprs bit for not removed stmts.
23596 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23598 PR middle-end/69921
23599 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
23600 parallelized" attribute for parallelized OpenACC kernels.
23601 * omp-offload.c (execute_oacc_device_lower): Use it.
23603 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
23604 Set "oacc kernels" attribute.
23605 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
23606 parameter. Adjust all users.
23607 (oacc_fn_attrib_kernels_p): Remove function.
23608 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
23609 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
23610 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
23611 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
23612 assert "oacc kernels" attribute is set.
23614 2017-05-11 Carl Love <cel@us.ibm.com>
23616 * config/rs6000/rs6000-c: Add support for built-in functions
23617 vector unsigned char vec_popcnt (vector signed char)
23618 vector unsigned char vec_popcnt (vector unsigned char)
23619 vector unsigned short vec_popcnt (vector signed short)
23620 vector unsigned short vec_popcnt (vector unsigned short)
23621 vector unsigned int vec_popcnt (vector signed int)
23622 vector unsigned int vec_popcnt (vector unsigned int)
23623 vector unsigned long long vec_popcnt (vector signed long long)
23624 vector unsigned long long vec_popcnt (vector unsigned long long)
23625 vector signed long long vec_slo (vector signed long long,
23626 vector signed char)
23627 vector signed long long vec_slo (vector signed long long,
23628 vector unsigned char)
23629 vector unsigned long long vec_slo (vector unsigned long long,
23630 vector signed char)
23631 vector unsigned long long vec_slo (vector unsigned long long,
23632 vector unsigned char)
23633 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23634 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23635 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23636 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23637 * doc/extend.texi: Update the built-in documentation file for the
23638 new built-in functions.
23640 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23642 * attribs.h (sorted_attr_string): Move machine independent
23643 functions for target clone support from the i386 port to common
23644 code. Rename ix86_function_versions to common_function_versions.
23645 Rename make_name to make_unique_name.
23646 (common_function_versions): Likewise.
23647 (make_unique_name): Likewise.
23648 (make_dispatcher_decl): Likewise.
23649 (is_function_default_version): Likewise.
23650 * attribs.c (attr_strcmp): Likewise.
23651 (sorted_attr_string): Likewise.
23652 (common_function_versions): Likewise.
23653 (make_unique_name): Likewise.
23654 (make_dispatcher_decl): Likewise.
23655 (is_function_default_version): Likewise.
23656 * config/i386/i386.c (attr_strcmp): Likewise.
23657 (sorted_attr_string): Likewise.
23658 (ix86_function_versions): Likewise.
23659 (make_name): Likewise.
23660 (make_dispatcher_decl): Likewise.
23661 (is_function_default_version): Likewise.
23662 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23664 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23667 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23668 Account for direct move costs for vec_construct of integer
23671 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23674 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23675 (UNSPEC_STX_ATOMIC): Ditto.
23676 (loaddi_via_sse): New insn.
23677 (storedi_via_sse): Ditto.
23678 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23679 Update corresponding peephole2 patterns.
23680 (atomic_storedi_fpu): Ditto.
23682 2017-05-11 Julia Koval <julia.koval@intel.com>
23684 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23685 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23687 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23688 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23689 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23691 2017-05-11 Nathan Sidwell <nathan@acm.org>
23693 * graphite-poly.c: Include dumpfile.h.
23695 * dumpfle.h (dump_function): Declare here ...
23696 * tree-dump.h (dump_function): ... not here.
23697 * dumpfile.c: #include tree-cfg.h.
23698 (dump_function): Move here from ...
23699 * tree-dump.c (dump_function): ... here.
23700 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23701 * graphite-poly.c: Don't include tree-dump.h.
23702 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23703 * print-tree.c: Likewise.
23704 * stor-layout.c: Likewise.
23705 * tree-nested.c: Likewise.
23707 * dumpfile.c (dump_start): Use TDF_FLAGS.
23708 (dump_enable_all): Fix TDF_KIND check thinko.
23710 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23712 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23713 array entries to represent two legal parameterizations of the
23714 overloaded __builtin_cmpb function, as represented by the
23715 P6_OV_BUILTIN_CMPB constant.
23716 (altivec_resolve_overloaded_builtin): Add special case handling
23717 for the __builtin_cmpb function, as represented by the
23718 P6_OV_BUILTIN_CMPB constant.
23719 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23720 (BU_P6_64BIT_2): New macro.
23721 (BU_P6_OVERLOAD_2): New macro
23722 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23723 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23724 (CMPB): Add overload support to represent both 32-bit and 64-bit
23725 compare-bytes function.
23726 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23727 support for TARGET_CMPB.
23728 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23729 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23730 documentation of the __builtin_cmpb overloaded built-in function.
23732 2017-05-11 Richard Biener <rguenther@suse.de>
23734 PR tree-optimization/80705
23735 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23736 bases are not vectorizable.
23738 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23740 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23741 when counting register pressure.
23743 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23745 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23746 (iv_ca_more_deps): Renamed to ...
23747 (iv_ca_compare_deps): ... this.
23748 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23750 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23752 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23754 (determine_group_iv_costs): ... here.
23755 (find_inv_vars_cb): Record inv var if it's not recorded before.
23757 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23759 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23760 (get_shiftadd_cost): Ditto.
23762 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23764 * tree-ssa-address.c: Include header file.
23765 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23767 (add_to_parts): Refactor.
23768 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23769 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23772 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23774 PR tree-optimization/53090
23775 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23777 (extract_cond_operands): Detect condition with IV on both sides
23778 and return COMP_IV_EXPR_2.
23779 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23780 (rewrite_use_compare): Simplify by removing call to function
23781 extract_cond_operands.
23783 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23785 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23786 (extract_cond_operands): Detect condition comparing against non-
23787 invariant bound and return appropriate enum value.
23788 (find_interesting_uses_cond): Update use of extract_cond_operands.
23789 Handle its return value accordingly.
23790 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23792 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23794 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23795 nonlinear iv_use computation in loop invariant sensitive way.
23797 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23799 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23800 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23802 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23804 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23805 (dump_cand): Support iv_cand.inv_exprs.
23806 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23808 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23811 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23813 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23815 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23816 as local function. Include necessary header files.
23817 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23819 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23821 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23823 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23825 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23826 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23827 RSHIFT_EXPR and BIT_NOT_EXPR.
23829 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23831 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23832 (adjust_setup_cost): New parameter supporting round up adjustment.
23833 (struct address_cost_data): Delete.
23834 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23835 (split_address_cost, ptr_difference_cost): Delete.
23836 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23837 (struct ainc_cost_data): New struct.
23838 (get_address_cost_ainc): New function.
23839 (get_address_cost, get_computation_cost): Reimplement.
23840 (determine_group_iv_cost_address): Record inv_expr for all uses of
23842 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23843 (iv_ca_has_deps): Reimplemented to ...
23844 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23846 (iv_ca_extend): Call iv_ca_more_deps.
23848 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23850 * tree-ssa-address.c (struct mem_address): Move to header file.
23851 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23852 * tree-ssa-address.h (struct mem_address): Move from C file.
23853 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23855 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23857 * tree-affine.h (aff_combination_type): New interface.
23858 (aff_combination_zero_p): Remove static.
23859 (aff_combination_const_p): New interface.
23860 (aff_combination_singleton_var_p): New interfaces.
23862 2017-05-11 Richard Biener <rguenther@suse.de>
23864 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23865 Skip unreachable blocks and destinations.
23866 (eliminate): Move stmt removal and fixup ...
23867 (fini_eliminate): ... here. Skip inserted exprs.
23868 (pass_pre::execute): Move fini_pre after fini_eliminate.
23869 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23870 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23871 PRE to get rid of dead code that has invalid SSA form and
23872 split critical edges again.
23874 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23876 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23878 2017-05-11 Richard Biener <rguenther@suse.de>
23880 * passes.c (execute_function_todo): Verify loops if they are
23881 said to be up-to-date.
23882 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23883 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23885 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23888 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23889 handle calling assemble_external ourself.
23892 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23893 modes with zero size. Enhance comment.
23895 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23897 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23898 built-ins for vec_xl and vec_xst with short and char pointer
23901 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23903 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23904 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23905 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23906 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23907 (_mm_maskz_min_round_ss): New intrinsics.
23908 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23909 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23910 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
23911 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
23912 (__builtin_ia32_minss_mask_round): New builtins.
23913 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23914 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23915 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
23917 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
23918 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
23920 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
23923 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23925 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
23926 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
23927 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
23928 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
23929 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
23930 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
23931 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
23932 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23933 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23934 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
23935 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
23936 (__builtin_ia32_mulss_mask_round): New builtins.
23937 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23938 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23939 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
23941 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
23942 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23944 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23947 2017-05-10 Julia Koval <julia.koval@intel.com>
23949 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
23950 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
23951 (_mm256_setr_m128i): New intrinsics.
23953 2017-05-10 Julia Koval <julia.koval@intel.com>
23955 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
23956 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
23957 (_mm_maskz_rcp14_ss): New intrinsics.
23958 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
23959 (__builtin_ia32_rcp14ss_mask): New builtins.
23960 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
23962 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
23964 PR tree-optimization/51513
23965 * tree-cfg.c (gimple_seq_unreachable_p): New function.
23966 (assert_unreachable_fallthru_edge_p): Use it.
23967 (group_case_labels_stmt): Likewise.
23968 * tree-cfg.h: Prototype it.
23969 * stmt.c: Include cfghooks.h and tree-cfg.h.
23970 (emit_case_dispatch_table) <gap_label>: New local variable.
23971 Use it to fill dispatch table gaps.
23972 Test for default_label before updating probabilities.
23973 (expand_case) <default_label>: Remove unneeded initialization.
23974 Test for unreachable default case statement and remove its edge.
23975 Set default_label accordingly.
23976 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
23978 2017-05-10 Carl Love <cel@us.ibm.com>
23980 * config/rs6000/rs6000-c: Add support for built-in functions
23981 vector signed char vec_neg (vector signed char)
23982 vector signed short int vec_neg (vector short int)
23983 vector signed int vec_neg (vector signed int)
23984 vector signed long long vec_neg (vector signed long long)
23985 vector float vec_neg (vector float)
23986 vector double vec_neg (vector double)
23987 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
23989 * config/rs6000/altivec.h: Add define for vec_neg
23990 * doc/extend.texi: Update the built-in documentation for the
23991 new built-in functions.
23993 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23995 PR tree-optimization/77644
23996 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
23998 2017-05-10 Nathan Sidwell <nathan@acm.org>
24000 * dumpfile.h (TDI_lang_all): New.
24001 (TDF_KIND): New. Renumber others
24002 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24004 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24006 (get_dump_file_name): Adjust suffix generation.
24007 (dump_enable_all): Use TDF_KIND.
24008 * doc/invoke.texi (-fdump-lang-all): Document.
24010 * dumpfile.h: Tabify.
24012 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24015 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24016 Move member access before delete.
24018 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24020 * tree-inline.c (expand_call_inline): Split block at stmt
24023 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24026 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24027 are now unused after splitting mov{sf,sd}_hardfloat.
24028 (f32_lr2): Likewise.
24029 (f32_lm): Likewise.
24030 (f32_lm2): Likewise.
24031 (f32_li): Likewise.
24032 (f32_li2): Likewise.
24033 (f32_lv): Likewise.
24034 (f32_sr): Likewise.
24035 (f32_sr2): Likewise.
24036 (f32_sm): Likewise.
24037 (f32_sm2): Likewise.
24038 (f32_si): Likewise.
24039 (f32_si2): Likewise.
24040 (f32_sv): Likewise.
24041 (f32_dm): Likewise.
24042 (f32_vsx): Likewise.
24043 (f32_av): Likewise.
24044 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24045 For movsf, order stores so the VSX stores occur before the GPR
24046 store which encourages the register allocator to use a traditional
24047 FPR instead of a GPR. For movsd, order the stores so that the GPR
24048 store comes before the VSX stores to allow the power6 to work.
24049 This is due to the power6 not having a 32-bit integer store
24050 instruction from a FPR.
24051 (movsf_hardfloat): Likewise.
24052 (movsd_hardfloat): Likewise.
24054 2017-05-09 Martin Sebor <msebor@redhat.com>
24056 PR translation/80280
24057 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24060 PR translation/80280
24061 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24062 data member added in r247778.
24063 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24065 2017-05-09 Nathan Sidwell <nathan@acm.org>
24067 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24069 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24072 2017-05-09 Marek Polacek <polacek@redhat.com>
24074 * doc/invoke.texi: Fix typo.
24076 2017-05-09 Richard Biener <rguenther@suse.de>
24078 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24079 (vrp_val_is_min): Likewise.
24080 (set_value_range_to_value): Likewise.
24081 (set_value_range_to_nonnegative): Likewise.
24082 (gimple_assign_nonzero_p): Likewise.
24083 (gimple_stmt_nonzero_p): Likewise.
24084 (vrp_int_const_binop): Likewise. Remove unreachable case.
24085 (adjust_range_with_scev): Adjust comments.
24086 (compare_range_with_value): Likewise.
24087 (extract_range_from_phi_node): Likewise.
24088 (test_for_singularity): Likewise.
24090 2017-05-09 Richard Biener <rguenther@suse.de>
24092 * tree-vrp.c (get_single_symbol): Add assert that we don't
24093 get overflowed constants as invariant part.
24094 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24095 checks. Use wi::cmp instead of recursing for integer constants.
24096 (compare_values): Just ignore whether we assumed undefined
24097 overflow instead of failing the compare.
24098 (extract_range_for_var_from_comparison_expr): Add comment before the
24099 TREE_NO_WARNING sets.
24100 (test_for_singularity): Likewise.
24101 (extract_range_from_comparison): Do not disable optimization
24102 when we assumed undefined overflow.
24103 (extract_range_basic): Remove init of unused var.
24105 2017-05-09 Richard Biener <rguenther@suse.de>
24107 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
24108 (extract_range_from_multiplicative_op_1): Adjust.
24109 (extract_range_from_binary_expr_1): Use int_const_binop.
24111 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
24114 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
24115 rs6000_store_data_bypass_p in seven define_bypass directives and
24116 in several comments.
24117 * config/rs6000/rs6000-protos.h: Add prototype for
24118 rs6000_store_data_bypass_p function.
24119 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
24120 function implements slightly different (rs6000-specific) semantics
24121 than store_data_bypass_p, returning false rather than aborting
24122 with assertion error when arguments do not satisfy the
24123 requirements of store data bypass.
24124 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
24125 rs6000_store_data_bypass_p.
24127 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
24129 * config/xtensa/xtensa-protos.h
24130 (xtensa_initial_elimination_offset): New declaration.
24131 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24132 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
24133 macro definition, add case for FRAME_POINTER_REGNUM when
24134 FRAME_GROWS_DOWNWARD.
24135 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
24136 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
24137 xtensa_initial_elimination_offset.
24139 2017-05-08 Nathan Sidwell <nathan@acm.org>
24141 * doc/invoke.texi: Alphabetize -fdump options.
24143 2017-05-08 Martin Sebor <msebor@redhat.com>
24145 PR translation/80280
24146 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
24148 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
24150 * target.def (compute_frame_layout): New optional target hook.
24151 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
24152 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
24153 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
24155 * reload1.c (verify_initial_elim_offsets): Likewise.
24156 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
24157 (use_simple_return_p): Call arm_compute_frame_layout if needed.
24158 (arm_get_frame_offsets): Split up into this ...
24159 (arm_compute_frame_layout): ... and this function.
24161 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24163 * config/aarch64/constraints.md (Usa): New constraint.
24164 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24166 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24168 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24169 with_multilib_list after it has been checked.
24171 2017-05-08 Richard Biener <rguenther@suse.de>
24173 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24174 (bitmap_set_subtract_values): Likewise.
24176 2017-05-08 Richard Biener <rguenther@suse.de>
24178 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24179 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24181 (gimple_stmt_nonzero_warnv_p): Rename to ...
24182 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24184 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24185 (extract_range_basic): Adjust, do not disable propagation on
24186 strict overflow sensitive simplification.
24187 (vrp_visit_cond_stmt): Likewise.
24189 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24191 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24192 body size unconditionally.
24194 2017-05-07 Jeff Law <law@redhat.com>
24197 2017-05-06 Jeff Law <law@redhat.com>
24198 PR tree-optimization/78496
24199 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24202 PR tree-optimization/78496
24203 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24204 (simplify_stmt_using_ranges): Call it.
24205 (vrp_dom_walker::before_dom_children): Extract equivalences
24206 from an ASSERT_EXPR with an equality comparison against a
24209 2017-05-06 Jeff Law <law@redhat.com>
24211 PR tree-optimization/78496
24212 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24215 PR tree-optimization/78496
24216 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24217 (simplify_stmt_using_ranges): Call it.
24218 (vrp_dom_walker::before_dom_children): Extract equivalences
24219 from an ASSERT_EXPR with an equality comparison against a
24222 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24224 * lra-constraints.c (lra_copy_reg_equiv): New function.
24225 (split_reg): Use it to copy equivalence information from the
24226 original register to the spill register.
24228 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24230 PR rtl-optimization/75964
24231 * simplify-rtx.c (simplify_const_relational_operation): Remove
24232 invalid handling of comparisons of integer ABS.
24234 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24236 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24237 initialize to zero.
24238 (init_regs): Remove declaration.
24239 (function_arg_advance_32): Initialize error_p as boolean variable.
24241 2017-05-05 Nathan Sidwell <nathan@acm.org>
24243 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24244 lines. Use for (;;).
24246 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24248 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24249 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24250 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24251 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24252 VF=2 that require versioning.
24254 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24256 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24259 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24261 * diagnostic.h (diagnostic_override_option_index): Convert from
24262 macro to inline function.
24264 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24266 * diagnostic.c (last_module_changed_p): New function.
24267 (set_last_module): New function.
24268 (diagnostic_report_current_module): Convert macro usage to
24269 the above functions.
24270 * diagnostic.h (diagnostic_context::last_module): Strengthen
24271 from const line_map * to const line_map_ordinary *.
24272 (diagnostic_last_module_changed): Delete macro.
24273 (diagnostic_set_last_module): Delete macro.
24275 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24277 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24278 with diagnostic_report_diagnostic.
24279 (diagnostic_n_impl_richloc): Likewise.
24280 * diagnostic.h (report_diagnostic): Delete macro.
24281 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24282 with diagnostic_report_diagnostic.
24283 * substring-locations.c (format_warning_va): Likewise.
24285 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24287 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24288 save/restor of format_spec. Move option-printing code to...
24289 (print_option_information): ...this new function, and
24290 reimplement by simply printing to the pretty_printer,
24291 rather than appending to the format string.
24293 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24295 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24296 handling logic into...
24297 (update_effective_level_from_pragmas): ...this new function.
24299 2017-05-04 Andrew Waterman <andrew@sifive.com>
24301 * config/riscv/riscv.opt (mstrict-align): New option.
24302 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24303 (SLOW_UNALIGNED_ACCESS): Define.
24304 (riscv_slow_unaligned_access): Declare.
24305 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24307 (riscv_slow_unaligned_access): New variable.
24308 (rocket_tune_info): Set slow_unaligned_access to true.
24309 (optimize_size_tune_info): Set slow_unaligned_access to false.
24310 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24311 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24312 (riscv_option_override): Set riscv_slow_unaligned_access.
24313 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24315 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24317 * config/riscv/riscv.md: Unify indentation.
24319 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24324 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24326 (extendsi<mode>2): Add support for doing sign extension via
24327 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24328 don't have ISA 3.0 instructions.
24329 (extendsi<mode>2 splitter): Likewise.
24330 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24331 generate the normal insns since SImode can now go in vector
24332 registers. Disallow the special UNSPECs needed for previous
24333 machines to hide SImode being used. Add new insns
24334 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24335 (fix_trunc<mode>si2_stfiwx): Likewise.
24336 (fix_trunc<mode>si2_internal): Likewise.
24337 (fixuns_trunc<mode>si2): Likewise.
24338 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24339 (fctiw<u>z_<mode>_smallint): Likewise.
24340 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24341 of floating point to 32-bit integer from doing a direct move to
24342 the GPR registers to do a store.
24343 (fctiwz_<mode>): Break long line.
24345 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24347 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24348 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24349 (addr_list, addr_offset_valid_p): New.
24350 (split_address_groups): Check offset validity with above function.
24351 (gt-tree-ssa-loop-ivopts.h): Include header file.
24353 2017-05-05 Nathan Sidwell <nathan@acm.org>
24355 * config.gcc (arm*-*-*): Add missing 'fi'.
24357 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24359 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24360 included in -fopt-info does not matter.
24361 * doc/optinfo.texi (-fopt-info): Fix description of default
24362 behavour. Explicitly say order of options included in -fopt-info
24365 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24367 * config.gcc: Allow combinations of aprofile and rmprofile values for
24368 --with-multilib-list.
24369 * config/arm/t-multilib: New file.
24370 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24371 variables. Remove setting of ISA and floating-point ABI in
24372 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24373 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24374 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24375 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24377 * config/arm/t-rmprofile: Likewise except for the matches changes.
24378 * doc/install.texi (--with-multilib-list): Document the combination of
24379 aprofile and rmprofile values and warn about pitfalls in doing that.
24381 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24383 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24384 (movdi_aarch64): Likewise.
24386 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24388 PR tree-optimization/80632
24389 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24391 (build_arrays): Initialize it for virtual phis.
24392 (fix_phi_nodes): Use it for virtual phis.
24394 PR tree-optimization/80558
24395 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24396 [x, y] op z into [x op, y op z] for op & or | if conditions
24399 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24400 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24403 * config/arm/arm.md (use_literal_pool): Remove.
24404 (64-bit immediate split): No longer takes cost into consideration
24405 if arm_disable_literal_pool is enabled.
24406 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24407 used when arm_disable_literal_pool is enabled.
24408 (arm_max_const_double_inline_cost): Remove use of
24409 arm_disable_literal_pool.
24410 (push_minipool_fix): Add assert.
24411 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24412 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24413 (no_literal_pool_sf_immediate): New.
24415 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24417 PR tree-optimization/80613
24418 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24419 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24421 2017-05-05 Richard Biener <rguenther@suse.de>
24423 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24425 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24427 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24428 of this flag from insn conditions due to removal from r247495.
24430 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24432 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24434 (arm_early_store_addr_dep_ptr): Likewise.
24435 * config/arm/aarch-common-protos.h
24436 (arm_early_load_addr_dep_ptr): Add prototype.
24437 (arm_early_store_addr_dep_ptr): Likewise.
24438 * config/arm/cortex-a53.md: Add new bypasses.
24440 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24442 * tree.c (next_type_uid): Change type to unsigned.
24443 (type_hash_canon): Decrement back next_type_uid if
24444 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24445 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24448 2017-05-04 Martin Sebor <msebor@redhat.com>
24450 * builtins.c: Fix a trivial typo in a comment.
24452 PR middle-end/79234
24453 * builtins.c (check_sizes): Adjust to handle reading past the end.
24454 Avoid printing excessive upper bound of ranges. Use %E to print
24455 tree nodes instead of converting them to %wu.
24456 (expand_builtin_memchr): New function.
24457 (compute_dest_size): Rename...
24458 (compute_objsize): ...to this.
24459 (expand_builtin_memcpy): Adjust.
24460 (expand_builtin_mempcpy): Adjust.
24461 (expand_builtin_strcat): Adjust.
24462 (expand_builtin_strcpy): Adjust.
24463 (check_strncat_sizes): Adjust.
24464 (expand_builtin_strncat): Adjust.
24465 (expand_builtin_strncpy): Adjust and simplify.
24466 (expand_builtin_memset): Adjust.
24467 (expand_builtin_bzero): Adjust.
24468 (expand_builtin_memcmp): Adjust.
24469 (expand_builtin): Handle memcmp.
24470 (maybe_emit_chk_warning): Check strncat just once.
24472 2017-05-04 Martin Sebor <msebor@redhat.com>
24474 PR preprocessor/79214
24475 PR middle-end/79222
24476 PR middle-end/79223
24477 * builtins.c (check_sizes): Add inlining context and issue
24478 warnings even when -Wno-system-headers is set.
24479 (check_strncat_sizes): Same.
24480 (expand_builtin_strncat): Same.
24481 (expand_builtin_memmove): New function.
24482 (expand_builtin_stpncpy): Same.
24483 (expand_builtin): Handle memmove and stpncpy.
24485 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24487 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24488 which is not used any more.
24490 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24492 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24494 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24496 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24497 (cortexa53_tunings): Likewise.
24498 (cortexa57_tunings): Likewise.
24499 (cortexa72_tunings): Likewise.
24500 (cortexa73_tunings): Likewise.
24502 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24504 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24505 Set loop alignment to 8.
24507 2017-05-04 Martin Sebor <msebor@redhat.com>
24509 PR translation/80280
24510 * builtins.c (expand_builtin_object_size): Add missing quoting to
24511 %D and like directives.
24512 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24513 (hsa_type_for_tree_type): Same.
24514 (verify_function_arguments): Same.
24515 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24516 * varasm.c (get_section): Same.
24519 2017-05-04 Martin Sebor <msebor@redhat.com>
24521 PR translation/80280
24522 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24524 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24526 * config/aarch64/aarch64.c (generic_addrcost_table):
24527 Change HI/TI mode setting.
24529 2017-05-04 Martin Jambor <mjambor@suse.cz>
24531 PR tree-optimization/80622
24532 * tree-sra.c (comes_initialized_p): New function.
24533 (build_accesses_from_assign): Only set write lazily when
24534 comes_initialized_p is false.
24535 (analyze_access_subtree): Use comes_initialized_p.
24536 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24537 instead of testing for PARM_DECL.
24539 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24541 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24542 constraint on operand 0 to allow more general addressing modes.
24543 Adjust output template.
24544 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24546 * config/aarch64/aarch64-protos.h
24547 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24548 * config/aarch64/constraints.md (Dp): New address constraint.
24549 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24552 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24554 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24555 update use of estimate_ipcp_clone_size_and_time.
24556 (estimate_local_effects): Update use of
24557 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24558 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24559 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24560 Return nonspecialized time.
24562 2017-05-04 Richard Biener <rguenther@suse.de>
24564 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24565 for the last VUSE which def dominates the PHI. Directly call
24567 (get_continuation_for_phi_1): Remove.
24569 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24571 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24572 to explain the use of truncating division. Cap the number of
24573 iterations to the maximum given by nb_iterations_upper_bound,
24576 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24578 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24579 * configure: Regenerate.
24580 * config.in: Regenerate.
24581 * config/i386/driver-mingw32.c: new file.
24582 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24583 * config.host: Link driver-mingw32.o on MinGW host.
24584 * doc/install.texi: Document new --enable-mingw-wildcard configure
24587 2017-05-04 Marek Polacek <polacek@redhat.com>
24589 PR tree-optimization/80612
24590 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24592 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24593 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24595 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24596 (movt splitter): Likewise.
24597 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24598 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
24599 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
24600 block for Thumb-1 with MOVT.
24601 (thumb2_legitimate_address_p): Move code block ...
24602 (can_avoid_literal_pool_for_label_p): ... into this new function.
24603 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
24605 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
24606 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
24607 "M-profile targets with the MOVT instruction".
24609 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24611 * config/arm/arm-builtins.c (arm_init_builtins): Rename
24612 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
24613 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24615 2017-05-04 Martin Liska <mliska@suse.cz>
24617 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24618 variable cond_code.
24620 2017-05-04 Richard Biener <rguenther@suse.de>
24622 * tree.c (array_at_struct_end_p): Handle arrays at struct
24623 end with flexarrays more conservatively. Refactor and treat
24624 arrays of arrays or aggregates more strict. Fix
24625 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24626 * tree.c (array_at_struct_end_p): Adjust prototype.
24627 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24628 * gimple-fold.c (get_range_strlen): Likewise.
24629 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24631 2017-05-04 Richard Biener <rguenther@suse.de>
24633 PR tree-optimization/31130
24634 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24636 (supports_overflow_infinity): Likewise.
24637 (is_negative_overflow_infinity): Likewise.
24638 (is_positive_overflow_infinity): Likewise.
24639 (is_overflow_infinity): Likewise.
24640 (stmt_overflow_infinity): Likewise.
24641 (overflow_infinity_range_p): Likewise.
24642 (usable_range_p): Remove as always returning true.
24643 (make_overflow_infinity): Remove.
24644 (negative_overflow_infinity): Likewise.
24645 (positive_overflow_infinity): Likewise.
24646 (avoid_overflow_infinity): Likewise.
24647 (set_value_range): Adjust accordingly.
24648 (set_value_range_to_nonnegative): Likewise, remove now unused
24649 overflow_infinity arg.
24650 (vrp_operand_equal_p): Adjust.
24651 (update_value_range): Likewise.
24652 (range_int_cst_singleton_p): Likewise.
24653 (operand_less_p): Likewise.
24654 (compare_values_warnv): Likewise.
24655 (extract_range_for_var_from_comparison_expr): Likewise.
24656 (vrp_int_const_binop): Likewise.
24657 (zero_nonzero_bits_from_vr): Likewise.
24658 (extract_range_from_multiplicative_op_1): Likewise.
24659 (extract_range_from_binary_expr_1): Likewise.
24660 (extract_range_from_unary_expr): Likewise.
24661 (extract_range_from_comparison): Likewise.
24662 (extract_range_basic): Likewise.
24663 (adjust_range_with_scev): Likewise.
24664 (compare_ranges): Likewise.
24665 (compare_range_with_value): Likewise.
24666 (dump_value_range): Likewise.
24667 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24669 (simplify_cond_using_ranges): Adjust.
24671 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24673 * brig-builtins.def: Added a builtin for class_f64.
24674 * builtin-types.def: Added a builtin type needed by class_f64.
24676 2017-05-03 Jason Merrill <jason@redhat.com>
24678 * timevar.def: Add TV_CONSTEXPR.
24680 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24682 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24684 2017-05-03 Martin Jambor <mjambor@suse.cz>
24686 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24687 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24688 * ipa-cp.c (ipcp_propagate_stage): Do not call
24689 ipa_update_after_lto_read.
24690 * ipa-inline.c (ipa_inline): Likewise.
24692 2017-05-03 Martin Jambor <mjambor@suse.cz>
24694 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24695 tag. Added a default constructor and a destructor.
24696 (ipa_edge_args_sum_t): New class;
24697 (ipa_edge_args_sum): Declare.
24698 (ipa_edge_args_vector): Remove declaration.
24699 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24700 (ipa_free_edge_args_substructures): Remove declaration.
24701 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24702 (ipa_edge_args_info_available_for_edge_p): Likewise.
24703 * ipa-prop.c (ipa_edge_args_vector): Removed.
24704 (edge_removal_hook_holder): Likewise.
24705 (edge_duplication_hook_holder): Likewise.
24706 (ipa_edge_args_sum): New variable.
24707 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24708 ipa_edge_args_vector.
24709 (ipa_free_edge_args_substructures): Likewise.
24710 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24711 ipa_edge_args_vector.
24712 (ipa_edge_removal_hook): Turned into method
24713 ipa_edge_args_sum_t::remove.
24714 (ipa_edge_duplication_hook): Turned into method
24715 ipa_edge_args_sum_t::duplicate.
24716 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24717 registering edge hooks.
24718 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24719 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24720 ipa_edge_args_sum instead of ipa_edge_args_vector.
24721 * ipa-profile.c (ipa_profile): Likewise.
24723 2017-05-03 Martin Jambor <mjambor@suse.cz>
24725 * symbol-summary.h (function_summary): New method exists.
24726 (function_summary::symtab_removal): Deallocate through release.
24727 (call_summary): New class.
24728 (gt_ggc_mx): New overload.
24729 (gt_pch_nx): Likewise.
24730 (gt_pch_nx): Likewise.
24732 2017-05-03 Jeff Law <law@redhat.com>
24734 PR tree-optimization/78496
24735 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24736 from simplify_cond_using_ranges. Split off code to walk
24737 backwards through casts into ...
24738 (simplify_cond_using_ranges_2): New function.
24739 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24740 (execute_vrp): After identifying jump threads, call
24741 simplify_cond_using_ranges_2.
24743 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24746 * ipa-inline.h (inline_summary): Add ctor.
24747 (create_ggc): Do not use ggc_cleared_alloc.
24749 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24750 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24752 * gcc.c (handle_braces): Support escaping in switch matching
24754 * doc/invoke.texi (Spec Files): Document it.
24755 Remove superfluous @code markup in items.
24757 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24759 * diagnostic-show-locus.c (struct column_range): New struct.
24760 (get_affected_columns): New function.
24761 (get_printed_columns): New function.
24762 (struct correction): New struct.
24763 (correction::ensure_capacity): New function.
24764 (correction::ensure_terminated): New function.
24765 (struct line_corrections): New struct.
24766 (line_corrections::~line_corrections): New dtor.
24767 (line_corrections::add_hint): New function.
24768 (layout::print_trailing_fixits): Reimplement in terms of the new
24770 (selftest::test_overlapped_fixit_printing): New function.
24771 (selftest::diagnostic_show_locus_c_tests): Call it.
24773 2017-05-03 Nathan Sidwell <nathan@acm.org>
24775 Canonicalize canonical type hashing
24776 * tree.h (type_hash_canon_hash): Declare.
24777 * tree.c (type_hash_list, attribute_hash_list): Move into
24778 type_hash_canon_hash.
24779 (build_type_attribute_qual_variant): Break out hash code calc into
24780 type_hash_canon_hash.
24781 (type_hash_canon_hash): New. Generic type hash computation.
24782 (build_range_type_1, build_array_type_1, build_function_type,
24783 build_method_type_directly, build_offset_type, build_complex_type,
24784 make_vector_type): Call it.
24786 2017-05-03 Richard Biener <rguenther@suse.de>
24788 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24789 When all DRs have unknown misaligned do not always peel
24790 when there is a store but apply the same costing model as if
24791 there were only loads.
24793 2017-05-03 Richard Biener <rguenther@suse.de>
24796 PR tree-optimization/80492
24797 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24798 compare_base_decls returning dont-know properly.
24800 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24802 * config/arm/iterators.md (CCSI): New mode iterator.
24803 (arch): New mode attribute.
24804 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24805 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24806 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24807 code iterator for success result mode.
24808 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24809 the corresponding new insn generators.
24811 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24814 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24815 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24817 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24819 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24820 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24821 (DDR_A): Wrap DDR argument in brackets.
24822 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24823 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24824 (DDR_REVERSED_P): Likewise.
24826 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24828 PR tree-optimization/79472
24829 * tree-switch-conversion.c (struct switch_conv_info): Add
24830 contiguous_range and default_case_nonstandard fields.
24831 (collect_switch_conv_info): Compute contiguous_range and
24832 default_case_nonstandard fields, don't clear final_bb if
24833 contiguous_range and only the default case doesn't have the required
24835 (check_all_empty_except_final): Set default_case_nonstandard instead
24836 of failing if contiguous_range and the default case doesn't have empty
24838 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24839 and only the default case doesn't have the required constants. Skip
24841 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24842 if default_case_nonstandard.
24843 (build_constructors): Build constant 1 just once. Assert that default
24844 values aren't inserted in between cases if contiguous_range. Skip
24846 (build_arrays): Skip virtual phis.
24847 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24848 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24849 Handle virtual phis.
24850 (gen_inbound_check): Handle default_case_nonstandard case.
24851 (process_switch): Adjust check_final_bb caller. Call
24852 gather_default_values with the first non-default case instead of
24853 default case if default_case_nonstandard.
24855 2017-05-02 Nathan Sidwell <nathan@acm.org>
24857 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24858 check. Fix formatting.
24860 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24862 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24863 errors when comparing specialized and unspecialized times.
24865 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24867 * diagnostic-show-locus.c
24868 (layout::should_print_annotation_line_p): Make private.
24869 (layout::print_annotation_line): Make private.
24870 (layout::annotation_line_showed_range_p): Make private.
24871 (layout::show_ruler): Make private.
24872 (layout::print_source_line): Make private. Pass in line and
24873 line_width, rather than calling location_get_source_line. Drop
24875 (layout::print_leading_fixits): New method.
24876 (layout::print_any_fixits): Rename to...
24877 (layout::print_trailing_fixits): ...this, and make private.
24878 Don't print newline fixits.
24879 (diagnostic_show_locus): Move logic for printing one row into...
24880 (layout::print_line): ...this new function. Move the
24881 location_get_source_line call and error-handling from
24882 print_source_line to here. Call print_leading_fixits, and rename
24883 print_any_fixits to print_trailing_fixits.
24884 (selftest::test_fixit_insert_containing_newline): Update now that
24885 newlines are partially supported.
24886 (selftest::test_fixit_insert_containing_newline_2): New test.
24887 (selftest::test_fixit_replace_containing_newline): Update comments.
24888 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24889 * edit-context.c (class added_line): New class.
24890 (class edited_line): Describe newline handling in comment.
24891 (edited_line::actually_edited_p): New method.
24892 (edited_line::print_content): Delete redundant decl.
24893 (edited_line::m_predecessors): New field.
24894 (edited_file::print_content): Call edited_line::print_content.
24895 (edited_file::print_diff): Update to support newlines.
24896 (edited_file::print_diff_hunk): Likewise.
24897 (edited_file::print_run_of_changed_lines): New function.
24898 (edited_file::print_diff_line): Convert to...
24899 (print_diff_line): ...this.
24900 (edited_file::get_effective_line_count): New function.
24901 (edited_line::edited_line): Initialize new field m_predecessors.
24902 (edited_line::~edited_line): Clean up m_predecessors.
24903 (edited_line::apply_fixit): Handle newlines.
24904 (edited_line::get_effective_line_count): New function.
24905 (edited_line::print_content): New function.
24906 (edited_line::print_diff_lines): New function.
24907 (selftest::test_applying_fixits_insert_containing_newline): New
24909 (selftest::test_applying_fixits_replace_containing_newline): New
24911 (selftest::insert_line): New function.
24912 (selftest::test_applying_fixits_multiple_lines): Add example of
24914 (selftest::edit_context_c_tests): Call the new tests.
24916 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24918 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
24919 parameter cand. Update dump information.
24920 (get_computation_cost): Update uses.
24922 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24924 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
24925 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
24926 (get_computation_at, rewrite_use_address): Update use of
24927 get_computation_aff.
24929 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24931 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
24932 (get_computation): Delete.
24933 (get_computation_cost): Implement like get_computation_cost_at.
24934 Use get_computation_at.
24935 (get_computation_cost_at): Delete.
24936 (rewrite_use_nonlinear_expr): Use get_computation_at.
24937 (rewrite_use_compare, remove_unused_ivs): Ditto.
24939 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24941 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
24943 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24945 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
24946 (ivopts_global_cost_for_size): Rename parameter and update uses.
24947 (iv_ca_recount_cost): Update uses.
24948 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
24949 candidates seperately in n_invs and n_cands.
24950 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
24952 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24954 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
24955 (find_inv_vars_cb): New.
24956 (find_depends): Renamed to ...
24957 (find_inv_vars): ... this.
24958 (add_candidate_1, force_var_cost): Call find_inv_vars.
24959 (split_address_cost, determine_group_iv_cost_cond): Ditto.
24961 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24963 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
24964 inv_vars. Add inv_exprs.
24965 (struct iv_cand): Rename depends_on to inv_vars.
24966 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
24967 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
24968 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
24969 (dump_cand): Dump inv_vars.
24970 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
24971 (record_invariant, find_depends, add_candidate_1): Ditto.
24972 (set_group_iv_cost, force_var_cost): Ditto.
24973 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
24974 (get_computation_cost_at, get_computation_cost): Ditto.
24975 (determine_group_iv_cost_generic): Ditto.
24976 (determine_group_iv_cost_address): Ditto.
24977 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
24978 (determine_group_iv_costs): Ditto.
24979 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
24980 (iv_ca_set_remove_invariants): Renamed to ...
24981 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
24982 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
24983 (iv_ca_set_add_invariants): Renamed to ...
24984 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
24985 (iv_ca_set_cp): Use iv_ca_set_add_invs.
24986 (iv_ca_has_deps): Support inv_vars and inv_exprs.
24987 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
24988 (create_new_ivs): Remove useless dump.
24990 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24992 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
24994 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
24995 (iv_ca_set_no_cp, create_new_iv): Ditto.
24997 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24999 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25001 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25003 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25004 function tree_check2.
25006 2017-05-02 Martin Liska <mliska@suse.cz>
25008 * doc/gcov.texi: Add missing preposition.
25009 * gcov.c (function_info::function_info): Properly fill up
25010 all member variables.
25012 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25014 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25016 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25018 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25020 2017-05-02 Martin Liska <mliska@suse.cz>
25023 * lto-streamer-in.c (lto_read_tree_1): Remove
25024 LTO_STREAMER_DEBUG.
25025 * lto-streamer.c (struct tree_hash_entry): Likewise.
25026 (struct tree_entry_hasher): Likewise.
25027 (tree_entry_hasher::hash): Likewise.
25028 (tree_entry_hasher::equal): Likewise.
25029 (lto_streamer_init): Likewise.
25030 (lto_orig_address_map): Likewise.
25031 (lto_orig_address_get): Likewise.
25032 (lto_orig_address_remove): Likewise.
25033 * lto-streamer.h: Likewise.
25034 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25035 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25037 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25039 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25040 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25041 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25042 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25043 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25044 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25045 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25046 (mm_maskz_sub_ss): New intrinsics.
25047 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25048 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25049 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25050 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25051 (__builtin_ia32_subss_mask_round): New builtins.
25052 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25053 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25054 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25056 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25057 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25059 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25062 2017-05-02 Martin Jambor <mjambor@suse.cz>
25064 PR tree-optimization/78687
25065 * tree-sra.c (access): New field parent.
25066 (process_subtree_disqualification): New function.
25067 (disqualify_candidate): Call it.
25068 (build_accesses_from_assign): Reset write flag if creating an
25070 (build_access_subtree): Fill in parent field and also prpagate
25071 down grp_write flag.
25072 (create_artificial_child_access): New parameter set_grp_write, set
25073 grp_write to its value.
25074 (propagate_subaccesses_across_link): Also propagate grp_write flag
25076 (propagate_all_subaccesses): Push the closest parent back to work
25077 queue if add_access_to_work_queue returned true.
25079 2017-05-02 Richard Biener <rguenther@suse.de>
25081 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25082 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25083 -fstrict-overflow documentation.
25084 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25085 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25086 flag_strict_overflow.
25087 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25088 * lto-opts.c (lto_write_options): Do not stream it.
25089 * lto-wrapper.c (merge_and_complain): Do not handle it.
25090 * opts.c (default_options_table): Do not set -fstrict-overflow.
25091 (finish_options): Likewise do not clear it when sanitizing.
25092 * simplify-rtx.c (simplify_const_relational_operation): Do not
25093 test flag_strict_overflow.
25095 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25097 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25098 using enabled attribute.
25099 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
25100 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
25101 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
25102 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
25103 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
25104 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
25105 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
25106 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
25107 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
25108 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
25110 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25112 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
25114 2017-05-02 Richard Biener <rguenther@suse.de>
25116 PR tree-optimization/80591
25118 2017-04-10 Richard Biener <rguenther@suse.de>
25120 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25123 2017-05-02 Richard Biener <rguenther@suse.de>
25125 PR tree-optimization/80549
25126 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
25127 (cleanup_tree_cfg_noloop): Create forwarders to known loop
25128 headers if they do not have a preheader.
25130 2017-05-02 Martin Liska <mliska@suse.cz>
25133 * common.opt: Fix typo.
25134 * doc/invoke.texi: Likewise.
25136 2017-05-01 Jan Beulich <jbeulich@suse.com>
25138 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
25139 swapping, add (x,x,m,x,n) alternative.
25141 2017-05-01 Nathan Sidwell <nathan@acm.org>
25143 * calls.c (combine_pending_stack_adjustment_and_call): Remove
25144 unnecessary unadjusted_alignment check.
25146 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25149 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
25151 * gimplify.c (gimplify_cilk_detach): New function.
25152 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
25153 * tree-core.h: Document EXPR_CILK_SPAWN.
25154 * tree.h (EXPR_CILK_SPAWN): Define.
25156 2017-05-01 David Malcolm <dmalcolm@redhat.com>
25158 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
25159 to use new fixit_hint representation, using the "replace" logic.
25160 (get_line_span_for_fixit_hint): Likewise.
25161 (layout::print_any_fixits): Likewise.
25162 (selftest::test_one_liner_many_fixits): Rename to...
25163 (selftest::test_one_liner_many_fixits_1): ...this, and update
25164 comment and expected output to reflect that the multiple fix-it
25165 hints are now consolidated into one insertion.
25166 (selftest::test_one_liner_many_fixits_2): New test.
25167 (selftest::test_diagnostic_show_locus_one_liner): Update for
25169 (selftest::test_fixit_consolidation): Update for fix-it API
25171 * diagnostic.c (print_parseable_fixits): Likewise.
25172 * edit-context.c (edited_line::m_line_events): Convert from
25173 auto_vec <line_event *> to auto_vec <line_event>.
25174 (class line_event): Convert from abstract base class to a concrete
25175 class, taking over the role of replace_event.
25176 (class insert_event): Delete.
25177 (class replace_event): Rename to class line_event. Convert to
25179 (edit_context::add_fixits): Reimplement.
25180 (edit_context::apply_insert): Delete.
25181 (edit_context::apply_replace): Rename to...
25182 (edit_context::apply_fixit): ...this. Convert to half-open range.
25183 (edited_file::apply_insert): Delete.
25184 (edited_file::apply_replace): Rename to...
25185 (edited_file::apply_fixit): ...this.
25186 (edited_line::~edited_line): Drop deletion of events.
25187 (edited_line::apply_insert): Delete.
25188 (edited_line::apply_replace): Rename to...
25189 (edited_line::apply_fixit): ...this. Convert to half-open range.
25190 Update for change to type of m_line_events.
25191 * edit-context.h (edit_context::apply_insert): Delete.
25192 (edit_context::apply_replace): Rename to...
25193 (edit_context::apply_fixit): ...this.
25195 2017-05-01 Martin Sebor <msebor@redhat.com>
25197 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25200 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25203 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25204 __get_cpuid_max returns 0.
25205 (__get_cpuid_count): Ditto.
25207 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25209 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25210 replacement expression is another instance of one of its arguments.
25212 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25215 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25216 check for stack push/pop autoinc.
25217 * config/i386/i386.c (ix86_agi_dependent): Return false
25218 if the only reason why modified_in_p returned true is that
25219 addr is SP based and set_insn is a push or pop.
25221 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25223 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25226 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25229 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25230 (account_size_time): Use two predicates - exec_pred and
25232 (evaluate_conditions_for_known_args): Compute both clause and
25234 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25235 (inline_summary_t::duplicate): Update.
25236 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25238 (compute_inline_parameters): Likewise.
25239 (estimate_edge_size_and_time): Update caluclation of time.
25240 (estimate_node_size_and_time): Compute both time and nonspecialized
25242 (estimate_ipcp_clone_size_and_time): Update.
25243 (inline_merge_summary): Update.
25244 (do_estimate_edge_time): Update.
25245 (do_estimate_edge_size): Update.
25246 (do_estimate_edge_hints): Update.
25247 (inline_read_section, inline_write_summary): Stream both new predicates.
25248 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25250 (compute_inlined_call_time): Cleanup.
25251 (big_speedup_p): Update.
25252 (edge_badness): Update.
25253 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25254 (size_time_entry): Replace predicate by exec_predicate and
25255 nonconst_predicate.
25256 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25257 (estimate_edge_time): Return also nonspec_time.
25258 (reset_edge_growth_cache): Update.
25260 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25262 PR rtl-optimization/80491
25263 * ifcvt.c (noce_process_if_block): When looking for x setter
25264 with missing else_bb, don't check only the insn right before
25265 cond_earliest, but look for the last insn that x is modified in
25266 within the same bb.
25268 PR rtl-optimization/80491
25269 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25271 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25273 PR tree-optimization/80487
25274 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25276 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25278 PR tree-optimization/79697
25279 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25280 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25281 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25283 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25284 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25286 2017-04-28 Martin Sebor <msebor@redhat.com>
25288 PR tree-optimization/80523
25289 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25290 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25292 (maybe_warn, format_directive, parse_directive): Use new functions.
25293 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25295 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25297 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25299 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25301 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25302 target_header_dir): Set correctly.
25303 * configure: Regenerated.
25304 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25305 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25306 instead of SYSTEM_HEADER_DIR.
25308 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25310 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25311 (estimate_local_effects): Likewise.
25312 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25313 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25314 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25315 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25316 do_estimate_edge_time, estimate_edge_time): Likewise.
25317 * ipa-inline-analysis.c (estimate_node_size_and_time,
25318 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25319 (estimate_time_after_inlining): Remove.
25321 2017-04-28 Martin Liska <mliska@suse.cz>
25323 * doc/gcov.texi: Enhance documentation of gcov.
25325 2017-04-28 Martin Liska <mliska@suse.cz>
25327 * doc/gcov.texi: Sort options in alphabetic order.
25328 * doc/gcov-dump.texi: Likewise.
25329 * doc/gcov-tool.texi: Likewise.
25330 * gcov.c (print_usage): Likewise.
25331 * gcov-dump.c (print_usage): Likewise.
25332 * gcov-tool.c (print_merge_usage_message): Likewise.
25333 (print_rewrite_usage_message): Likewise.
25334 (print_overlap_usage_message): Likewise.
25336 2017-04-28 Martin Liska <mliska@suse.cz>
25338 PR gcov-profile/53915
25339 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25341 2017-04-28 Martin Liska <mliska@suse.cz>
25343 PR gcov-profile/79891
25344 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25345 is marked by compiler as living on a line.
25346 (get_cycles_count): Remove usage of the union.
25347 (output_intermediate_file): Likewise.
25348 (find_source): Fix GNU coding style.
25349 (accumulate_line_counts): Remove old non-all block mode.
25350 (output_lines): Remove usage of the union.
25351 * profile.c (output_location): Include all BBs, even if
25352 belonging to a same line (and file) as a previous BB.
25354 2017-04-28 Martin Liska <mliska@suse.cz>
25356 * gcov.c (process_args): Handle new argument 'w'.
25357 (read_graph_file): Assign ID to BBs.
25358 (output_branch_count): Display BB # if verbose flag is set.
25359 (output_lines): Likewise for arcs.
25360 (print_usage): Add '--verbose' option help.
25361 * doc/gcov.texi: Document --verbose (-w) option.
25363 2017-04-28 Martin Liska <mliska@suse.cz>
25365 * gcov.c (struct block_location_info): New struct.
25366 (process_file): Fill up the new structure.
25367 (read_graph_file): Replace usage of encoding by the newly added
25369 (add_line_counts): Likewise.
25370 (accumulate_line_counts): Remove usage of the union.
25371 (function_info::function_info): New function.
25372 (function_info::~function_info): Likewise.
25373 (process_file): Call delete instead of release_function.
25374 (release_function): Release the function.
25375 (release_structures): Call delete instead of release_function.
25376 (solve_flow_graph): Replace usage of num_blocks.
25377 (find_exception_blocks): Likewise.
25378 (output_lines): Fix GNU coding style.
25380 2017-04-28 Martin Liska <mliska@suse.cz>
25383 * coverage.c (coverage_remove_note_file): New function.
25384 * coverage.h: Declare the function.
25385 * toplev.c (finalize): Clean if an error has been seen.
25387 2017-04-28 Martin Liska <mliska@suse.cz>
25389 PR gcov-profile/80031
25390 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25391 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25392 * gcov.c (read_graph_file): Read just number of blocks.
25393 * profile.c (branch_prob): Do not stream 0 flags per a basic
25396 2017-04-28 Martin Liska <mliska@suse.cz>
25398 * gcov-dump.c (tag_*): Add new argument to declarations.
25399 (dump_gcov_file): Likewise.
25400 (tag_blocks): Add and use new argument depth.
25401 (tag_arcs): Likewise.
25402 (tag_lines): Likewise.
25403 (tag_counters): Likewise.
25404 (tag_summary): Likewise.
25405 (dump_working_sets): Use depth to do a proper indentation.
25407 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25410 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25411 * symtab.c (symtab_node::debug_symtab): Move definition here.
25413 2017-04-28 Richard Biener <rguenther@suse.de>
25415 * lto-streamer.h (LTO_major_version): Bump to 7.
25417 2017-04-28 Richard Biener <rguenther@suse.de>
25419 * tree-vrp.c (assert_info): New struct.
25420 (add_assert_info): New helper.
25421 (register_edge_assert_for_2): Refactor to add asserts to a vector
25423 (register_edge_assert_for_1): Likewise.
25424 (register_edge_assert_for): Likewise.
25425 (finish_register_edge_assert_for): New helper actually registering
25426 asserts where live on edge.
25427 (find_conditional_asserts): Adjust.
25428 (find_switch_asserts): Likewise.
25429 (evrp_dom_walker::try_find_new_range): Generalize.
25430 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25432 2017-04-27 Marek Polacek <polacek@redhat.com>
25435 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25436 arg10 and arg11 to itype.
25438 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25440 * doc/extend.texi (Object Size Checking): Improve grammar.
25442 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25445 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25446 that the logic for permitting reciprocal estimates matches that
25449 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25452 * tree.c (type_cache_hasher::equal): Only compare
25453 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25454 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25455 non-aggregate element types.
25456 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25457 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25459 2017-04-27 Richard Biener <rguenther@suse.de>
25461 PR middle-end/80533
25462 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25463 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25464 keeping a reference to a trailing array.
25466 2017-04-27 Richard Biener <rguenther@suse.de>
25468 PR middle-end/80539
25469 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25470 being in loop-closed SSA form conservatively.
25471 (chrec_fold_multiply_poly_poly): Likewise.
25473 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25475 PR middle-end/79665
25476 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25477 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25479 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25482 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25483 (aarch64_function_arg_alignment): Return unsigned int again, but still
25484 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25485 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25486 Don't emit -Wpsabi note.
25487 (aarch64_function_arg_boundary): Likewise.
25488 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25491 2017-04-26 Nathan Sidwell <nathan@acm.org>
25493 * tree.h (crc32_unsigned_n): Declare.
25494 (crc32_unsigned, crc32_unsigned): Make inline.
25495 * tree.c (crc32_unsigned_bits): Replace with ...
25496 (crc32_unsigned_n): ... this.
25497 (crc32_unsigned, crc32_byte): Remove.
25498 (crc32_string): Remove unnecessary braces.
25500 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25502 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25503 * ipa-inline-analysis.c (MAX_TIME): Remove.
25504 (account_size_time): Use sreal for time.
25505 (dump_inline_summary): Update.
25506 (estimate_function_body_sizes): Update.
25507 (estimate_edge_size_and_time): Update.
25508 (estimate_calls_size_and_time): Update.
25509 (estimate_node_size_and_time): Update.
25510 (inline_merge_summary): Update.
25511 (inline_update_overall_summary): Update.
25512 (estimate_time_after_inlining): Update.
25513 (inline_read_section): Update.
25514 (inline_write_summary): Update.
25515 * ipa-inline.c (compute_uninlined_call_time): Update.
25516 (compute_inlined_call_time): Update.
25517 (recursive_inlining): Update.
25518 (inline_small_functions): Update.
25519 (dump_overall_stats): Update.
25520 * ipa-inline.h: Include sreal.h.
25521 (size_time_entry): Turn time to sreal.
25522 (inline_summary): Turn self_time nad time to sreal.
25524 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25526 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25528 (sreal::stream_out, sreal::stream_in): New.
25529 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25531 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25533 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25536 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25539 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25541 (dimode_scalar_chain::compute_convert_gain): Ditto.
25542 (dimode_scalar_chain::make_vector_copies): Ditto.
25543 (dimode_scalar_chain::convert_reg): Ditto.
25544 (dimode_scalar_chain::convert_insn): Ditto.
25545 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25546 (VI248_AVX512BW_1): New mode iterator.
25547 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25548 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25551 2017-04-25 Martin Sebor <msebor@redhat.com>
25553 PR tree-optimization/80497
25554 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25555 constants are representable in HOST_WIDE_INT.
25556 (parse_directive): Ditto.
25558 2017-04-25 Martin Sebor <msebor@redhat.com>
25561 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25562 (new_zero_array): Adjust signature.
25563 (dom_info::dom_init): Used unsigned rather that size_t.
25564 (dom_info::dom_info): Same.
25566 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25567 Jakub Jelinek <jakub@redhat.com>
25570 * config/arm/arm.c: Include gimple.h.
25571 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25572 returns negative, increment ncrn only if it returned positive.
25573 (arm_needs_doubleword_align): Return int instead of bool,
25574 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25575 members, but if there is any such non-FIELD_DECL
25576 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25577 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25578 returns negative, increment nregs only if it returned positive.
25579 (arm_setup_incoming_varargs): Likewise.
25580 (arm_function_arg_boundary): Emit -Wpsabi note if
25581 arm_needs_doubleword_align returns negative, return
25582 DOUBLEWORD_ALIGNMENT only if it returned positive.
25584 2017-04-25 Marek Polacek <polacek@redhat.com>
25587 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25588 first argument to type.
25590 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25593 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25594 type checks to test for compatibility instead of equality.
25596 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25597 Jakub Jelinek <jakub@redhat.com>
25600 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
25602 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
25603 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
25604 the alignment computation, but return their maximum in warn_alignment.
25605 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25606 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
25608 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
25609 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25612 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25614 * config/arc/simdext.md (dmpyh): Fix typo.
25616 2017-04-25 Richard Biener <rguenther@suse.de>
25618 PR tree-optimization/80492
25619 * alias.c (compare_base_decls): Handle registers with asm
25620 specification conservatively.
25621 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25622 compare_base_decls returning dont-know properly.
25624 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25626 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25627 (legitimate_offset_address_p): New function.
25628 (arc_legitimate_address_p): Use above function.
25630 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25632 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25634 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25636 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25637 ACCH registers whenever they are available.
25639 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25641 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25642 double regs fix when not used.
25644 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25646 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25648 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25649 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25651 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25653 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25654 when emitting short ADD instructions.
25656 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25658 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25660 (cmpsi_cc_c_insn): Likewise.
25661 (cbranchsi4_scratch): Compute proper instruction length using
25662 compact_hreg_operand.
25663 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25665 2017-04-25 Richard Biener <rguenther@suse.de>
25667 PR middle-end/80509
25668 * passes.c (pass_manager::pass_manager): Initialize
25669 m_name_to_pass_map.
25671 2017-04-25 Richard Biener <rguenther@suse.de>
25673 PR tree-optimization/79201
25674 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25676 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25679 * config/s390/vector.md: Split MEM->GPR vector moves for
25680 non-s_operand addresses.
25682 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25685 * config/s390/predicates.md (reload_const_wide_int_operand): New
25687 * config/s390/s390.md ("movti"): Remove d/P alternative.
25688 ("movti_bigconst"): New pattern definition.
25690 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25693 * s390-protos.h (s390_expand_cs_hqi): Removed.
25694 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25695 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25696 modes as well as CCZ1mode and CCZmode.
25697 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25698 signature of s390_emit_compare_and_swap.
25699 (s390_expand_cs_hqi): Likewise, make static.
25700 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25701 compare-and-swap, in some cases.
25702 (s390_expand_cs): Wrapper function.
25703 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25705 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25706 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25707 patterns for small and large integers. Forbid symref memory operands.
25708 Move expander to s390.c. Require cc register.
25709 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25710 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25711 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25712 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25713 symref memory operands. Remove CC mode and call s390_match_ccmode
25715 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25717 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25719 * config/s390/s390.md (define_peephole2): New peephole to help
25720 combining the load-and-test pattern with volatile memory.
25722 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25724 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25725 with CCZmode for TARGET_Z196.
25727 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25729 PR rtl-optimization/80501
25730 * combine.c (make_compound_operation_int): Set subreg_code to SET
25731 even for AND with mask of the sign bit of mode.
25733 PR rtl-optimization/80500
25734 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25735 sum's initial value.
25737 2017-04-25 Julian Brown <julian@codesourcery.com>
25738 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25740 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25742 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25744 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25746 2017-04-25 Julian Brown <julian@codesourcery.com>
25747 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25749 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25750 (thunderx2t99_sha): New Reservation.
25752 2017-04-25 Julian Brown <julian@codesourcery.com>
25753 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25755 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25756 type for 1-element load.
25758 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25760 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25762 2017-04-24 Martin Jambor <mjambor@suse.cz>
25764 PR tree-optimization/80293
25765 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25766 char arrays not totally scalarizable if it is false.
25767 (analyze_all_variable_accesses): Pass correct value in the new
25768 parameter. Add a statistics counter.
25770 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25772 PR middle-end/79931
25773 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25775 2017-04-24 Richard Biener <rguenther@suse.de>
25777 PR tree-optimization/80494
25778 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25779 out for complex types.
25781 2017-04-24 Richard Biener <rguenther@suse.de>
25783 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25784 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25785 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25786 (DFS): Adjust and never fail.
25787 (sccvn_dom_walker::fail): Remove.
25788 (sccvn_dom_walker::before_dom_children): Adjust.
25789 (run_scc_vn): Likewise and never fail.
25790 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25791 (pass_fre::execute): Likewise.
25793 2017-04-24 Richard Biener <rguenther@suse.de>
25795 PR tree-optimization/79725
25796 * tree-ssa-sink.c (statement_sink_location): Return whether
25797 failure reason was zero uses. Move that check later.
25798 (sink_code_in_bb): Deal with zero uses by removing the stmt
25801 2017-04-24 Richard Biener <rguenther@suse.de>
25804 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25805 pointer-based references.
25807 2017-04-24 Richard Biener <rguenther@suse.de>
25810 * pass_manager.h (pass_manager::operator new): Remove.
25811 (pass_manager::operator delete): Likewise.
25812 * passes.c (pass_manager::operator new): Remove.
25813 (pass_manager::operator delete): Likewise.
25814 (pass_manager::pass_manager): Zero individual pass members.
25816 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25819 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25820 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25821 Check "XEXP (src, 1)" operand here.
25822 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25823 Check "XEXP (src, 1)" operand here.
25824 (dimode_scalar_chain::make_vector_copies): Detect count register
25825 of a shift instruction. Zero extend count register from QImode
25826 to DImode to satisfy vector shift pattern count operand predicate.
25827 Substitute vector shift count operand with a DImode copy.
25828 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25831 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25833 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25834 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25835 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25836 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25837 (UNSPEC_NOREX_MEM): Remove definition.
25839 2017-04-21 Richard Biener <rguenther@suse.de>
25841 PR tree-optimization/79547
25842 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25843 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25844 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25845 without any constraints.
25847 2017-04-21 Richard Biener <rguenther@suse.de>
25849 PR tree-optimization/78847
25850 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25852 2017-04-21 Richard Biener <rguenther@suse.de>
25854 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25855 (build_distinct_type_copy): Likewise.
25856 (build_variant_type_copy): Likewise.
25857 * tree.c (build_qualified_type): Pass down mem-stat info.
25858 (build_distinct_type_copy): Likewise.
25859 (build_variant_type_copy): Likewise.
25861 2017-04-21 Richard Biener <rguenther@suse.de>
25863 PR tree-optimization/80237
25864 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25866 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25867 for a simplified result.
25869 2016-04-21 Richard Biener <rguenther@suse.de>
25871 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25872 sth as strict as a simple_iv but a chrec without symbols and an
25873 operand defined in the loop we are peeling (and not some subloop).
25874 (propagate_constants_for_unrolling): Propagate all constants.
25876 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25879 * config/i386/i386.c (print_reg): Remove assert for disalowed
25880 regno values, call output_operand_lossage instead.
25882 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25885 * config/i386/constraints.md (Yc): New register constraint.
25886 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25887 Use Yc constraint for alternative 2 of operand 0. Remove
25888 preferred_for_speed attribute.
25890 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25892 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25893 lastprivate clauses in SIMT case.
25895 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25897 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25899 2017-04-20 Richard Biener <rguenther@suse.de>
25901 PR tree-optimization/57796
25902 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25903 as N scalar stores.
25904 (vect_model_load_cost): Cost gathers as N scalar loads.
25906 2017-04-20 Richard Biener <rguenther@suse.de>
25908 * ggc-page.c (ggc_allocated_p): Rename to ...
25909 (safe_lookup_page_table_entry): ... this and return the lookup
25911 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
25913 2017-04-20 Richard Biener <rguenther@suse.de>
25915 PR tree-optimization/80453
25916 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
25917 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
25918 from the conditions.
25919 (vn_phi_eq): Pass them down.
25920 (vn_phi_lookup): Record them.
25921 (vn_phi_insert): Likewise.
25923 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
25925 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
25926 uninitialized variable warning to avoid buffer overrun.
25928 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25931 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
25932 is suppressed for '{ 0 }' in C.
25934 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25936 * BASE-VER: Set to 8.0.0.
25938 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
25940 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
25941 priority .init_array and .fini_array section with SECTION_NOTYPE
25944 2017-04-20 Jakub Jelinek <jakub@redhat.com>
25946 PR middle-end/80423
25947 * tree.h (build_array_type): Add typeless_storage default argument.
25948 * tree.c (type_cache_hasher::equal): Also compare
25949 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
25950 (build_array_type): Add typeless_storage argument, set
25951 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
25953 (build_nonshared_array_type): Adjust build_array_type_1 caller.
25954 (build_array_type): Likewise. Add typeless_storage argument.
25956 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25957 Jakub Jelinek <jakub@redhat.com>
25959 PR tree-optimization/80426
25960 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
25961 operation on symbolic operands, also compute the overflow for the
25962 invariant part when the operation degenerates into a negation.
25964 2017-04-19 Jakub Jelinek <jakub@redhat.com>
25967 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
25968 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
25971 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
25973 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
25976 * config/avr/avr.c (tree.h): Include it.
25977 (cgraph.h): Include it.
25978 (avr_encode_section_info): Don't warn for uninitialized progmem
25979 variable if it's just an alias.
25981 2017-04-19 Richard Biener <rguenther@suse.de>
25984 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
25985 when needed by AutoPGO.
25987 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
25990 * doc/lto.texi: Remove an extra 'that'.
25992 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
25994 PR rtl-optimization/80429
25995 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
25996 are only used in debug insns.
25998 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
25999 Vladimir Makarov <vmakarov@redhat.com>
26001 * config/sparc/predicates.md (input_operand): Add comment. Return
26002 true for any memory operand when LRA is in progress.
26003 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26005 2017-04-18 Jeff Law <law@redhat.com>
26008 * mips.md ({return,simple_return}_internal): Do not overwrite
26011 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26013 PR tree-optimization/80443
26014 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26015 instead of adding 1, subtract -1 and similarly instead of subtracting
26018 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26020 PR rtl-optimization/80357
26021 * haifa-sched.c (tmp_bitmap): New variable.
26022 (model_recompute): Handle duplicate use records.
26023 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26024 (free_global_sched_pressure_data): Free it.
26026 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26029 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26030 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26031 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26032 instead of SYSTEM_HEADER_DIR.
26034 2017-04-18 Jeff Law <law@redhat.com>
26036 PR middle-end/80422
26037 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26038 predecessors after walking up the insn chain.
26040 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26043 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26044 sizetype type into debug info.
26046 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26049 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26050 unneeded test for TARGET_UPPER_REGS_SF.
26051 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26053 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26056 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26057 instead of gsi_after_labels.
26059 2017-04-18 Jeff Law <law@redhat.com>
26061 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26065 2017-04-13 Jeff Law <law@redhat.com>
26066 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26067 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26069 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26072 * config/avr/avr.c (intl.h): Include it.
26073 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26075 2017-04-18 Martin Liska <mliska@suse.cz>
26077 PR gcov-profile/78783
26078 * gcov-tool.c (gcov_output_files): Validate that destination
26079 file is either removed by the tool or by a user.
26081 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26082 Guy Benyei <guybe@mellanox.com>
26084 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26085 block, and do not negate it, the stored id is already negative.
26087 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26089 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26091 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26094 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26095 masks of options that should be turned off if the VSX vector
26096 options are turned off.
26097 (OTHER_P8_VECTOR_MASKS): Likewise.
26098 (OTHER_VSX_VECTOR_MASKS): Likewise.
26099 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
26100 rs6000_disable_incompatible_switches to validate no type switches
26102 (rs6000_incompatible_switch): New function to disallow turning on
26103 other vector options if -mno-vsx, -mno-power8-vector, or
26104 -mno-power9-vector are specified.
26106 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26108 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
26110 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26112 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
26113 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
26114 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
26115 (ARG_POINTER_CFA_OFFSET): Likewise.
26117 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26119 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
26120 conditions to take advantage of various optimizations.
26122 2017-04-13 Jeff Law <law@redhat.com>
26124 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26125 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26126 (zero_extendsidi2_dext): Likewise.
26128 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26131 * fold-const.c (fold_ternary_loc): Revert
26132 use op0 instead of fold_convert_loc (loc, type, arg0) part of
26135 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
26137 PR rtl-optimization/80343
26138 * lra-remat.c (update_scratch_ops): Assign original hard reg to
26139 new scratch pseudo.
26141 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
26144 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
26145 to ubsan_encode_value.
26147 2017-04-13 Jeff Law <law@redhat.com>
26149 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
26150 appearing in DEBUG_INSNs.
26152 2017-04-13 Martin Liska <mliska@suse.cz>
26154 PR gcov-profile/80413
26155 * gcov-io.c (gcov_write_string): Copy to buffer just when
26156 allocated size is greater than zero.
26158 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26161 * dwarf2out.c (decls_for_scope): Ignore declarations of
26162 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26164 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26167 * ipa-visibility.c (non_local_p): Fix typos.
26168 (localize_node): When localizing symbol in same comdat group,
26169 dissolve the group only when we know external symbols are going
26171 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26173 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26175 PR tree-optimization/79390
26176 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26177 order does not result in usable sequence, retry with reversed operand
26183 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26184 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26185 op0 instead of fold_convert_loc (loc, type, arg0).
26187 2017-04-12 Jeff Law <law@redhat.com>
26189 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26190 has a delay slot in the generated code.
26192 * config/cris/cris.md (cris_preferred_reload_class): Return
26193 GENNONACR_REGS rather than GENERAL_REGS.
26195 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26198 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26199 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26200 signedness of the result type.
26202 2017-04-12 Richard Biener <rguenther@suse.de>
26203 Jeff Law <law@redhat.com>
26205 PR tree-optimization/80359
26206 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26207 trim stores to TARGET_MEM_REFs.
26209 2017-04-12 Richard Biener <rguenther@suse.de>
26211 PR tree-optimization/79390
26212 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26213 threading case even more.
26215 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26218 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26219 for quad_address_p for TImode, instead of just not indexed_address.
26221 2017-04-12 Richard Biener <rguenther@suse.de>
26222 Bernd Edlinger <bernd.edlinger@hotmail.de>
26224 PR middle-end/79671
26225 * alias.c (component_uses_parent_alias_set_from): Handle
26226 TYPE_TYPELESS_STORAGE.
26227 (get_alias_set): Likewise.
26228 * tree-core.h (tree_type_common): Add typeless_storage flag.
26229 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26230 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26231 for types containing members with TYPE_TYPELESS_STORAGE.
26232 (place_field): Likewise.
26233 (layout_type): Likewise for ARRAY_TYPE.
26234 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26235 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26236 TYPE_TYPELESS_STORAGE.
26237 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26239 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26242 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26243 first argument to type.
26245 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26249 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26250 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26251 (rs6000_expand_binop_builtin): Likewise.
26252 (rs6000_expand_ternop_builtin): Likewise; also add missing
26253 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26254 vshasigma built-ins.
26255 * doc/extend.texi: Document that vec_xxpermdi's third argument
26256 must be a constant.
26258 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26260 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26261 Use shift_const cost parameter when calculating gain of STV shifts.
26263 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26265 PR rtl-optimization/70478
26266 * lra-constraints.c (process_alt_operands): Check memory for
26267 disfavoring memory insn operand.
26269 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26271 PR middle-end/80100
26272 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26273 left shift in unsigned HOST_WIDE_INT type.
26275 PR rtl-optimization/80385
26276 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26277 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26280 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26281 if they have any depend clauses.
26283 2017-04-11 Martin Liska <mliska@suse.cz>
26286 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26287 * ipa-split.c (split_function): Create a local comdat symbol
26288 if caller is in a comdat group.
26290 2017-04-11 Martin Liska <mliska@suse.cz>
26293 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26296 2017-04-11 Martin Sebor <msebor@redhat.com>
26298 PR middle-end/80364
26299 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26300 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26302 (directive::set_width, directive::set_precision, format_character):
26304 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26307 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26310 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26311 conflict, set target->arch_name instead of target->cpu_name.
26313 2017-04-11 Richard Biener <rguenther@suse.de>
26315 PR tree-optimization/80374
26316 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26317 build_zero_cst, remove fold_convertible_p check again.
26319 2017-04-11 Martin Liska <mliska@suse.cz>
26322 * ubsan.c (instrument_object_size): Do not instrument register
26325 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26328 * config/i386/i386-builtin-types.def
26329 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26330 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26331 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26332 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26333 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26334 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26335 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26336 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26337 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26338 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26339 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26340 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26341 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26342 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26343 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26344 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26345 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26346 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26347 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26348 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26349 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26350 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26351 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26352 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26353 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26354 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26355 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26356 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26357 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26358 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26359 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26360 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26361 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26362 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26363 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26364 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26365 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26366 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26367 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26368 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26369 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26370 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26371 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26372 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26373 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26374 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26376 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26377 flag to second_arg_count, handle 4 argument function type _COUNT
26378 aliases, handle second_arg_count on second argument rather than last.
26380 2017-04-10 Jeff Law <law@redhat.com>
26382 PR tree-optimization/80374
26383 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26384 record anything if we can not convert integer_zero_node to the
26387 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26390 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26391 Enhance special handling given to the TARGET_P9_MINMAX option in
26392 relation to certain other options.
26394 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26396 PR tree-optimization/80153
26397 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26398 remove POINTER_PLUS_EXPR's base part directly, rather than through
26401 2017-04-10 Richard Biener <rguenther@suse.de>
26402 Bin Cheng <bin.cheng@arm.com>
26404 PR tree-optimization/80153
26405 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26406 the first element of pointer type aff_tree. Build result expr in
26408 (add_elt_to_tree): Convert to type unconditionally. Remove other
26409 fold_convert calls.
26410 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26411 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26413 2017-04-10 Richard Biener <rguenther@suse.de>
26415 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26418 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26420 PR rtl-optimization/70478
26421 * lra-constraints.c (curr_small_class_check): New.
26422 (update_and_check_small_class_inputs): New.
26423 (process_alt_operands): Update curr_small_class_check. Disfavor
26424 alternative insn memory operands. Check available regs for small
26427 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26430 * config/mips/mips.opt (-mvirt): Update description.
26431 * doc/invoke.texi (-mvirt): Likewise.
26433 2017-04-10 Richard Biener <rguenther@suse.de>
26435 PR middle-end/80362
26436 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26437 looking for NEGATE_EXPR in -A / -B to A / B folding.
26439 2017-04-10 Martin Liska <mliska@suse.cz>
26441 PR gcov-profile/80224
26442 * gcov.c (print_usage): Fix usage string.
26443 (get_gcov_intermediate_filename): Remove.
26444 (output_gcov_file): Use both for normal and intermediate format.
26445 (generate_results): Do not initialize special file for
26446 intermediate format.
26448 2017-04-10 Richard Biener <rguenther@suse.de>
26450 PR tree-optimization/80304
26451 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26454 2017-04-10 Nathan Sidwell <nathan@acm.org>
26457 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26458 (rs6000_init_builtins): Use it.
26460 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26462 * config/arm/arm.md (<mrc>): Add mode to SET source.
26463 (<mrrc>): Likewise.
26465 2017-04-10 Richard Biener <rguenther@suse.de>
26467 PR middle-end/80344
26468 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26470 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26473 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26474 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26475 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26476 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26477 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26478 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26479 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26480 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26481 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26482 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26483 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26484 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26485 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26486 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26487 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26488 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26489 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26490 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26491 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26492 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26493 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26494 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26495 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26497 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26499 PR rtl-optimization/70478
26500 * lra-constraints.c: Reverse the last patch.
26502 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26504 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26505 Add comment for WCHAR_T.
26507 2017-04-08 Martin Liska <mliska@suse.cz>
26510 2017-04-07 Martin Liska <mliska@suse.cz>
26513 * ipa-split.c (split_function): Add function part to a same comdat
26516 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26519 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26521 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26523 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26524 (vec_pairing): ...to this.
26525 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26526 (rs6000_sched_init): Adjust for name change.
26527 (struct rs6000_sched_context): Likewise.
26528 (rs6000_init_sched_context): Likewise.
26529 (rs6000_set_sched_context): Likewise.
26531 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26537 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26539 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26540 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26541 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26543 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26545 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26547 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26549 PR rtl-optimization/70703
26550 * ira-color.c (update_conflict_hard_regno_costs): Use
26551 int64_t instead of HOST_WIDE_INT.
26553 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26555 PR rtl-optimization/70478
26556 * lra-constraints.c (process_alt_operands): Disfavor alternative
26557 insn memory operands.
26559 2017-04-07 Jeff Law <law@redhat.com>
26561 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26562 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26564 2017-04-07 Martin Liska <mliska@suse.cz>
26567 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26568 Show error message instead of an ICE.
26570 2017-04-07 Martin Liska <mliska@suse.cz>
26573 * ipa-split.c (split_function): Add function part to a same comdat
26576 2017-04-07 Richard Biener <rguenther@suse.de>
26578 PR middle-end/80341
26579 * tree.c (get_unwidened): Also handle ! for_type case for
26581 * convert.c (do_narrow): Split out from ...
26582 (convert_to_integer_1): ... here. Do not pass final truncation
26583 type to get_unwidened for TRUNC_DIV_EXPR.
26585 2017-04-07 Richard Biener <rguenther@suse.de>
26587 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26589 (aff_combination_const): Adjust.
26590 (aff_combination_scale): Likewise.
26591 (aff_combination_add_elt): Likewise.
26592 (aff_combination_add_cst): Likewise.
26593 (aff_combination_convert): Likewise.
26594 (add_elt_to_tree): Likewise. Remove unused argument.
26595 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26597 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
26599 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
26601 * config/arm/arm.c (arm_default_short_enums): Use
26602 ARM_DEFAULT_SHORT_ENUMS.
26603 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
26605 2017-04-06 Jakub Jelinek <jakub@redhat.com>
26608 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
26609 members with redundant out-of-class redeclaration.
26611 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26614 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26615 * config/i386/i386.md (*zero_extendsidi2):
26616 Add (?*x,*x) and (?*v,*v) alternatives.
26618 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26621 * config/i386/i386.c (ix86_expand_builtin)
26622 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26623 mode from insn data. Convert operands to insn operand mode.
26624 Copy operands that don't satisfy insn predicate to a register.
26626 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26628 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26631 2017-04-06 Richard Biener <rguenther@suse.de>
26633 PR tree-optimization/80334
26634 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26635 preserve alignment of accesses.
26637 2017-04-06 Richard Biener <rguenther@suse.de>
26639 PR tree-optimization/80262
26640 * tree-sra.c (build_ref_for_offset): Preserve address-space
26642 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26643 Drop useless address-space information on MEM_REF offsets.
26645 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26647 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26649 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26651 PR rtl-optimization/70703
26652 * ira-color.c (update_conflict_hard_regno_costs): Use
26653 HOST_WIDE_INT instead of long.
26655 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26658 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26659 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26661 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26662 for x86_64 target. Handle -m3dnowa option.
26664 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26666 PR rtl-optimization/70703
26667 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26668 (update_conflict_hard_regno_costs): Use long instead of unsigned
26669 arithmetic for cost calculation.
26671 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26672 Bernd Edlinger <bernd.edlinger@hotmail.de>
26675 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26678 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26681 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26682 ptr_mode with Pmode throughout.
26683 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26684 into probe_stack_range and use DImode.
26686 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26689 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26690 call_eh_return is true.
26692 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26694 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26695 Initialize last_match_fntype_index.
26697 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26700 * tree-nvr.c: Include internal-fn.h.
26701 (pass_return_slot::execute): Ignore internal calls without
26704 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26705 Richard Biener <rguenther@suse.de>
26708 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26709 captures used multiple times, except for the last use.
26710 * generic-match-head.c: Include gimplify.h.
26712 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26714 PR tree-optimization/79390
26715 * target.h (struct noce_if_info): Declare.
26716 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26717 * target.def (noce_conversion_profitable_p): New target hook.
26718 * ifcvt.h (struct noce_if_info): New type, moved from ...
26719 * ifcvt.c (struct noce_if_info): ... here.
26720 (noce_conversion_profitable_p): Renamed to ...
26721 (default_noce_conversion_profitable_p): ... this. No longer
26723 (noce_try_store_flag_constants, noce_try_addcc,
26724 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26725 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26726 instead of noce_conversion_profitable_p.
26727 * config/i386/i386.c: Include ifcvt.h.
26728 (ix86_option_override_internal): Don't override
26729 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26730 (ix86_noce_conversion_profitable_p): New function.
26731 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26732 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26733 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26734 * doc/tm.texi: Regenerated.
26736 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26738 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26741 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26744 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26745 instructions for small multiply cores.
26747 2017-04-04 Jeff Law <law@redhat.com>
26749 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26751 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26752 that are not set by the loop over the elements.
26754 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26757 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26758 int mode, convert_modes it to mode as unsigned, otherwise use
26759 lowpart_subreg to mode rather than SImode.
26760 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26761 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26762 Use DImode instead of SImode for the shift count operand.
26763 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26766 2017-04-04 Richard Biener <rguenther@suse.de>
26768 PR middle-end/80281
26769 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26770 arithmetic done for the negate or the plus. Simplify.
26771 (A - (-B) -> A + B): Likewise.
26772 * fold-const.c (split_tree): Make sure to not negate pointers.
26774 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26776 PR rtl-optimization/60818
26777 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26778 a compare of comparisons with the thing compared if this results
26779 in a different machine mode.
26781 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26783 * alias.c (base_alias_check): Fix typo in comment.
26784 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26785 * cgraphunit.c (symbol_table::compile): Likewise.
26786 * collect2.c (maybe_run_lto_and_relink): Likewise.
26787 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26788 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26789 * config/avr/avr.c (avr_map_op_t): Likewise.
26790 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26791 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26792 * config/epiphany/epiphany.md (movcc): Likewise.
26793 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26794 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26796 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26797 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26798 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26799 * config/sh/sh.c (sh_rtx_costs): Likewise.
26800 * fold-const.c (fold_truth_andor): Likewise.
26801 * genautomata.c (collapse_flag): Likewise.
26802 * gengtype.h (struct type::u::s): Likewise.
26803 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26804 * input.c (FORMAT_AMOUNT): Likewise.
26805 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26806 (known_aggs_to_agg_replacement_list): Likewise.
26807 * ipa-inline-analysis.c: Likewise.
26808 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26809 * ipa-polymorphic-call.c
26810 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26811 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26812 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26814 * modulo-sched.c (apply_reg_moves): Likewise.
26815 * omp-expand.c (build_omp_regions_1): Likewise.
26816 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26817 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26818 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26819 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26820 * value-prof.c: Likewise.
26821 * var-tracking.c (val_reset): Likewise.
26823 2017-04-03 Richard Biener <rguenther@suse.de>
26825 PR tree-optimization/80275
26826 * fold-const.c (split_address_to_core_and_offset): Handle
26829 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26831 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26832 descriptors is at least equal to that of functions.
26834 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26836 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26838 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26841 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26842 (mov<IMOD4:mode>): New expander.
26843 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26845 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26847 PR rtl-optimization/79405
26848 * fwprop.c (propagations_left): New variable.
26849 (forward_propagate_into): Decrement it.
26850 (fwprop_init): Initialize it.
26851 (fw_prop): If the variable has reached zero, stop propagating.
26852 (fwprop_addr): Ditto.
26854 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26857 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26858 a FUNCTION_DECL, pass it as decl instead of origin to
26861 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26863 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26866 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26869 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26870 TARGET_VSX_SMALL_INTEGER.
26872 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26874 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26875 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26877 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26879 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26880 extraction from odd-numbered MSA register.
26882 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26884 PR middle-end/80173
26885 * expmed.c (store_bit_field_1): Don't attempt to create
26886 a word subreg out of hard registers wider than word if they
26887 have HARD_REGNO_NREGS of 1 for their mode.
26889 PR middle-end/80163
26890 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26891 conversions to integer types wider than word and pointer.
26894 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26895 (rtx_equal_for_cselib_p): Pass 0 to it.
26896 * cselib.c (cselib_hasher::equal): Likewise.
26897 (rtx_equal_for_cselib_1): Add depth argument. If depth
26898 is 128, don't look up VALUE locs and punt. Increment
26899 depth in recursive calls when walking VALUE locs.
26901 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26903 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26904 (make_gcov_file_name): Use the canonical path name for generating
26906 (read_line): Fix handling of files with ascii null bytes.
26908 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26910 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
26911 to initialise a vector register instead
26912 of using a const_int.
26914 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26916 PR translation/80189
26917 * gimplify.c (omp_default_clause): Use %qs instead of %s in
26918 diagnostic messages.
26920 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
26923 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
26924 (dfp_diex_<mode>): Update mode of operand 1.
26925 * doc/extend.texi (dxex, dxexq): Document change to return type.
26926 (diex, diexq): Document change to argument type.
26928 2017-03-30 Martin Jambor <mjambor@suse.cz>
26931 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
26932 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
26933 it reflects the signature changes performed at the callee side.
26934 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
26935 to cgraph_build_function_type_skip_args.
26936 (build_function_decl_skip_args): Adjust call to the above function.
26938 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26941 * config/i386/sse.md
26942 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
26943 register as dest whenever it is a MEM not rtx_equal_p to the
26944 corresponding dup operand, and when forcing into reg move the
26945 reg into the memory afterwards.
26946 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
26947 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
26948 for the force_reg mode.
26949 (avx512vl_vextractf128<mode>): Use register as dest either
26950 always when a MEM, or when it is a MEM not rtx_equal_p to the
26951 corresponding dup operand, or even not when it is a CONST_VECTOR
26952 depending on the mode and lo vs. hi.
26953 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
26955 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
26956 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
26957 Likewise. Require that operands[2] is even.
26958 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
26959 Remove extraneous parens. Require that operands[2] is a multiple
26961 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
26962 operands[0] is a MEM if <mask_applied>, the predicates/constraints
26963 disallow memory then.
26965 2017-03-30 Richard Biener <rguenther@suse.de>
26967 PR tree-optimization/77498
26968 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
26969 to non-constants over backedges.
26971 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
26973 PR rtl-optimization/80233
26974 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
26975 as last_combined_insn. Do not test for BARRIER_P separately.
26977 2017-03-29 Andreas Schwab <schwab@suse.de>
26980 * calls.c (prepare_call_address): Convert funexp to Pmode before
26981 copying to temp reg.
26983 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26985 PR tree-optimization/80158
26986 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
26987 Handle possible future case of more than one alternate
26989 (replace_rhs_if_not_dup): Likewise.
26990 (replace_one_candidate): Likewise.
26992 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
26994 PR rtl-optimization/80193
26995 * ira.c (ira): Do not check allocation for LRA.
26997 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
26999 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27000 (nvptx_output_simt_exit): Declare.
27001 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27002 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27003 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27004 (nvptx_declare_function_name): ...here. Emit declaration of local
27005 memory space buffer for omp_simt_enter insn.
27006 (nvptx_output_unisimt_switch): New.
27007 (nvptx_output_softstack_switch): New.
27008 (nvptx_output_simt_enter): New.
27009 (nvptx_output_simt_exit): New.
27010 * config/nvptx/nvptx.h (struct machine_function): New fields
27011 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27012 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27013 (UNSPECV_SIMT_EXIT): Ditto.
27014 (omp_simt_enter_insn): New insn.
27015 (omp_simt_enter): New expansion.
27016 (omp_simt_exit): New insn.
27017 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27019 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27020 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27021 (expand_GOMP_SIMT_EXIT): New.
27022 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27023 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27024 (GOMP_SIMT_EXIT): Ditto.
27025 * target-insns.def (omp_simt_enter): New insn.
27026 (omp_simt_exit): Ditto.
27027 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27029 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27030 (lower_rec_input_clauses): Likewise.
27031 (lower_lastprivate_clauses): Handle SIMT privatization.
27033 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27034 (ompdevlow_adjust_simt_enter): New.
27035 (find_simtpriv_var_op): New.
27036 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27037 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27039 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27040 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27041 (copy_decl_for_dup_finish): Ditto.
27043 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27045 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27048 * config/i386/i386.c (ix86_option_override_internal): Always
27049 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27051 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27053 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27055 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27057 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27058 mark new edge's irreducible flag accordign to it.
27059 (vect_do_peeling): Check loop preheader edge's irreducible flag
27060 and pass it to function slpeel_add_loop_guard.
27062 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27064 PR tree-optimization/80218
27065 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27066 Update block frequencies and counts.
27068 2017-03-28 Richard Biener <rguenther@suse.de>
27070 PR tree-optimization/78644
27071 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27072 of a simplification result we may not use it at all.
27074 2017-03-28 Richard Biener <rguenther@suse.de>
27077 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27078 without arguments, generate default definition of a SSA name.
27080 2017-03-28 Richard Biener <rguenther@suse.de>
27082 PR middle-end/80222
27083 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27084 TYPE_REF_CAN_ALIAS_ALL references.
27085 * fold-const.c (fold_indirect_ref_1): Likewise.
27087 2017-03-28 Martin Liska <mliska@suse.cz>
27090 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27091 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27093 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27094 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27096 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27097 (EXTRA_SPECS): Define.
27098 (SUBTARGET_EXTRA_SPECS): Likewise.
27099 (SUBTARGET_CPP_SPEC): Likewise.
27100 * config/arc/elf.h (EXTRA_SPECS): Renamed to
27101 SUBTARGET_EXTRA_SPECS.
27102 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
27104 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27106 * config/arc/simdext.md (vst64_insn): Update pattern.
27107 (vld32wh_insn): Likewise.
27108 (vld32wl_insn): Likewise.
27109 (vld64_insn): Likewise.
27110 (vld32_insn): Likewise.
27112 2017-03-28 Marek Polacek <polacek@redhat.com>
27115 * fold-const.c (fold_comparison): Use protected_set_expr_location
27116 instead of SET_EXPR_LOCATION.
27118 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
27120 * tree.c (add_expr): Avoid name lookup warning.
27122 2017-03-27 Jeff Law <law@redhat.com>
27124 PR tree-optimization/80216
27125 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
27126 function name. Limit recursion depth.
27127 (record_temporary_equivalences): Corresponding changes.
27129 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
27131 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
27134 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27137 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
27139 * cfgcleanup.c (reg_note_cfa_p): New array.
27140 (insns_have_identical_cfa_notes): New function.
27141 (old_insns_match_p): Don't cross-jump in between /f
27142 and non-/f instructions. If both i1 and i2 are frame related,
27143 verify all CFA notes, their order and content.
27145 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
27148 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
27149 HImode and SImode with zero extend to DImode to one insn.
27150 (bswap<mode>2_extenddi): Likewise.
27151 (bswapsi2_extenddi): Likewise.
27152 (bswaphi2_extendsi): Likewise.
27153 (bswaphi2): Combine bswap HImode and SImode into one insn.
27154 Separate memory insns from swapping register.
27155 (bswapsi2): Likewise.
27156 (bswap<mode>2): Likewise.
27157 (bswaphi2_internal): Delete, no longer used.
27158 (bswapsi2_internal): Likewise.
27159 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
27160 store, and gpr<-gpr swap insns.
27161 (bswap<mode>2_store): Likewise.
27162 (bswaphi2_reg): Register only splitter, combine with the splitter.
27163 (bswaphi2 splitter): Likewise.
27164 (bswapsi2_reg): Likewise.
27165 (bswapsi2 splitter): Likewise.
27166 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27167 the insns into load, store, and register/register insns.
27168 (bswapdi2_ldbrx): Likewise.
27169 (bswapdi2_load): Likewise.
27170 (bswapdi2_store): Likewise.
27171 (bswapdi2_reg): Likewise.
27173 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27175 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27176 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27178 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27181 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27183 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27184 special handling for target option conflicts between dform
27185 options (-mpower9-dform, -mpower9-dform-vector,
27186 -mpower9-dform-scalar) and -mno-direct-move.
27188 2017-03-27 Richard Biener <rguenther@suse.de>
27190 PR tree-optimization/80181
27191 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27193 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27195 * config/arc/predicates.md (move_double_src_operand): Replace the
27196 call to move_double_src_operand with a call to address_operand.
27198 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27200 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27201 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27202 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27204 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27206 * config/arc/predicates.md (long_immediate_loadstore_operand):
27207 Consider scaled addresses cases.
27209 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27211 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27212 restored when in interrupt.
27213 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27214 doesn't have delay slot.
27216 2017-03-27 Richard Biener <rguenther@suse.de>
27219 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27220 inlined thunk clones.
27222 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27225 * asan.c (instrument_derefs): Copy over last operand from
27226 original COMPONENT_REF to the new COMPONENT_REF with
27227 DECL_BIT_FIELD_REPRESENTATIVE.
27228 * ubsan.c (instrument_object_size): Likewise.
27230 2017-03-27 Richard Biener <rguenther@suse.de>
27232 PR tree-optimization/80170
27233 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27234 sure DR/SCEV didnt fold in constants we do not see when looking
27235 at the reference base alignment.
27237 2017-03-27 Richard Biener <rguenther@suse.de>
27239 PR middle-end/80171
27240 * gimple-fold.c (fold_ctor_reference): Properly guard against
27241 NULL return value from canonicalize_constructor_val.
27243 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27246 * config/i386/i386.c (ix86_expand_builtin)
27247 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27248 flags reg setting and flags reg using instructions.
27249 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27250 clobbering instructions to zero extend op2.
27252 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27254 * doc/install.texi (Configuration) <--with-aix-soname>:
27255 Update link to AIX ld.
27257 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27259 PR rtl-optimization/80160
27260 PR rtl-optimization/80159
27261 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27262 reg_alternate_class into account.
27264 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27267 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27268 to consider in curr_insn_transform.
27270 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27272 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27273 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27274 and emit_mode_inner.
27276 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27278 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27279 argument to the overloaded builtin variants. Use the new flag to
27280 deprecate certain builtin variants.
27281 * config/s390/s390-builtin-types.def: Add new builtin types.
27282 * config/s390/s390-builtins.h: Support new flags field for
27283 overloaded builtins.
27284 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27285 (s390_macro_to_expand): Enable vector float data type.
27286 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27287 builtins by incrementing the __VEC__ version number.
27288 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27290 (s390_resolve_overloaded_builtin): Emit error messages depending
27291 on the builtin flags.
27292 * config/s390/s390.c (s390_expand_builtin): Support additional
27293 flags argument. Change error message to match the messages
27294 emitted in s390-c.c.
27295 * config/s390/s390.md: New UNSPEC_* constants.
27296 (op_type): Add new instruction types.
27297 * config/s390/vecintrin.h: Add new builtins and test data class
27299 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27300 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27301 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27302 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27303 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27304 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27306 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27307 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27308 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27309 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27311 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27312 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27313 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27314 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27316 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27317 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27318 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27319 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27320 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27321 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27322 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27324 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27325 ("vec_scatter_element<V_HW_4:mode>_DI")
27326 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27327 ("vec_fpint<mode>", "vflls")
27328 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27329 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27330 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27331 ("*vec_cmphe<mode>_cc"): ... these.
27333 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27334 mode constant instead of magic value.
27336 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27338 * config/s390/s390.c (s390_expand_vec_compare): Support other
27339 vector floating point modes than just V2DF.
27340 (s390_expand_vcond): Likewise.
27341 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27342 (s390_cannot_change_mode_class): Prevent mode changes between TF
27343 and V1TF in vector registers.
27344 * config/s390/s390.md (DF, SF): New mode attributes.
27345 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27346 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27347 SFmode support for VRs.
27348 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27350 (VFT, VF_HW): New mode iterators.
27351 (vw, sdx): New mode attributes.
27352 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27353 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27354 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27355 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27356 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27357 also the new vector floating point modes. Renaming to ...
27359 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27360 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27361 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27362 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27363 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27364 ("vec_unordered<mode>"): ... these.
27366 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27367 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27368 ("*vec_extendv2df"): New insn definitions.
27370 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27372 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27373 ("mulditi3_2", "*muldi3_sign"): New patterns.
27374 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27375 rename the pattern definition.
27377 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27379 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27381 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27383 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27385 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27386 instruction if possible.
27387 * config/s390/vector.md (vec_halfnumelts): New mode
27389 ("*vec_vllezlf<mode>"): New pattern.
27391 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27393 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27394 ("popcountv4si2", "popcountv2di2"): Rename to ...
27395 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27396 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27398 ("popcount<mode>2_vxe"): New pattern.
27400 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27402 * common/config/s390/s390-common.c (processor_flags_table): Add
27404 * config.gcc: Add arch12.
27405 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27406 Default to arch12 for unknown CPU model numbers.
27407 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27408 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27409 PROCESSOR_max sanity check.
27410 * config/s390/s390-opts.h (enum processor_type): Add
27412 * config/s390/s390.c (processor_table): Add arch12.
27413 (s390_expand_builtin): Add check for B_VXE flag.
27414 (s390_issue_rate): Add PROCESSOR_ARCH12.
27415 (s390_get_sched_attrmask): Likewise.
27416 (s390_get_unit_mask): Likewise.
27417 (s390_sched_score): Enable z13 scheduling for arch12.
27418 (s390_sched_reorder): Likewise.
27419 (s390_sched_variable_issue): Likewise.
27420 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27422 (s390_tune_attr): Use z13 scheduling also for arch12.
27423 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27424 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27425 (TARGET_VXE_P): New macros.
27426 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27427 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27428 * config/s390/s390.opt: Add arch12 as processor_type.
27430 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27432 * config/s390/s390.md
27433 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27434 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27435 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27437 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27438 Rename expanders to ...
27440 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27441 ("fixuns_truncdddi2_emu"): ... these.
27443 ("fixuns_trunc<mode>si2_emu"): New expander.
27445 ("*fixuns_truncdfdi2_z13"): Rename to ...
27446 ("*fixuns_truncdfdi2_vx"): ... this.
27448 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27450 * config/s390/2964.md: Remove the single element vector compare
27451 instructions which are no longer used.
27452 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27454 (s390_canonicalize_comparison): Remove handling of DFmode
27456 (s390_expand_vec_compare_scalar): Remove function.
27457 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27458 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27460 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27462 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27464 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27466 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27467 will anyway by matched by mov<mode>_64dfp.
27469 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27471 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27472 vlef/vstef. Add missing operand to vleif.
27474 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27476 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27477 pair for all vector types with 64 bit elements.
27478 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27479 * config/s390/vector.md (V_HW_64): ... here.
27480 (V_128_NOSINGLE): New mode iterator.
27481 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27482 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27483 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27484 ("*vec_load_pairv2di"): Change to ...
27485 ("*vec_load_pair<mode>"): ... this one.
27487 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27489 * config/s390/constraints.md: Add comments.
27490 (jKK): Reject element sizes > 8 bytes.
27491 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27493 * config/s390/s390.md: Add the s_operand checks formerly in
27494 s390_split_ok_p to various splitters where they are still
27496 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27497 for 128 bit vectors. Plus two splitters.
27499 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27501 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27504 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27507 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27508 error if the boundary argument is not constant.
27510 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27512 PR rtl-optimization/80112
27513 * loop-doloop.c (doloop_condition_get): Don't check condition
27514 if cmp isn't SET with IF_THEN_ELSE src.
27516 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27518 PR tree-optimization/80158
27519 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27520 replacing a candidate statement, also replace it for the
27521 candidate's alternate interpretation.
27522 (replace_rhs_if_not_dup): Likewise.
27523 (replace_one_candidate): Likewise.
27525 2017-03-24 Richard Biener <rguenther@suse.de>
27527 PR tree-optimization/80167
27528 * graphite-isl-ast-to-gimple.c
27529 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27531 (translate_isl_ast_to_gimple::get_rename): Likewise.
27533 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27535 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27536 handling of certain combinations of target options, including the
27537 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27538 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27540 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27543 * config/arm/arm.md (*load_multiple): Add reload_completed to
27544 matching condition.
27546 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27547 Richard Biener <rguenth@suse.de>
27549 PR tree-optimization/79908
27550 PR tree-optimization/80136
27551 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27552 been cast away, gimplify_and_add suffices.
27554 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27556 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27558 2017-03-23 Richard Biener <rguenther@suse.de>
27560 PR tree-optimization/80032
27561 * gimplify.c (gimple_push_cleanup): Forced unconditional
27562 cleanups still have to go to the conditional_cleanups
27565 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27567 PR tree-optimization/80072
27568 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27570 (next_operand_entry_id): Change type to unsigned int.
27571 (sort_by_operand_rank): Make sure to return the right return value
27572 even if unsigned fields are bigger than INT_MAX.
27573 (struct oecount): Change cnt and id type to unsigned int.
27574 (oecount_hasher::equal): Formatting fix.
27575 (oecount_cmp): Make sure to return the right return value
27576 even if unsigned fields are bigger than INT_MAX.
27577 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27580 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27581 TREE_READONLY on result if writing it more than once.
27584 * doc/invoke.texi (-fsanitize=thread): Document that with
27585 -fnon-call-exceptions atomics are not able to throw
27589 * tsan.c: Include tree-eh.h.
27590 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27591 maybe_clean_or_replace_eh_stmt where needed.
27592 (instrument_memory_accesses): Add cfg_changed argument.
27593 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27594 if it returned true.
27595 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27597 PR rtl-optimization/63191
27598 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
27599 wrapper function, moved the whole old content into ...
27600 (ix86_delegitimize_address_1): ... this. New inline function.
27601 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
27602 true as last argument instead of ix86_delegitimize_address.
27604 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27606 * config/aarch64/aarch64.c (generic_branch_cost): Copy
27607 cortexa57_branch_cost.
27609 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27611 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
27613 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27616 * doc/md.texi (Constraints): Document wA constraint.
27617 * config/rs6000/constraints.md (wA): New.
27618 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27619 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27620 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27621 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27623 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27626 * gimplify.c (is_oacc_declared): New function.
27627 (oacc_default_clause): Use it to set default flags for acc declared
27628 variables inside parallel regions.
27629 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27630 declared variables.
27631 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27632 declare attribute to any decl as necessary.
27634 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27637 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27638 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27639 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27640 (arm_arch_lpae): This.
27641 * config/arm/arm.c (arm_arch7ve): Rename into ...
27642 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27643 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27646 2017-03-22 Martin Liska <mliska@suse.cz>
27649 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27650 error message instead of an ICE.
27652 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27654 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27656 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27658 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27660 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27663 2017-03-21 Martin Sebor <msebor@redhat.com>
27665 * doc/extend.texi: Use "cannot" instead of "can't."
27666 * doc/hostconfig.texi: Same.
27667 * doc/install.texi: Same.
27668 * doc/invoke.texi: Same.
27669 * doc/loop.texi: Same.
27670 * doc/md.texi: Same.
27671 * doc/objc.texi: Same.
27672 * doc/rtl.texi: Same.
27673 * doc/tm.texi: Same.
27674 * doc/tm.texi.in: Same.
27675 * doc/trouble.texi: Same.
27677 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27680 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27681 (collect_checksum_attributes): Set it.
27682 (die_checksum_ordered): Use it.
27684 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27686 PR tree-optimization/79908
27687 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27688 change: For a VA_ARG whose LHS has been cast away, use
27689 force_gimple_operand to construct the side effects.
27691 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27693 PR translation/80001
27694 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27695 more amenable to translation.
27696 (oacc_loop_auto_partitions): Likewise.
27698 2017-03-21 Marek Polacek <polacek@redhat.com>
27699 Martin Sebor <msebor@redhat.com>
27701 PR tree-optimization/80109
27702 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27703 on INTEGRAL_TYPE_P.
27705 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27706 Segher Boessenkool <segher@kernel.crashing.org>
27709 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27710 check reg_used_between_p between insn and one of succ or succ2
27711 depending on if succ is artificial insn not inserted into insn
27714 2017-03-21 Martin Liska <mliska@suse.cz>
27716 PR gcov-profile/80081
27717 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27718 * doc/gcc.texi: Include gcov-dump stuff.
27719 * doc/gcov-dump.texi: New file.
27721 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27723 PR rtl-optimization/79150
27724 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27725 conditional jump, if the jump is the last insn of the loop.
27727 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27728 Richard Biener <rguenth@suse.de>
27730 PR tree-optimization/79908
27731 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27732 been cast away, use force_gimple_operand to construct the side
27735 2017-03-21 Martin Liska <mliska@suse.cz>
27737 PR libfortran/79956
27738 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27741 2017-03-21 Brad Spengler <spender@grsecurity.net>
27744 * plugin.c (htab_hash_plugin): New function.
27745 (add_new_plugin): Use it and adjust.
27746 (parse_plugin_arg_opt): Adjust.
27747 (init_one_plugin): Likewise.
27749 2017-03-21 Richard Biener <rguenther@suse.de>
27751 PR tree-optimization/80032
27752 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27753 if set force the cleanup to happen unconditionally.
27754 (gimplify_target_expr): Push inserted clobbers with force_uncond
27755 to avoid them being removed by control-dependent DCE.
27757 2017-03-21 Richard Biener <rguenther@suse.de>
27759 PR tree-optimization/80122
27760 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27761 va_arg_pack_len when the inlined call stmt requires pack
27763 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27765 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27768 * tsan.c (instrument_builtin_call): If the memory model argument
27769 is not a constant, assume it is valid.
27772 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27775 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27777 PR rtl-optimization/79910
27778 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27779 if its dest is used by an insn before I2 (other than the combined
27780 insns themselves, which are properly handled already).
27782 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27785 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27787 * combine.c (record_used_regs): New static function.
27788 (try_combine): Handle situations where there is an additional
27789 instruction between I2 and I3 which needs to have a LOG_LINK
27793 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27795 * combine.c (try_combine): Delete redundant i1 test. Call
27796 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27798 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27801 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27802 alternatives 13/14.
27804 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27806 PR tree-optimization/80054
27807 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27808 the optimization if a PHI or any of its arguments is not dominated
27809 by the candidate's basis. Use gphi* rather than gimple* as
27811 (replace_profitable_candidates): Clean up a gimple* variable that
27812 should be a gphi* variable.
27814 2017-03-20 Martin Sebor <msebor@redhat.com>
27817 * doc/extend.texi (attribute constructor): Document present limitation.
27819 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27822 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27823 __POWER9_VECTOR__ #ifdef control, change template definition to
27824 use Power9-specific built-in function.
27825 (vec_any_eq): Likewise.
27826 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27827 to control outcomes from this test.
27828 (vector_ae_<mode>p): For VEC_F modes, likewise.
27830 2017-03-20 Ian Lance Taylor <iant@google.com>
27832 * config/i386/i386.c (ix86_function_regparm): Save an extra
27833 register for -fsplit-stack with DECL_STATIC_CHAIN.
27835 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27838 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27839 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27841 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27843 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27845 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27848 2017-03-20 Marek Polacek <polacek@redhat.com>
27851 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27853 2017-03-20 Richard Biener <rguenther@suse.de>
27855 PR tree-optimization/80113
27856 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27857 allocate extra SSA name for PHI def.
27858 (add_close_phis_to_outer_loops): Likewise.
27859 (add_close_phis_to_merge_points): Likewise.
27860 (copy_loop_close_phi_args): Likewise.
27861 (copy_cond_phi_nodes): Likewise.
27863 2017-03-20 Martin Liska <mliska@suse.cz>
27865 PR middle-end/79753
27866 * tree-chkp.c (chkp_build_returned_bound): Do not build
27867 returned bounds for a LHS that's not a BOUNDED_P type.
27869 2017-03-20 Martin Liska <mliska@suse.cz>
27873 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27874 COMPLEX_CST and VECTOR_CST.
27876 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27879 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27880 target operand. A new splitter adds the clobber statement in case
27881 the target operand is dead anyway.
27883 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27885 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27886 to age-old versions of binutils and glibc.
27888 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27890 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27892 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27894 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27896 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27898 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27899 requirement for binutils 2.13.
27901 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27903 * combine.c (try_combine): Delete redundant i1 test. Call
27904 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27906 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27908 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27909 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
27911 <riscv64-*-elf>: Re-arrange section
27912 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
27913 <riscv32-*-linux>: Likewise.
27914 <riscv64-*-elf>: Likewise
27915 <riscv64-*-linux>: Likewise.
27917 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
27920 * aarch64.opt(verbose-cost-dump): Fix typo.
27922 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
27925 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
27926 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
27928 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27930 * reload.c (find_reloads): When reloading a nonoffsettable address,
27931 use RELOAD_OTHER for it and its address reloads.
27933 PR rtl-optimization/79910
27934 * combine.c (record_used_regs): New static function.
27935 (try_combine): Handle situations where there is an additional
27936 instruction between I2 and I3 which needs to have a LOG_LINK
27939 2017-03-17 Jeff Law <law@redhat.com>
27941 PR tree-optimization/71437
27942 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
27943 conditional in the hash table first.
27944 (vrp_dom_walker::before_dom_children): Extract condition from
27945 ASSERT_EXPR. Record condition, its inverion and any implied
27946 conditions as well.
27948 2017-03-17 Marek Polacek <polacek@redhat.com>
27949 Markus Trippelsdorf <markus@trippelsdorf.de>
27951 PR tree-optimization/80079
27952 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
27955 2017-03-17 Richard Biener <rguenther@suse.de>
27957 PR middle-end/80075
27958 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
27959 Properly verify the LHS before the RHS possibly claims to be
27961 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
27964 2017-03-17 Martin Jambor <mjambor@suse.cz>
27966 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
27967 (List of -O2 options): Likewise.
27968 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
27971 2017-03-17 Tom de Vries <tom@codesourcery.com>
27973 * gcov-dump.c (print_usage): Print bug_report_url.
27975 2017-03-17 Richard Biener <rguenther@suse.de>
27977 PR middle-end/80050
27978 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
27979 (parser::peek): Likewise.
27981 2017-03-17 Richard Biener <rguenther@suse.de>
27983 PR tree-optimization/80048
27984 * sese.c (free_sese_info): Properly release rename_map and
27985 copied_bb_map elements.
27987 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
27989 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
27990 Add linked-list forward and backlinks. Insert on
27991 construction, remove on destruction.
27992 (class pass_store_merging): Add m_stores_head field.
27993 (pass_store_merging::terminate_and_process_all_chains):
27994 Iterate over m_stores_head list.
27995 (pass_store_merging::terminate_all_aliasing_chains):
27997 (pass_store_merging::execute): Check for debug stmts first.
27998 Push new chains onto the m_stores_head stack.
28000 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28003 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28004 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28005 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28007 2017-03-16 Jeff Law <law@redhat.com>
28009 PR tree-optimization/71437
28010 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28011 member function. Implementation moved into after_dom_children
28012 member function and into the threader's thread_outgoing_edges
28014 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28015 some code into new thread_outgoing_edges.
28016 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28017 definition. Simplify marker handling (do it here). Assume we always
28018 have the available expression and the const/copies tables.
28019 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28021 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28022 * tree-vrp.c (equiv_stack): No longer file scoped.
28023 (vrp_dom_walker): New class.
28024 (vrp_dom_walker::before_dom_children): New member function.
28025 (vrp_dom_walker::after_dom_children): Likewise.
28026 (identify_jump_threads): Setup domwalker. Use it rather than
28027 walking edges in a random order by hand. Simplify setup/finalization.
28028 (finalize_jump_threads): Remove.
28029 (vrp_finalize): Do not call identify_jump_threads here.
28030 (execute_vrp): Do it here instead and call thread_through_all_blocks
28033 PR tree-optimization/71437
28034 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28036 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28038 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28039 (dom_opt_dom_walker::thread_across_edge): Remove
28040 handle_dominating_asserts argument. All callers changed.
28041 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28042 changes. Remove calls to lhs_of_dominating_assert. Other
28043 uses of handle_dominating_asserts turn into unconditional code
28044 (simplify_control_stmt_condition_1): Likewise.
28045 (simplify_control_stmt_condition): Likewise.
28046 (thread_through_normal_block, thread_across_edge): Likewise.
28047 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28048 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28049 object if it is not an SSA_NAME.
28050 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28051 before calling into the VRP specific simplifiers.
28052 (identify_jump_threads): Remove handle_dominating_asserts
28055 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28058 * tree-diagnostic.c (default_tree_printer): No longer static.
28059 * tree-diagnostic.h (default_tree_printer): New prototype.
28061 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28063 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28064 Change ins into fmov.
28066 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28068 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28069 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28070 Use h_con constraint for operand 1.
28071 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28072 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28074 2017-03-15 Jeff Law <law@redhat.com>
28076 PR tree-optimization/71437
28077 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28078 (record_temporary_equivalences): Use it.
28080 PR tree-optimization/71437
28081 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28082 tree-ssa-scopedtables.
28083 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28084 (record_conditions, record_cond, vuse_eq): Likewise.
28085 (record_edge_info): Adjust to API tweak of record_conditions.
28086 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28087 (record_temporary_equivalences, optimize_stmt): Likewise.
28088 (eliminate_redundant_computations): Likewise.
28089 (record_equivalences_from_stmt): Likewise.
28090 * tree-ssa-scopedtables.c: Include options.h and params.h.
28091 (vuse_eq): New function, moved from tree-ssa-dom.c
28092 (build_and_record_new_cond): Likewise.
28093 (record_conditions): Likewise. Accept vector of conditions rather
28094 than edge_equivalence structure for first argument.
28095 for the first argument.
28096 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28097 from tree-ssa-dom.c.
28098 (avail_exprs_stack::record_cond): Likewise.
28099 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
28100 from tree-ssa-dom.c.
28101 (avail_exprs_stack): Add new member functions lookup_avail_expr
28103 (record_conditions): Declare.
28105 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
28108 * lra-constraints.c (process_alt_operands): Increase reject for
28109 reloading an input/output operand.
28111 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28114 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
28115 insns to convert from signed/unsigned char/short to IEEE 128-bit
28117 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
28119 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
28122 * config/i386/i386.c (ix86_vector_duplicate_value): Create
28123 subreg of inner mode for values already in registers.
28125 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
28127 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
28128 iteration reg is used after the loop.
28130 2017-03-14 Martin Sebor <msebor@redhat.com>
28132 PR tree-optimization/79800
28133 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
28134 precision in negative-positive range.
28135 (format_floating): Call non-const overload with adjusted precision.
28137 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28140 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
28143 2017-03-14 Martin Sebor <msebor@redhat.com>
28145 PR middle-end/80020
28146 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
28147 * builtins.def (aligned_alloc): Use it.
28150 * Makefile.in (GTFILES): Add calls.c.
28151 * calls.c: Include "gt-calls.h".
28153 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
28155 PR rtl-optimization/79728
28156 * regs.h (struct target_regs): New field
28157 x_contains_allocatable_regs_of_mode.
28158 (contains_allocatable_regs_of_mode): New macro.
28159 * reginfo.c (init_reg_sets_1): Initialize it, and change
28160 contains_reg_of_mode so it includes global regs as well.
28161 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28162 rather than contains_regs_of_mode.
28164 2017-03-14 Martin Liska <mliska@suse.cz>
28166 * doc/invoke.texi: Document options that can't be combined with
28167 -fcheck-pointer-bounds.
28169 2017-03-14 Martin Liska <mliska@suse.cz>
28171 PR middle-end/79831
28172 * doc/invoke.texi (-Wchkp): Document the option.
28174 2017-03-14 Martin Liska <mliska@suse.cz>
28176 * Makefile.in: Install gcov-dump.
28178 2017-03-14 Martin Liska <mliska@suse.cz>
28180 * multiple_target.c (expand_target_clones): Bail out for
28181 an invalid attribute.
28183 2017-03-14 Richard Biener <rguenther@suse.de>
28185 * alias.c (struct alias_set_entry): Pack properly.
28186 * cfgloop.h (struct loop): Likewise.
28187 * cse.c (struct set): Likewise.
28188 * ipa-utils.c (struct searchc_env): Likewise.
28189 * loop-invariant.c (struct invariant): Likewise.
28190 * lra-remat.c (struct cand): Likewise.
28191 * recog.c (struct change_t): Likewise.
28192 * rtl.h (struct address_info): Likewise.
28193 * symbol-summary.h (function_summary): Likewise.
28194 * tree-loop-distribution.c (struct partition): Likewise.
28195 * tree-object-size.c (struct object_size_info): Likewise.
28196 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28197 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28198 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28199 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28200 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28201 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28202 (struct _stmt_vec_info): Likewise.
28204 2017-03-14 Martin Liska <mliska@suse.cz>
28207 * multiple_target.c (create_dispatcher_calls): Check that
28208 a target can create a function dispatcher.
28210 2017-03-14 Martin Liska <mliska@suse.cz>
28213 * multiple_target.c (expand_target_clones): Drop local.local
28214 flag for default implementation.
28216 2017-03-14 Richard Biener <rguenther@suse.de>
28218 PR tree-optimization/80030
28219 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28221 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28223 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28224 gcc_fallthrough() instead of __attribute__((fallthrough));
28226 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28228 * doc/gcc.texi: Remove "up" link to (DIR).
28229 * doc/gccint.texi: Ditto.
28231 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28233 * doc/install.texi (Specific) <avr>: Remove reference to
28236 2017-03-13 Jeff Law <law@redhat.com>
28238 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28239 attribute rather than comments.
28241 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28242 match_scratch operand is highest.
28244 2017-03-13 Martin Liska <mliska@suse.cz>
28246 PR middle-end/78339
28247 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28248 is a CHKP clone, use original declaration.
28250 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28252 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28253 (arc_conditional_register_usage): Use a different allocation order
28254 when optimizing for size.
28255 * common/config/arc/arc-common.c (arc_option_optimization_table):
28256 Section anchors default on when optimizing for size.
28258 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28260 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28262 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28264 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28265 * config/arc/arc.md (cpu_facility): Add cd variant.
28266 (*movqi_insn): Add code density variant.
28267 (*movhi_insn): Likewise.
28268 (*movqi_insn): Likewise.
28269 (*addsi3_mixed): Likewise.
28270 (subsi3_insn): Likewise.
28272 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28274 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28276 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28278 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28279 expressions with MINUS and UNARY ops.
28281 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28284 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28286 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28287 between vec_select and vector argument.
28288 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28289 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28290 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28291 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28292 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28293 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28295 2017-03-13 Richard Biener <rguenther@suse.de>
28298 * params.def (vect-max-peeling-for-alignment): Fix typo.
28300 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28302 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28303 issue that only occurred with binutils below 2.18.
28305 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28307 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28308 refer to binutils 2.11/2.12 minimum.
28310 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28312 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28313 ftp.kernel.org and simplify binutils requirement.
28315 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28317 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28319 (Optimize Options): Ditto. Also remove redundancy.
28321 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28323 PR translation/79848
28324 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28326 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28327 to G_ to avoid double translation.
28329 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28331 PR translation/79923
28332 * auto-profile.c (get_combined_location): Convert leading
28333 character of diagnostics to lower case and remove trailing period.
28334 (read_profile): Likewise for various diagnostics.
28335 * config/arm/arm.c (arm_option_override): Remove trailing period
28336 from various diagnostics.
28337 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28338 (msp430_expand_delay_cycles): Likewise.
28340 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28343 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28344 full command-line argument, rather than just "str".
28345 (aarch64_validate_march): Likewise.
28346 (aarch64_validate_mtune): Likewise.
28348 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28350 PR rtl-optimization/78911
28351 * lra-assigns.c (must_not_spill_p): New function.
28352 (spill_for): Use it.
28354 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28356 PR tree-optimization/79981
28357 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28358 ATOMIC_COMPARE_EXCHANGE ifn result.
28359 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28360 IFN_ATOMIC_COMPARE_EXCHANGE.
28362 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28365 * opts.c (parse_sanitizer_options): Add missing question mark to
28366 "did you mean" message.
28368 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28370 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28372 (VMULEUH_UNS): Likewise.
28373 (VMULOUB_UNS): Likewise.
28374 (VMULOUH_UNS): Likewise.
28375 * config/rs6000/rs6000.c (builtin_function_type): Remove
28376 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28378 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28381 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28382 x with result of extra_parsing_for_operand_code_0.
28383 (function_reader::extra_parsing_for_operand_code_0): Convert
28384 return type from void to rtx, returning x. When reading
28385 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28386 larger size containing struct block_symbol.
28388 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28390 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28391 -mfloat128-hardware without -m64.
28393 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28396 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28397 entries to the case statement that marks unsigned arguments to
28398 overloaded functions.
28400 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28402 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28403 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28405 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28408 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28409 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28411 2017-03-10 Martin Liska <mliska@suse.cz>
28415 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28416 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28419 2017-03-10 Olivier Hainque <hainque@adacore.com>
28421 * tree-switch-conversion (array_value_type): Start by resetting
28422 candidate type to it's main variant.
28424 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28426 PR rtl-optimization/79909
28427 * combine.c (try_combine): Use simplify_replace_rtx on individual
28428 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28429 of the whole CALL_INSN_FUNCTION_USAGE.
28431 PR tree-optimization/79972
28432 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28433 get_range_info on SSA_NAMEs. Formatting fixes.
28435 2017-03-10 Richard Biener <rguenther@suse.de>
28436 Jakub Jelinek <jakub@redhat.com>
28438 PR tree-optimization/77975
28439 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28440 edge to be constant.
28441 (get_val_for): For constant x return it. Formatting fix.
28442 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28443 would use the same bases as the current one.
28445 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28447 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28448 instead of vec_select for V1TImode.
28449 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28451 (VSX_LE_128): Add V1TI to this mode iterator.
28452 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28453 (*vsx_le_perm_store_<mode>): Likewise.
28454 (pre-reload splitter for VSX stores): Likewise.
28455 (post-reload splitter for VSX stores): Likewise.
28456 (*vsx_xxpermdi2_le_<mode>): Likewise.
28457 (*vsx_lxvd2x2_le_<mode>): Likewise.
28458 (*vsx_stxvd2x2_le_<mode>): Likewise.
28460 2017-03-09 Michael Eager <eager@eagercon.com>
28462 Correct failures with --enable-checking=yes,rtl.
28464 * config/microblaze/microblaze.c (microblaze_expand_shift):
28465 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28466 test for const0_rtx.
28467 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28468 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28470 2017-03-09 Richard Biener <rguenther@suse.de>
28472 PR tree-optimization/79977
28473 * graphite-scop-detection.c (scop_detection::merge_sese):
28474 Handle the case of extra exits to blocks dominating the entry.
28476 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28478 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28481 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28483 PR rtl-optimization/79949
28484 * lra-constraints.c (process_alt_operands): Check memory when
28485 trying to predict a cycle. Print about the overall increase.
28487 2017-03-09 Richard Biener <rguenther@suse.de>
28489 PR middle-end/79971
28490 * gimple-expr.c (useless_type_conversion_p): Preserve
28491 TYPE_SATURATING for fixed-point types.
28493 2017-03-09 Richard Biener <rguenther@suse.de>
28496 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28497 alignment of BLKmode params.
28499 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28502 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28503 (VALL_NO_V2Q): Likewise.
28505 * config/aarch64/aarch64-simd.md
28506 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28508 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28509 VALL_NO_V2Q mode iterator.
28510 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28512 2017-03-09 Martin Liska <mliska@suse.cz>
28514 PR tree-optimization/79631
28515 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28516 tree_int_cst_sign_bit just for INTEGER constants.
28518 2017-03-09 Martin Liska <mliska@suse.cz>
28522 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28525 2017-03-09 Marek Polacek <polacek@redhat.com>
28528 * tree.c (inchash::add_expr): Handle TREE_VEC.
28530 2017-03-09 Martin Liska <mliska@suse.cz>
28533 (chkp_narrow_size_and_offset): New function.
28534 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28535 (void chkp_parse_bit_field_ref): New function.
28536 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28537 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28539 2017-03-09 Martin Liska <mliska@suse.cz>
28542 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28543 (chkp_find_bounds_1): Remove gcc_unreachable.
28545 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28548 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28549 BUILT_IN_SYNC*, determine the access type from the size suffix and
28550 always build a MEM_REF with that type. Handle forgotten
28551 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28554 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28555 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28556 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28557 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28558 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28559 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28560 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28561 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28562 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28563 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28564 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28565 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28566 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28567 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28568 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28569 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28570 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28571 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28572 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28573 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28574 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28575 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28576 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28577 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28578 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28579 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28580 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28581 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28582 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28583 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28584 definitions outside of __OPTIMIZE__ guarded section.
28587 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28588 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28589 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28590 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28593 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28595 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28596 ("vfenez<mode>"): Add missing constraints.
28598 2017-03-08 Martin Sebor <msebor@redhat.com>
28601 * config/nds32/nds32.c (nds32_option_override):
28602 Fix misspelled diagnostic.
28604 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28607 * gimplify.c (gimplify_omp_for): Replace index var in outer
28608 taskloop statement with an artificial variable and add
28609 OMP_CLAUSE_PRIVATE clause for it.
28611 2017-03-08 Richard Biener <rguenther@suse.de>
28613 PR tree-optimization/79955
28614 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28615 for accesses that are completely outside of the variable.
28617 2017-03-08 Andrew Haley <aph@redhat.com>
28619 PR tree-optimization/79943
28620 * tree-ssa-loop-split.c (compute_new_first_bound): When
28621 calculating the new upper bound, (END-BEG) should be added, not
28624 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28626 * config/avr/avr.md (setmemhi): Make sure match_dup
28627 operand number comes before match_scratch.
28629 2017-03-08 Richard Biener <rguenther@suse.de>
28631 PR tree-optimization/79920
28632 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28633 with ncopies == 1 to ...
28634 (vect_transform_slp_perm_load): ... here. Properly compute
28635 all element loads by iterating VF times over the group. Do
28636 not handle ncopies (computed in a broken way) in
28637 vect_create_mask_and_perm.
28639 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28642 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28643 is a uniform vector, use uniform_vector_p return value instead of
28644 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28646 2017-03-07 Marek Polacek <polacek@redhat.com>
28648 PR middle-end/79809
28649 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28650 (alloca_call_type): Likewise.
28652 2017-03-07 Martin Liska <mliska@suse.cz>
28654 * gcov.c (process_args): Put comment to correct location.
28656 2017-03-07 Martin Liska <mliska@suse.cz>
28658 PR middle-end/68270
28659 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28660 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28661 (chkp_narrow_bounds_for_field): Likewise.
28662 (chkp_parse_array_and_component_ref): Pass one more argument to
28665 2017-03-07 Richard Biener <rguenther@suse.de>
28667 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28670 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28672 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28673 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28675 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28678 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28679 to end of description.
28680 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28682 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28684 PR rtl-optimization/79901
28685 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28687 (*avx512f_<code><mode>3<mask_name>): ... this.
28688 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28689 iterator instead of VI8_AVX2_AVX512BW.
28691 PR rtl-optimization/79901
28692 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28693 min/max expander, expand it using expand_vec_cond_expr.
28696 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28699 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28702 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28703 to void * for PCH reasons.
28704 * dwarf2out.c (output_loc_operands, output_die): Cast
28705 v.val_vec.array to unsigned char *.
28707 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28710 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28713 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28715 PR rtl-optimization/79571
28716 * lra-constraints.c (process_alt_operands): Calculate static
28717 reject and subtract it from overall when only addresses will be
28720 2017-03-06 Julia Koval <julia.koval@intel.com>
28723 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28724 incoming stack boundary to 128 for 64-bit targets.
28726 2017-03-06 Richard Biener <rguenther@suse.de>
28728 PR tree-optimization/79894
28729 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28730 to NULL after folding it.
28732 2017-03-06 Richard Biener <rguenther@suse.de>
28734 PR tree-optimization/79824
28735 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28736 check disabling peeling for gaps.
28738 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28740 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28741 attributes): Document gettimeofday.
28743 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28745 * config/s390/s390.c (s390_option_override_internal): Set
28746 PARAM_MIN_VECT_LOOP_BOUND
28748 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28750 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28751 * config/s390/s390.md: Likewise.
28753 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28756 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28757 (<avx2_avx512>_perm<mode>): Rename to ...
28758 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28760 (<avx512>_perm<mode>_mask): Rename to ...
28761 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28763 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28764 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28765 instead of VI8F_256_512.
28766 (avx512f_perm<mode>): New define_expand.
28767 (avx512f_perm<mode>_mask): Likewise.
28768 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28769 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28771 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28773 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28774 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28776 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28778 2017-03-06 Martin Liska <mliska@suse.cz>
28781 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28782 when having a SSA NAME w/o VAR_DECL assigned to it.
28784 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28786 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28787 msa_dpsub_<su>_d): Fix MODE for vec_select.
28789 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28791 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28793 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28795 2017-03-06 Richard Biener <rguenther@suse.de>
28797 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28798 * plugin.c (register_plugin_info): Likewise.
28799 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28801 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28803 * config/i386/sse.md (sse_storehps, sse_storelps,
28804 avx_<castmode><avxsizesuffix>_<castmode>,
28805 avx512f_<castmode><avxsizesuffix>_<castmode>,
28806 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28807 in condition that at least one operand is not a MEM.
28809 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28811 PR middle-end/79805
28812 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28813 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28815 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28816 gimple_call_nothrow_p flag based on whether original builtin can throw.
28817 If it can, emit following stmts on the fallthrough edge.
28818 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28819 don't create new bb if inserting just debug stmts on the edge, try to
28820 insert them on the fallthru bb or just reset debug stmts.
28822 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28825 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28826 restore recog_data (including the operand rtxes inside it) around
28827 the call to get_insn_template.
28829 2017-03-03 Martin Sebor <msebor@redhat.com>
28831 PR tree-optimization/79699
28832 * context.c (context::~context): Free MPFR caches to avoid
28833 a memory leak on program exit.
28835 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28837 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28838 Use wide_int::ulow () instead of .elt (0).
28840 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28842 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28843 (*pushxf): Limit oF constraint to 32bit targets and add oC
28844 constraint for 64bit targets.
28845 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28846 (*pushdf): Change rmF constraint to rmC.
28848 2017-03-03 Martin Liska <mliska@suse.cz>
28850 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28851 Remove unused variable.
28853 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28856 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28857 is a memory operand, increase num_memory.
28858 (ix86_expand_args_builtin): Likewise.
28860 2017-03-03 Jan Hubicka <jh@suse.cz>
28863 * ipa-devirt.c (maybe_record_node): Properly handle
28864 __cxa_pure_virtual visibility.
28866 2017-03-03 Martin Liska <mliska@suse.cz>
28868 PR tree-optimization/79803
28869 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28871 (pass_loop_prefetch::execute): Disabled optimization if an
28872 assumption about L1 cache size is not met.
28874 2017-03-03 Martin Liska <mliska@suse.cz>
28876 PR rtl-optimization/79574
28877 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28878 (hash_scan_set): Likewise.
28879 (dump_hash_table): Likewise.
28880 (hoist_code): Likewise.
28882 2017-03-03 Richard Biener <rguenther@suse.de>
28884 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28885 in place of uhigh ().
28886 (fixed_convert_from_real): Likewise.
28888 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28891 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28893 2017-03-03 Richard Biener <rguenther@suse.de>
28895 PR middle-end/79818
28896 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28897 TYPE_OVERFLOW_UNDEFINED check.
28899 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28901 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28903 (vector_ae_<mode>_p): Likewise.
28904 (vector_nez_<mode>_p): Likewise.
28905 (vector_ne_v2di_p): Likewise.
28906 (vector_ae_v2di_p): Likewise.
28907 (vector_ne_<mode>_p): Likewise.
28908 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28910 (vsx_tsqrt<mode>2_fe): Likewise.
28912 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
28915 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
28917 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28919 PR rtl-optimization/79780
28920 * cprop.c (one_cprop_pass): When second and further conditional trap
28921 in a single basic block is turned into an unconditional trap, turn it
28922 into a deleted note to avoid RTL verification failures.
28924 2017-03-02 Richard Biener <rguenther@suse.de>
28926 * fold-const.c (const_binop): Use ulow () instead of elt (0).
28928 2017-03-02 Richard Biener <rguenther@suse.de>
28930 PR tree-optimization/79345
28932 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
28933 param and abort the walk, returning -1 if it is hit.
28934 (walk_aliased_vdefs): Take a limit param and pass it on.
28935 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
28936 defaulting to 0 and return a signed int.
28937 * tree-ssa-uninit.c (struct check_defs_data): New struct.
28938 (check_defs): New helper.
28939 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
28940 about uninitialized memory.
28941 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
28942 bogus uninitialized warning.
28943 (fixed_convert_from_real): Likewise.
28945 2017-03-02 Bin Cheng <bin.cheng@arm.com>
28947 PR tree-optimization/66768
28948 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
28949 iv_use if base object can't be determined.
28951 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28953 PR tree-optimization/79345
28954 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
28955 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
28956 (get_pattern_stats): Initialize it.
28957 * genemit.c (gen_expand): Verify match_scratch numbers come after
28958 match_operand/match_dup numbers.
28959 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
28960 match_scratch numbers.
28961 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
28963 * config/s390/s390.md (trunctdsd2): Likewise.
28965 2017-03-02 Richard Biener <rguenther@suse.de>
28967 * wide-int.h (wide_int_storage::operator=): Implement in terms
28970 2017-03-02 Richard Biener <rguenther@suse.de>
28972 PR tree-optimization/79777
28973 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
28974 the to insert expression to sth existing.
28976 2017-03-01 Martin Sebor <msebor@redhat.com>
28978 PR middle-end/79692
28979 * gimple-ssa-sprintf.c
28980 (directive::known_width_and_precision): New function.
28981 (format_integer): Use it.
28982 (get_mpfr_format_length): Consider the full range of precision
28983 when computing %g output with the # flag. Set the likely byte
28984 count to 3 rather than 1 when precision is indeterminate.
28985 (format_floating): Correct the lower bound of precision.
28987 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28989 * doc/invoke.texi: Document default code model for 64-bit Linux.
28991 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28994 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
28995 udiv rather than div since input pattern is unsigned.
28997 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
28999 * config/i386/i386.c (print_reg): Warn for values of
29000 unsupported size in integer register.
29002 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29005 * config/rs6000/predicates.md (current_file_function_operand): Do
29006 not allow self calls to be local if the function is replaceable.
29008 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29011 * config/rs6000/altivec.h (vec_ctz and others): Change the
29012 preprocessor macro that controls conditional compilation from
29013 _ARCH_PWR9 to __POWER9_VECTOR__.
29014 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29015 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29016 control (instead of _ARCH_PWR9 control) so that template
29017 definition uses power9-specific function.
29018 (vec_any_eq): Likewise.
29019 (vec_all_ne): Change macro definition to use a power9-specific
29020 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29021 _ARCH_PWR9 control).
29022 (vec_any_eq) Likewise.
29023 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29024 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29025 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29026 support for xvcmpnedp instruction.
29027 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29028 macro expansion so that Power9 implementation of vec_all_ne does
29029 not use the AltiVec predicate framework.
29030 (VCMPNEH_P): Likewise.
29031 (VCMPNEW_P): Likewise.
29032 (VCMPNED_P): Likewise.
29033 (VCMPNEFP_P): Likewise.
29034 (VCMPNEDP_P): Likewise.
29035 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29036 implementation of vec_any_eq to not use AltiVec predicate
29038 (VCMPAEH_P): Likewise.
29039 (VCMPAEW_P): Likewise.
29040 (VCMPAED_P): Likewise.
29041 (VCMPAEFP_P): Likewise.
29042 (VCMPAEDP_P): Likewise.
29043 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29044 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29045 not use the AltiVec predicate framework.
29046 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29047 of vec_any_eq to not use AltiVec predicate framework.
29048 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29049 support for predefined __POWER9_VECTOR__ macro to indicate that
29050 Power9 instruction selection is enabled.
29051 (altivec_overloaded_builtins): Remove extraneous
29052 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29053 function argument types RS6000_BTI_bool_V16QI and
29054 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29055 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29056 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29057 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29058 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29059 Power9 for implementations of vec_cmpne. Change the signature for
29060 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29061 (representing vec_all_ne) to remove the previously described first
29062 argument of type RS6000_BTI_INTSI, as this was an artifact of
29063 reliance on the AltiVec predicate framework, which is no longer
29064 used in the implementation of these functions. Add
29065 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29066 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29067 since, unlike the AltiVec predicate framework implementation, we
29068 do not share function descriptors between vec_alle and vec_anyeq.
29069 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29070 set of modes that receive special treatment even when
29071 TARGET_P9_VECTOR is true. The special treatment emits code that
29072 does not depend on Power9 instructions.
29073 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29074 define_expand to not rely on AltiVec predicate framework.
29075 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29077 (vector_ne_v2di_p): Change this define_expand to not rely on
29078 AltiVec predicate framework.
29079 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29081 (vector_ne_<mode>_p): Change this define_expand to not rely on
29082 AltiVec predicate framework.
29083 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29085 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29086 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29087 define_insn pattern.
29088 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29089 define_insn pattern because the xvcmpne<VSs>. instruction is not
29091 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29092 instruction is not supported.
29094 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29096 * config/nvptx/nvptx.c: Include intl.h.
29098 2017-03-01 Martin Jambor <mjambor@suse.cz>
29101 * ipa-prop.h (ipa_bits): Removed field known.
29102 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
29103 to pointers. Adjusted their comments to warn about their sharing.
29104 (ipcp_transformation_summary): Change bits to a vector of pointers.
29105 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
29106 (ipa_get_ipa_bits_for_value): Declare.
29107 * tree-vrp.h (value_range): Mark as GTY((for_user)).
29108 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
29109 (ipa_bits_hash_table): Likewise.
29110 (ipa_vr_ggc_hash_traits): Likewise.
29111 (ipa_vr_hash_table): Likewise.
29112 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
29113 being pointers and vr_known being removed.
29114 (ipa_set_jf_unknown): Likewise.
29115 (ipa_get_ipa_bits_for_value): New function.
29116 (ipa_set_jfunc_bits): Likewise.
29117 (ipa_get_value_range): New overloaded functions.
29118 (ipa_set_jfunc_vr): Likewise.
29119 (ipa_compute_jump_functions_for_edge): Use the above functions to
29120 construct bits and vr parts of jump functions.
29121 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
29122 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29124 (ipcp_grow_transformations_if_necessary): Also allocate
29125 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29127 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
29128 them. Fix too long lines.
29129 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
29130 vr_known being removed.
29131 (ipa_read_jump_function): Use new setter functions to construct bits
29132 and vr parts of jump functions or set them to NULL.
29133 (write_ipcp_transformation_info): Adjust for bits being pointers.
29134 (read_ipcp_transformation_info): Likewise.
29135 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
29137 Include gt-ipa-prop.h.
29138 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
29140 (ipcp_store_bits_results): Likewise.
29141 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
29142 Do not write to existing jump functions but use a temporary instead.
29144 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29147 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
29148 attempt to use its first operand as BIT_FIELD_REF base.
29150 2017-03-01 Richard Biener <rguenther@suse.de>
29152 PR middle-end/79721
29153 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
29154 interpolating formula in wrapping arithmetic.
29155 (chrec_apply): Convert chrec_evaluate return value to wanted type.
29157 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29159 PR tree-optimization/79734
29160 * tree-vect-generic.c (expand_vector_condition): Optimize
29161 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29162 Handle VEC_COND_EXPR where comparison has different inner width from
29163 type's inner width.
29165 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29167 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29168 markup, and similar issues. Remove @opindex entries for things
29169 that aren't options. Add missing -mmpy-option entries.
29171 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29173 PR tree-optimization/79737
29174 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29175 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29176 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29177 instead of byte_size. Formatting fix.
29178 (shift_bytes_in_array_right): Formatting fix.
29180 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29183 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29184 condition on optimize for the leaf function test.
29186 2017-02-28 Martin Liska <mliska@suse.cz>
29189 * read-rtl-function.c (function_reader::handle_unknown_directive):
29190 Bail out when one uses -flto.
29192 2017-02-28 Martin Liska <mliska@suse.cz>
29194 * common.opt: Replace space with tabular for options of <number>
29196 * config/i386/i386.opt: Show <number> value for
29197 -mlarge-data-threshold.
29198 * opts.c (print_filtered_help): Do not display number in hexadecimal
29201 2017-02-28 Martin Liska <mliska@suse.cz>
29203 * common.opt: Fix --help=option -Q for options which are of
29206 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29208 * config/i386/i386.c (print_reg): Error out for values
29209 of 8-bit size in invalid integer register.
29211 2017-02-28 Martin Sebor <msebor@redhat.com>
29213 PR tree-optimization/79691
29214 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29216 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29219 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29220 gcc_unreachable with output_operand_lossage.
29222 2017-02-28 Richard Biener <rguenther@suse.de>
29224 PR tree-optimization/79740
29225 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29227 (visit_nary_op): Insert the nary into the hashtable if we
29228 pattern-matched sth.
29229 * tree-ssa-pre.c (eliminate_insert): Robustify.
29231 2017-02-28 Richard Biener <rguenther@suse.de>
29233 PR middle-end/79731
29234 * fold-const.c (decode_field_reference): Reject out-of-bound
29237 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29239 * config/i386/i386.c: Include intl.h.
29240 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29241 instead of just cond ? "..." : "...".
29242 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29243 * coverage.c (read_counts_file): Likewise.
29244 * omp-offload.c: Include intl.h.
29245 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29246 of just cond ? "..." : "...".
29247 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29248 of just cond ? "..." : "...".
29250 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29253 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29255 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29256 'tune for' CPU name.
29257 * config/arm/arm-cpu-data.h: Regenerated.
29259 2017-02-28 Richard Biener <rguenther@suse.de>
29261 PR tree-optimization/79732
29262 * tree-inline.c (expand_call_inline): Do not shadow var.
29264 2017-02-28 Richard Biener <rguenther@suse.de>
29266 PR tree-optimization/79723
29267 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29268 address-space properly.
29270 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29272 * doc/optinfo.texi (Optimization groups): Fix option used for
29274 * doc/invoke.texi (-fopt-info): Document "omp".
29275 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29276 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29277 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29278 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29279 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29281 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29283 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29284 OPTGROUP_OMP with "omp".
29286 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29289 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29290 for arithmetic shift of unsigned V2DI.
29292 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29294 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29295 arc/linux.h headers.
29296 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29297 (LINK_SPEC): Likewise.
29298 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29299 (EXTRA_SPECS): Likewise.
29300 (STARTFILE_SPEC): Likewise.
29301 (ENDFILE_SPEC): Likewise.
29302 (LIB_SPEC): Likewise.
29303 (TARGET_SDATA_DEFAULT): Likewise.
29304 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29305 (MULTILIB_DEFAULTS): Likewise.
29306 (DWARF2_UNWIND_INFO): Likewise.
29307 * config/arc/big.h: New file.
29308 * config/arc/elf.h: Likewise.
29309 * config/arc/linux.h: Likewise.
29310 * config/arc/t-uClibc: Remove.
29312 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29314 PR tree-optimization/77536
29315 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29316 (tree_transform_and_unroll_loop): Use above function to compute the
29317 estimated niter of unrolled loop and use it when scaling profile.
29318 Also use count info rather than frequency if it's non-zero.
29319 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29320 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29321 (vect_transform_loop): Call above function.
29323 2017-02-27 Richard Biener <rguenther@suse.de>
29325 PR tree-optimization/45397
29326 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29327 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29328 (visit_nary_op): Add pattern matching for CSEing sign-changed
29329 or truncated operations with wider ones.
29331 2017-02-27 Richard Biener <rguenther@suse.de>
29333 PR tree-optimization/79690
29334 * tree-vect-stmts.c (vectorizable_store): Use vector type
29335 built from the DR with address-space.
29337 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29339 * doc/invoke.texi (Optimize Options): Refine the description
29340 of asan-use-after-return.
29342 2017-02-25 Alan Modra <amodra@gmail.com>
29344 PR rtl-optimization/79584
29345 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29346 base, not ad->base_term, the reg within base. Remove assertion
29347 that ad->base == ad->base_term. Replace gen_int_mode using
29348 bogus mode with const0_rtx.
29350 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29352 PR middle-end/79396
29353 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29354 FMA_EXPR like tcc_binary or tcc_unary.
29356 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29359 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29361 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29362 (output_loc_operands): Handle DW_OP_call_ref and
29363 DW_OP_GNU_variable_value.
29364 (struct variable_value_struct): New type.
29365 (struct variable_value_hasher): Likewise.
29366 (variable_value_hash): New variable.
29367 (string_types): Remove.
29368 (copy_loc_descr): New function.
29369 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29370 (prepend_loc_descr_to_each): New function.
29371 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29372 instead of add_loc_descr_to_each if the first argument is single
29373 location list and the second has multiple.
29374 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29375 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29376 when looking for variable value which doesn't have other location info.
29377 (loc_list_from_tree): Formatting fix.
29378 (gen_array_type_die): Simplify DW_AT_string_length handling.
29379 (adjust_string_types): Remove.
29380 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29381 string_types. Call resolve_variable_values.
29382 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29383 (resolve_addr_in_expr): Likewise. Add A argument.
29384 (copy_deref_exprloc): Remove deref argument. Adjust for the
29385 original expression being DW_OP_GNU_variable_value with optionally
29386 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29387 optionally after it.
29388 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29389 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29390 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29391 (variable_value_hasher::hash, variable_value_hasher::equal): New
29393 (resolve_variable_value_in_expr, resolve_variable_value,
29394 resolve_variable_values, note_variable_value_in_expr,
29395 note_variable_value): New functions.
29396 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29399 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29402 * opts.h (handle_generated_option): Add GENERATED_P argument.
29403 * opts-common.c (handle_option): Adjust function comment.
29404 (handle_generated_option): Add GENERATED_P argument, pass it to
29406 (control_warning_option): Pass false to handle_generated_option
29408 * opts.c (maybe_default_option): Pass true to handle_generated_option
29410 * optc-gen.awk: Likewise.
29412 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29414 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29415 a REG, look at the REG it is a SUBREG of.
29416 (splitter for cmpeqsi_t): Ditto.
29418 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29420 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29421 the special USEs with the pattern of the insn, not the insn itself.
29423 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29426 * doc/invoke.texi: Document -mload-store-pairs.
29428 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29429 Sandra Loosemore <sandra@codesourcery.com>
29431 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29432 argument isn't a CONST_INT.
29433 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29434 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29435 (nios2_validate_compare): Bypass alternate compare logic if *op2
29436 is not a CONST_INT.
29437 (ldstwm_operation_p): Return false if first_base is not a REG or
29438 if first_offset is not a CONST_INT.
29440 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29442 * config/cris/cris.md: Use correct operand in a define_peephole2.
29444 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29446 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29448 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29450 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29451 this_insn if it is an INSN or JUMP_INSN.
29452 (force_offsettable): Look at base, not at addr.
29453 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29454 on things that aren't necessarily CONST_INTs.
29456 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29458 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29459 -mfpmath=sse is the default also for x86-32 targets with SSE2
29460 instruction set when @option{-ffast-math} is enabled
29462 2017-02-24 Jeff Law <law@redhat.com>
29464 PR rtl-optimizatoin/79286
29465 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29468 2017-02-24 Richard Biener <rguenther@suse.de>
29470 PR tree-optimization/79389
29471 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29474 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29476 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29477 function comment to reflect reality.
29478 (loop_exits_before_overflow): Fix typo in function description.
29480 2017-02-24 Richard Biener <rguenther@suse.de>
29482 PR tree-optimization/79389
29483 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29484 properly that a threading opportunity exists. Detect conditional
29485 copy/constant propagation opportunities.
29487 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29489 * config/visium/visium.md (type): Add trap.
29490 (b): New mode attribute.
29491 (*btst): Rename into...
29492 (*btst<mode>): ...this and adjust.
29493 (*cbranchsi4_btst_insn): Rename into...
29494 (*cbranch<mode>4_btst_insn): ...this and adjust.
29495 (trap): New define_insn.
29497 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29499 PR tree-optimization/79389
29500 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29501 (noce_reversed_cond_code): New function.
29502 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29503 reversed_comparison_code. Formatting fix.
29504 (noce_try_store_flag): Test rev_cond != NULL in addition to
29505 reversed_comparison_code.
29506 (noce_try_store_flag_constants): Likewise.
29507 (noce_try_store_flag_mask): Likewise.
29508 (noce_try_addcc): Use rev_cond if non-NULL instead of
29509 reversed_comparison_code.
29510 (noce_try_cmove_arith): Likewise. Formatting fixes.
29511 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29512 (noce_find_if_block): Initialize rev_cond.
29513 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29514 instead of false as last argument never attempt to reverse it
29517 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29519 PR tree-optimization/79663
29520 * tree-predcom.c (combine_chains): Process refs in reverse order
29521 only for ZERO length chains, and add explaining comment.
29523 2017-02-23 Jeff Law <law@redhat.com>
29525 PR tree-optimization/79578
29526 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29527 in call to operand_equal_p.
29529 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29532 * config/i386/cpuid.h: Fix another undefined behavior.
29534 2017-02-23 Richard Biener <rguenther@suse.de>
29536 PR tree-optimization/79683
29537 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29538 vector types for data-refs.
29540 2017-02-23 Martin Liska <mliska@suse.cz>
29542 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29544 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29546 PR middle-end/79665
29547 * internal-fn.c (get_range_pos_neg): Moved to ...
29548 * tree.c (get_range_pos_neg): ... here. No longer static.
29549 * tree.h (get_range_pos_neg): New prototype.
29550 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29551 are known to be in between 0 and signed maximum inclusive, try to
29552 expand both unsigned and signed divmod and use the cheaper one from
29555 2017-02-22 Jeff Law <law@redhat.com>
29557 PR tree-optimization/79578
29558 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29559 to compare base operands.
29561 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29564 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29565 gpc_reg_operand instead of fpr_reg_operand.
29567 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29569 * config/mips/mips.c (mips_return_in_memory): Force FP
29570 vector types to be returned in memory for o32 ABI.
29572 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29574 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29575 instead of DW_TAG_member for static data member declarations and don't
29576 set no_linkage_name for static inline data members.
29577 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29580 2017-02-22 Martin Liska <mliska@suse.cz>
29582 * doc/invoke.texi: Replace inequality signs with square brackets
29585 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29588 * lra-constraints.c (simplify_operand_subreg): Handle
29589 WORD_REGISTER_OPERATIONS targets.
29591 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29594 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29595 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29596 elimination by swapping fld*.
29598 2017-02-22 Richard Biener <rguenther@suse.de>
29600 PR tree-optimization/79673
29601 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
29602 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
29603 irrelevant address-space qualifiers and avoiding a
29604 ADDR_SPACE_CONVERT_EXPR from fold_convert.
29606 2017-02-22 Richard Biener <rguenther@suse.de>
29608 PR tree-optimization/79666
29609 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
29610 to not symbolically negate if that may introduce undefined
29613 2017-02-22 Martin Liska <mliska@suse.cz>
29616 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29617 * data-streamer-out.c (streamer_write_gcov_count_stream):
29619 * value-prof.c (stream_out_histogram_value): Make assert more
29620 precise based on type of counter.
29622 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29625 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29626 Use nonimmediate_operand instead of memory_operand for operand 1.
29627 (float-extend standard_x87sse_constant_load splitter): Ditto.
29629 2017-02-21 Jeff Law <law@redhat.com>
29631 PR tree-optimization/79621
29632 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29633 blocks with edges to themselves.
29635 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29638 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29639 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29640 Use gimple_call_builtin_p.
29643 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29644 on temporarily removed DEBUG_INSNs.
29646 PR tree-optimization/79649
29647 * tree-loop-distribution.c (classify_partition): Give up on
29648 non-generic address space loads/stores.
29650 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29652 * doc/loop.texi (Loop manipulation): Remove nonexistent
29653 tree_ssa_loop_version from the documentation.
29654 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29656 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29659 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29660 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29661 * config/rs6000/rs6000.c: Include except.h.
29662 (rs6000_expand_split_stack_prologue): Call
29663 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29665 2017-02-21 Martin Jambor <mjambor@suse.cz>
29668 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29669 have been analyzed.
29671 2017-02-21 Martin Jambor <mjambor@suse.cz>
29673 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29674 for backward compatibility only.
29675 * doc/invoke.texi (Option Summary): Remove all references to
29676 -fipa-cp-alignment.
29678 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29682 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29684 * lra-constraints.c (curr_insn_transform): Handle
29685 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29687 2017-02-21 Martin Liska <mliska@suse.cz>
29689 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29691 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29693 PR translation/79638
29694 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29696 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29699 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29700 (arm_function_ok_for_sibcall): Return false for an indirect call by
29701 descriptor if all the argument registers are used.
29702 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29703 alignment of the function.
29705 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29707 PR tree-optimization/61441
29708 * simplify-rtx.c (simplify_const_unary_operation): For
29709 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29710 the sNaN unmodified.
29712 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29714 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29715 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29716 instead of SYSTEM_HEADER_DIR.
29718 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29719 Martin Liška <mliska@suse.cz>
29721 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29722 Fix typos and grammar, use active voice, and clarify.
29724 2017-02-20 Marek Polacek <polacek@redhat.com>
29726 PR middle-end/79537
29727 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29730 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29732 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29735 * config/i386/i386.c (ix86_expand_builtin): Handle
29736 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29737 ix86_builtins_isa[fcode].isa as a requirement of those
29738 flags and any other flag in the bitmask.
29739 (ix86_init_mmx_sse_builtins): Use 0 instead of
29740 ~OPTION_MASK_ISA_64BIT as mask.
29741 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29742 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29743 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29744 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29746 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29749 * lra-constraints.c (split_reg): Check requested split mode
29750 is supported by the register.
29752 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29754 * lra-constraints.c (simplify_operand_subreg): Remove early
29757 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29760 * lra-constraints.c (curr_insn_transform): Tighten condition
29761 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29763 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29766 * lra-constraints.c (curr_insn_transform): Handle
29767 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29769 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29772 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29774 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29776 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29779 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29782 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29784 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29787 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29789 * final.c (last_columnnum, override_columnnum): New variables.
29790 (final_start_function): Set last_columnnum, pass it to begin_prologue
29791 hook and pass 0 to dwarf2out_begin_prologue.
29792 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29793 to source_line debug hook.
29794 (notice_source_line): Compute last_columnnum and for debug_column_info
29795 return true on column changes.
29796 * debug.h (struct gcc_debug_hooks): Add column argument to
29797 source_line and begin_prologue hooks.
29798 (debug_nothing_int_charstar_int_bool): Remove prototype.
29799 (debug_nothing_int_int_charstar,
29800 debug_nothing_int_int_charstar_int_bool): New prototypes.
29801 (dwarf2out_begin_prologue): Add column argument.
29802 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29803 begin_prologue hooks.
29804 (debug_nothing_int_charstar_int_bool): Remove.
29805 (debug_nothing_int_int_charstar,
29806 debug_nothing_int_int_charstar_int_bool): New functions.
29807 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29808 through to dwarf2out_source_line.
29809 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29810 (dwarf2out_source_line): Add column argument, emit it if requested.
29811 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29813 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29814 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29815 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29816 through to dwarf2out_begin_prologue.
29817 (vmsdbgout_source_line): Add column argument, pass it through to
29818 dwarf2out_source_line.
29819 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29820 dbxout_source_line caller.
29821 (dbxout_source_line): Add column argument.
29823 * common.opt (gno-column-info, gcolumn-info): New options.
29824 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29825 (check_die): Also test for multiple DW_AT_decl_column attributes.
29826 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29827 DW_AT_decl_column if requested.
29828 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29830 (gen_variable_die): Likewise.
29831 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29832 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29835 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29836 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29837 (ix86_handle_option): Handle OPT_m3dnowa.
29838 * doc/invoke.texi (-m3dnowa): Document.
29839 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29840 -m3dnowa instead of -m3dnow -march=athlon.
29843 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29844 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29846 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29849 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29850 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29851 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29852 generator for vsx_xxpermdi_<mode>_be.
29853 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29854 force big-endian semantics.
29855 (vsx_xxpermdi_<mode>_be): New define_expand with same
29856 implementation as previous version of vsx_xxpermdi_<mode>.
29858 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29860 PR tree-optimization/79327
29861 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29862 variable, its initialization and use.
29864 2017-02-17 Julia Koval <julia.koval@intel.com>
29866 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29867 (OPTION_MASK_ISA_PKU_UNSET): New.
29868 (ix86_handle_option): Handle -mrdpid.
29869 * config/i386/cpuid.h (bit_RDPID): New.
29870 * config/i386/driver-i386.c (host_detect_local_cpu):
29871 Detect RDPID feature.
29872 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29873 * config/i386/i386-c.c (ix86_target_macros_internal):
29875 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29876 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29877 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29878 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29879 * config/i386/i386.md (define_insn "rdpid"): New.
29880 * config/i386/i386.opt Add -mrdpid.
29881 * config/i386/immintrin.h (_rdpid_u32): New.
29883 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29885 PR rtl-optimization/79541
29886 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29887 instead of transforming it into USE.
29889 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29891 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29892 If HONOR_SNANS (SFmode) force the input to a register.
29893 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29894 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29895 an frsp or similar insn.
29897 2017-02-17 Martin Liska <mliska@suse.cz>
29899 PR rtl-optimization/79577
29900 * params.def (selsched-max-sched-times): Increase minimum to 1.
29902 2017-02-17 Martin Liska <mliska@suse.cz>
29904 PR rtl-optimization/79574
29905 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29907 2017-02-17 Martin Liska <mliska@suse.cz>
29909 PR tree-optimization/79529
29910 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
29911 ssa_defined_default_def_p to handle cases which are implicitly
29913 * tree-ssa.c (ssa_defined_default_def_p): New function.
29914 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
29915 which are implicitly defined.
29916 * tree-ssa.h (ssa_defined_default_def_p): Declare.
29918 2017-02-17 Richard Biener <rguenther@suse.de>
29920 PR middle-end/79576
29921 * params.def (max-ssa-name-query-depth): Limit to 10.
29923 2017-02-17 Richard Biener <rguenther@suse.de>
29925 PR tree-optimization/79552
29926 * tree-ssa-structalias.c (visit_loadstore): Properly verify
29929 2017-02-17 Richard Biener <rguenther@suse.de>
29932 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
29934 2017-02-17 Marek Polacek <polacek@redhat.com>
29936 PR middle-end/79536
29937 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
29938 (fold_negate_expr): New wrapper.
29940 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
29942 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
29943 Correct terminology and de-emphasize pre-standard behavior.
29945 2017-02-16 Alan Modra <amodra@gmail.com>
29947 PR rtl-optimization/79286
29948 * ira.c (def_dominates_uses): New function.
29949 (update_equiv_regs): Don't create an equivalence for insns that
29950 may trap where the register def does not dominate the use.
29952 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
29954 PR rtl-optimization/78127
29955 * lra.c (lra): Call lra_eliminate before finish the loop after
29958 2017-02-16 Richard Biener <rguenther@suse.de>
29960 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
29962 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
29963 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
29964 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
29965 (isl_val_int_from_wi): New function.
29966 (extract_affine_gmp): Rename to ...
29967 (extract_affine_wi): ... this, take a widest_int.
29968 (extract_affine_int): Just wrap extract_affine_wi.
29969 (add_param_constraints): Use isl_val_int_from_wi.
29970 (add_loop_constraints): Likewise, and extract_affine_wi.
29972 2017-02-15 Jeff Law <law@redhat.com>
29974 PR middle-end/79521
29975 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
29976 ira_init_register_move_cost_if_necessary.
29978 2017-02-15 Martin Sebor <msebor@redhat.com>
29980 PR middle-end/32003
29981 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
29982 removed in a prior commit.
29984 2017-02-15 Bin Cheng <bin.cheng@arm.com>
29986 PR tree-optimization/79347
29987 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
29988 counters during peeling.
29990 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
29992 * Makefile.in (site.exp): Remove "set ISLVER".
29994 2017-02-15 Jakub Jelinek <jakub@redhat.com>
29997 * real.c (real_from_integer): Call real_convert even for decimal.
29999 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30002 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30004 2017-02-14 Andrew Pinski <apinski@cavium.com>
30006 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30007 cores and change the partno/implementer to be correct.
30008 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30009 the 'B" as the implementer.
30010 * config/aarch64/aarch64-tune.md: Regenerate.
30012 2017-02-14 Carl Love <cel@us.ibm.com>
30014 * config/rs6000/rs6000.c: Add case statement entry to make the
30015 xvcvuxdsp built-in argument unsigned.
30016 * config/rs6000/vsx.md: Fix the source and return operand types so they
30017 match the instruction definitions from the ISA document. Fix typo
30018 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30021 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30024 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30025 member early_clobber_alts.
30026 * lra-lives.c (reg_early_clobber_p): New.
30027 (process_bb_lives): Use it.
30028 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30029 (debug_operand_data): Initialize early_clobber_alts.
30030 (setup_operand_alternative): Set up early_clobber_alts.
30031 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30032 alternatives to new_insn_reg.
30033 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30035 (lra_update_insn_regno_info): Pass the new arg.
30037 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30039 PR middle-end/79505
30040 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30041 (new_oacc_loop_raw): Don't clear already cleared fields.
30044 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30045 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30046 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30048 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30051 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30053 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30056 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30057 the extra instruction to the right place to store 128-bit constant
30060 2017-02-14 Martin Sebor <msebor@redhat.com>
30062 PR middle-end/79448
30063 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30064 warning for strings of unknown length.
30066 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30068 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30070 2017-02-14 Jeff Law <law@redhat.com>
30073 * ira-costs.c (scan_one_insn): Initialize register move costs
30074 for pseudos seen in USE/CLOBBER insns.
30076 PR tree-optimization/79095
30077 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30078 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30079 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30080 if the operands are known to be not equal, then the resulting range
30082 (intersect_ranges): If the new range is ~[0,0] and the old range is
30083 wide, then prefer ~[0,0].
30084 * tree-vrp.c (overflow_comparison_p_1): New function.
30085 (overflow_comparison_p): New function.
30086 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30087 if NAME is used in an overflow test.
30088 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30089 overflow check that can be expressed as an equality test, then adjust
30090 ops to be that equality test.
30092 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30094 * config/s390/s390-builtin-types.def: Remove flags argument.
30095 * config/s390/s390.c (s390_init_builtins): Likewise.
30097 2017-02-14 Martin Liska <mliska@suse.cz>
30099 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
30100 vector. Fix trailing white spaces.
30102 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
30104 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
30107 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30109 PR rtl-optimization/68664
30110 * config/arm/arm.c (arm_sched_can_speculate_insn):
30111 New function. Declare prototype.
30112 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30114 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30116 PR rtl-optimization/68664
30117 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
30119 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30121 2017-02-14 Amit Pawar <amit.pawar@amd.com>
30123 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
30124 max skip bytes for function, loop and jump.
30126 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30128 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
30129 ABS_EXPR for gimple dump.
30131 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30134 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
30136 PR tree-optimization/79408
30137 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
30138 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
30139 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
30140 also if rhs1 is INTEGER_CST.
30142 2017-02-14 Richard Biener <rguenther@suse.de>
30144 PR middle-end/79432
30145 * tree-into-ssa.c (insert_phi_nodes): When the function can
30146 have abnormal edges rewrite SSA names with broken use-def
30147 dominance out of SSA and register them for PHI insertion.
30149 2017-02-13 Martin Sebor <msebor@redhat.com>
30151 PR middle-end/79496
30152 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
30153 clearing info.nowrite flag when snprintf size argument is a range.
30155 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30157 * cprop.c (cprop_jump): Add missing space in string literal.
30158 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
30159 (get_constraint_for_component_ref): Likewise.
30160 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
30161 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30162 * lra-constraints.c (process_alt_operands): Likewise.
30163 * ipa-inline.c (inline_small_functions): Likewise.
30164 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30165 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30166 * trans-mem.c (diagnose_tm_1_op): Likewise.
30167 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30168 (grid_parallel_clauses_gridifiable): Likewise.
30170 * config/nvptx/mkoffload.c (process): Add space in between
30173 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30174 "MOD4_SSE_REGS" and "ALL_REGS".
30176 * spellcheck.c (test_data): Add , in between "foo" and "food".
30178 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30181 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30182 boundary crossing check and subsequent code generation agree.
30184 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30186 * config/aarch64/aarch64.c (has_memory_op): Delete.
30187 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30190 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30192 PR rtl-optimization/79388
30193 PR rtl-optimization/79450
30194 * combine.c (distribute_notes): When removing TEM_INSN for which
30195 corresponding dest has last value recorded, invalidate that last
30198 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30200 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30201 of explicit '@'. Add missing assembly comment marker on branch costs
30204 2017-02-13 Nathan Sidwell <nathan@acm.org>
30206 * gengtype-lex.l (<in_struct>): Add '/'.
30208 2017-02-13 Martin Liska <mliska@suse.cz>
30211 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30213 2017-02-13 Richard Biener <rguenther@suse.de>
30215 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30217 * configure: Re-generate.
30218 * config.in: Likewise.
30219 * graphite-dependences.c: Simplify as if
30220 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30221 * graphite-isl-ast-to-gimple.c: Likewise.
30222 * graphite-optimize-isl.c: Likewise.
30223 * graphite-poly.c: Likewise.
30224 * graphite-sese-to-poly.c: Likewise.
30225 * graphite.h: Likewise.
30226 * toplev.c: Include isl/version.h and use isl_version () for
30227 printing the ISL version.
30228 * doc/install.texi: Update ISL requirement.
30230 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30232 * doc/standards.texi (Standards): Update reference to
30235 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30237 * doc/extend.texi (Named Address Spaces): sourceware.org now
30239 * doc/install.texi (Binaries): Ditto.
30242 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30244 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30245 standard terminology "stringize"/"stringizing" throughout.
30246 * doc/cppinternals.texi: Likewise.
30248 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30250 * doc/extend.texi: Fix some spelling mistakes and typos.
30251 * doc/invoke.texi: Likewise.
30253 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30256 * params.def (inline-min-speedup) Change from 10 to 8.
30258 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30260 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30263 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30266 * ipa-inline-analysis.c (get_minimal_bb): New function.
30267 (record_modified): Use it.
30268 (remap_edge_change_prob): Handle also ancestor functions.
30270 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30272 * doc/contrib.texi (Contributors): Remove broken link into
30273 the Mauve CVS repository.
30275 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30277 PR middle-end/79454
30278 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30279 result computation whenever lhs doesn't have vector mode, not
30280 just when it has BLKmode.
30282 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30284 * doc/makefile.texi (profiledbootstrap): Refer to the
30285 installation instructions only in textual form.
30287 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30290 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30292 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30294 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30295 (Specific): Update mingw-w64 reference.
30297 (Specific): Remove broken link to Renesas RX processor.
30299 2017-02-10 Richard Biener <rguenther@suse.de>
30301 * toplev.c (process_options): Do not mention obsolete graphite
30302 options when printing sorry message about missing graphite support.
30303 Mention -floop-nest-optimize.
30305 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30307 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30308 (vtst_p16): Likewise.
30309 (vtstq_p8): Likewise.
30310 (vtstq_p16): Likewise.
30312 (vtstq_p64): Likewise.
30313 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30314 (vset_lane_p64): New.
30315 (vsetq_lane_p64): New.
30317 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30319 PR tree-optimization/79411
30320 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30321 stmt operands are SSA_NAMEs used in abnormal phis.
30322 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30325 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30328 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30331 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30333 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30335 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30337 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30341 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30342 not arbitrary TREE_CONSTANT.
30345 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30346 "omp declare target link" attribute unless is_global_var.
30347 * omp-offload.c (find_link_var_op): Likewise.
30349 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30350 Chung-Lin Tang <cltang@codesourcery.com>
30352 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30354 (gimplify_adjust_omp_clauses): Don't delete TILE.
30355 (gimplify_omp_for): Deal with TILE.
30356 * internal-fn.c (expand_GOACC_TILE): New function.
30357 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30359 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30360 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30362 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30363 avoid DIV for outermost collapse var.
30364 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30365 Remove out of date comments, fix whitespace.
30366 * omp-general.c (omp_extract_for_data): Deal with tiling.
30367 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30368 adjust OLF_DIM_BASE value.
30369 (struct omp_for_data): Add tiling field.
30370 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30371 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30372 for auto loops. Remove default auto determining, moved to
30373 oacc_loop_fixed_partitions.
30374 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30375 stmts, add e_mask field.
30376 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30377 (oacc_thread_numbers): Use oacc_dim_call.
30378 (oacc_xform_tile): New.
30379 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30380 (finish_oacc_loop): Adjust for ifns vector.
30381 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30382 add case for GOACC_TILE fns.
30383 (oacc_loop_xform_loop): Delete.
30384 (oacc_loop_process): Iterate over call list directly, and add
30385 handling for GOACC_TILE fns.
30386 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30388 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30389 vector partitioning to outer loops. Assign 2 partitions to loops
30390 when available. Add TILE handling.
30391 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30392 (execite_oacc_device_lower): Process GOACC_TILE fns,
30393 ignore unknown specs.
30394 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30395 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30396 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30398 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30400 * configure.ac (ACX_BUGURL): Update.
30401 * configure: Regenerate.
30403 2017-02-09 Richard Biener <rguenther@suse.de>
30405 PR tree-optimization/69823
30406 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30407 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30409 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30411 * config/arc/arc-c.def: Add __NPS400__ definition.
30412 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30413 (TARGET_NPS400): Define.
30415 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30417 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30419 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30420 pointer, arch_info.
30421 (arc_cpu_types): Fill the arch_info field with a pointer into the
30422 arc_arch_types table.
30423 (arc_selected_cpu): Declare.
30424 * config/arc/arc.c (arc_selected_cpu): Make global.
30425 (arc_selected_arch): Delete.
30426 (arc_base_cpu): Delete.
30427 (arc_override_options): Remove references to deleted variables,
30428 update access to arch information.
30429 (ARC_OPT): Update access to arch information.
30430 (ARC_OPTX): Likewise.
30431 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30432 (TARGET_ARC600): Update access to arch information.
30433 (TARGET_ARC601): Likewise.
30434 (TARGET_ARC700): Likewise.
30435 (TARGET_EM): Likewise.
30436 (TARGET_HS): Likewise.
30437 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30440 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30443 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30444 condition/operands for integer GE/LE/GEU/LEU operations.
30446 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30448 PR translation/79397
30449 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30452 2017-02-08 Martin Jambor <mjambor@suse.cz>
30455 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30456 whether allocation happened.
30457 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30458 nothing was allocated.
30460 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30462 PR tree-optimization/79408
30463 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30464 constant, but SSA_NAME with a known integer range, use the minimum
30465 of that range instead of op1 to determine if modulo can be replaced
30466 with its first operand.
30468 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30470 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30472 2017-02-08 Richard Biener <rguenther@suse.de>
30474 PR tree-optimization/71824
30475 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30476 Check all loops contained in the merged region.
30478 2017-02-07 Andrew Pinski <apinski@cavium.com>
30480 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30482 2017-02-07 Andrew Pinski <apinski@cavium.com>
30484 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30485 (thunderxt88): Likewise.
30486 (thunderxt81): Disable LSE and change v8.1 to v8.
30487 (thunderxt83): Likewise.
30489 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30490 Richard Biener <rguenther@suse.de>
30492 PR middle-end/79399
30493 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30494 type from int to size_t.
30495 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30497 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30499 PR rtl-optimization/79386
30500 * cprop.c (bypass_conditional_jumps): Initialize
30501 bypass_last_basic_block already before splitting bbs after
30502 unconditional traps...
30503 (bypass_conditional_jumps): ... rather than here.
30506 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30507 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30508 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30509 fix -masm=intel patterns.
30511 2017-02-07 Richard Biener <rguenther@suse.de>
30513 PR tree-optimization/79256
30514 PR middle-end/79278
30515 * builtins.c (get_object_alignment_2): Use min_align_of_type
30516 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30517 and ADJUST_FIELD_ALIGN.
30519 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30521 * doc/tm.texi: Regenerate.
30522 * stor-layout.c (layout_decl): Adjust.
30523 (update_alignment_for_field): Likewise.
30524 (place_field): Likewise.
30525 (min_align_of_type): Likewise.
30526 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30527 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30528 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30529 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30530 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30531 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30532 * config/i386/i386.c (x86_field_alignment): Likewise.
30533 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30534 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30535 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30536 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30537 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30538 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30542 2017-01-30 Richard Biener <rguenther@suse.de>
30544 PR tree-optimization/79256
30545 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30546 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30549 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30551 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30552 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30553 builtins to SImode and emit a zero-extend, if necessary.
30555 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30557 * docs/invoke.texi (RISC-V Options): Alphabetize.
30559 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30561 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30564 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30566 * config/riscv/riscv.c: New file.
30567 * common/config/riscv/riscv-common.c: Likewise.
30568 * config.gcc: Likewise.
30569 * config/riscv/constraints.md: Likewise.
30570 * config/riscv/elf.h: Likewise.
30571 * config/riscv/generic.md: Likewise.
30572 * config/riscv/linux.h: Likewise.
30573 * config/riscv/multilib-generator: Likewise.
30574 * config/riscv/peephole.md: Likewise.
30575 * config/riscv/pic.md: Likewise.
30576 * config/riscv/predicates.md: Likewise.
30577 * config/riscv/riscv-builtins.c: Likewise.
30578 * config/riscv/riscv-c.c: Likewise.
30579 * config/riscv/riscv-ftypes.def: Likewise.
30580 * config/riscv/riscv-modes.def: Likewise.
30581 * config/riscv/riscv-opts.h: Likewise.
30582 * config/riscv/riscv-protos.h: Likewise.
30583 * config/riscv/riscv.h: Likewise.
30584 * config/riscv/riscv.md: Likewise.
30585 * config/riscv/riscv.opt: Likewise.
30586 * config/riscv/sync.md: Likewise.
30587 * config/riscv/t-elf-multilib: Likewise.
30588 * config/riscv/t-linux: Likewise.
30589 * config/riscv/t-linux-multilib: Likewise.
30590 * config/riscv/t-riscv: Likewise.
30591 * configure.ac: Likewise.
30592 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30593 Waterman as RISC-V maintainers.
30594 * doc/install.texi: Add RISC-V entries.
30595 * doc/invoke.texi: Add RISC-V options section.
30596 * doc/md.texi: Add RISC-V constraints section.
30597 * configure: Regenerated.
30599 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
30602 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
30603 false values to be constant vectors with all 0 or all 1 bits set.
30604 (vcondu<mode><mode>): Likewise.
30605 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
30607 (fpmask_comparison_operator): Update comment.
30608 (vecint_comparison_operator): New predicate.
30609 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
30610 vector conditionals when the true and false values are constant
30611 vectors with all 0 bits or all 1 bits set.
30613 2017-02-06 Martin Sebor <msebor@redhat.com>
30615 PR tree-optimization/79376
30616 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30618 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30620 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30621 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30622 to simplify split condition.
30624 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30626 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30627 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30630 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30632 PR rtl-optimization/68664
30633 * target.def (can_speculate_insn): New hook.
30634 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30635 * doc/tm.texi: Regenerate.
30636 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30637 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30638 (rs6000_sched_can_speculate_insn): New function.
30640 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30642 PR tree-optimization/79284
30643 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30644 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30645 vectorizable_mask_load_store, vectorizable_operation,
30646 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30647 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30648 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30649 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30650 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30651 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30652 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30653 is_gimple_assign (stmt). Replace another such test with
30654 is_gimple_assign (stmt).
30656 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30659 * config/avr/avr.c (rtl-iter.h): Include it.
30660 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30661 (avr_legitimate_combined_insn): ...and implementation.
30663 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30665 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30666 * config/s390/s390.c (s390_const_operand_ok)
30667 (s390_canonicalize_comparison, s390_extract_part)
30668 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30669 (s390_contiguous_bitmask_p, s390_rtx_costs)
30670 (legitimize_pic_address): Likewise.
30671 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30672 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30673 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30674 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30675 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30677 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30679 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30680 REGNO($0) == REGNO($1).
30682 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30684 * config/s390/linux.h(SIZE_TYPE): Add comment.
30686 2017-02-06 Julian Brown <julian@codesourcery.com>
30687 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30688 Virendra Pathak <virendra.pathak@broadcom.com>
30690 * config/aarch64/aarch64-cores.def: Change the scheduler
30692 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30693 * config/aarch64/thunderx2t99.md: New file.
30695 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30697 * doc/standards.texi (Go Language): Update link to language
30700 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30702 * tree-eh.c (lower_resx): Sanitize profile.
30703 (cleanup_empty_eh_move_lp): Likewise.
30705 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30708 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30710 * cfgloopmanip.h (loop_version): Update prototype.
30711 * modulo-sched.c (sms_schedule): Update call of loop_version.
30712 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30713 * tree-parloops.c (gen_parallel_loop): Likewise.
30714 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30715 * tree-ssa-loop-split.c (split_loop): Likewise.
30716 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30717 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30719 2017-02-05 Martin Liska <mliska@suse.cz>
30722 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30724 (print_operand_address): Initialize a struct to zero.
30726 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30728 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30729 garbage collector only in textual form.
30731 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30733 * doc/extend.texi (x86 specific memory model extensions for
30734 transactional memory): Simplify a phrase.
30736 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30739 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30740 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30741 (atomic_storedi_1): Likewise.
30743 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30745 PR tree-optimization/79338
30746 * tree-parloops.c (gather_scalar_reductions): Don't call
30747 vect_analyze_loop_form for loop->inner before destroying loop's
30750 2017-02-03 Martin Sebor <msebor@redhat.com>
30752 PR tree-optimization/79327
30753 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30754 when precision has resulted in leading zeros.
30755 (format_integer): Adjust the likely counter to assume an unknown
30756 argument that may be zero is non-zero.
30758 2017-02-03 Jason Merrill <jason@redhat.com>
30761 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30762 avoid copying non-taken branch.
30764 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30766 PR tree-optimization/79340
30767 * tree-vect-loop.c (vectorizable_reduction): Release
30768 vec_defs elements after safe_splicing them into other vectors.
30771 PR tree-optimization/79327
30772 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30773 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30775 (format_integer): Use wide_int_to_tree instead of build_int_cst
30776 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30777 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30778 of shortest and longest sequence.
30780 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30782 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30783 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30785 2017-02-03 Walter Lee <walt@tilera.com>
30788 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30789 after initial stackframe link reg save.
30790 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30792 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30795 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30796 wu for stxssp alternative.
30798 2017-02-03 Martin Sebor <msebor@redhat.com>
30800 PR tree-optimization/79352
30801 * gimple-fold.c (get_range_strlen): Add argument.
30802 (get_range_strlen): Change return type to bool.
30803 (get_maxval_strlen): Pass in a dummy argument.
30804 * gimple-fold.h (get_range_strlen): Change return type to bool.
30805 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30806 * tree.h (array_at_struct_end_p): Add argument.
30807 * tree.c (array_at_struct_end_p): Handle it.
30809 2017-02-03 Martin Liska <mliska@suse.cz>
30812 * multiple_target.c (create_dispatcher_calls): Redirect edge
30813 from a caller of a dispatcher.
30814 (expand_target_clones): Make the clones local.
30815 (ipa_target_clone): Do both target clones and resolvers.
30816 (ipa_dispatcher_calls): Remove the pass.
30817 (pass_dispatcher_calls::gate): Likewise.
30818 (make_pass_dispatcher_calls): Likewise.
30819 * passes.def (pass_target_clone): Put as very first IPA early
30822 2017-02-03 Martin Liska <mliska@suse.cz>
30824 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30825 in case of a function with ifunc attribute.
30827 2017-02-03 Martin Liska <mliska@suse.cz>
30829 * cgraph.c (cgraph_node::dump): Dump function version info.
30830 * symtab.c (symtab_node::dump_base): Add missing new line.
30832 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30834 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30835 (ifcombine_ifandif): Use it.
30837 2017-02-03 Martin Liska <mliska@suse.cz>
30839 * doc/invoke.texi: Document default value for
30840 use-after-scope-direct-emission-threshold.
30842 2017-02-03 Martin Liska <mliska@suse.cz>
30844 PR tree-optimization/79339
30845 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30846 (format_floating): Likewise.
30848 2017-02-03 Martin Liska <mliska@suse.cz>
30851 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30853 (ipa_node_params_t::remove): Likewise.
30854 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30855 initialization from removed ipa_node_params_t::insert.
30856 (ipa_node_params::~ipa_node_params): Move from removed
30857 ipa_node_params_t::release.
30858 * symbol-summary.h (symbol_summary::m_released): New member.
30859 Do not release a summary twice. Do not allow to call finalizer
30860 for types of a summary that live in GGC memory.
30862 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30864 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30867 2017-02-02 Martin Sebor <msebor@redhat.com>
30869 PR middle-end/79275
30870 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30871 (format_string): Tighten up the range of output for non-constant
30872 strings and correct the expected range for wide non-constant strings.
30874 2017-02-02 Martin Sebor <msebor@redhat.com>
30876 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30878 PR middle-end/32003
30879 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30881 (-fdump-tree-@var): Add to index and document how to come up
30882 with pass-specific option and dump file names.
30883 (-fdump-passes): Clarify where to look for output.
30885 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30887 PR middle-end/77445
30888 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30889 statistics of the analyzed path; allow threading for speed when
30890 any of BBs along the path are optimized for speed.
30892 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30894 PR middle-end/78468
30895 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30896 settings of the virtual registers.
30899 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30901 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30902 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30905 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30907 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30908 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30910 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30912 * config/s390/s390.md: Add missing comments with the expanded
30914 * config/s390/vector.md: Likewise.
30915 * config/s390/vx-builtins.md: Likewise.
30917 2017-02-02 Jakub Jelinek <jakub@redhat.com>
30920 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
30921 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
30922 conditions on a single line.
30924 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30926 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30927 __S390_VX__ to __VX__.
30929 2017-02-01 Andrew Pinski <apinski@cavium.com>
30931 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
30932 stmt_info to record_stmt_cost.
30933 (vect_get_known_peeling_cost): Pass stmt_info if known to
30935 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
30936 cpu_vector_cost field into
30937 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
30938 field into vec_int_stmt_cost and vec_fp_stmt_cost.
30939 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
30940 splitting of scalar_stmt_cost and vec_stmt_cost.
30941 (thunderx_vector_cost): Likewise.
30942 (cortexa57_vector_cost): LIkewise.
30943 (exynosm1_vector_cost): Likewise.
30944 (xgene1_vector_cost): Likewise.
30945 (thunderx2t99_vector_cost): Improve after the splitting of the two
30947 (aarch64_builtin_vectorization_cost): Update for the splitting of
30948 scalar_stmt_cost and vec_stmt_cost.
30950 2017-02-01 Torvald Riegel <triegel@redhat.com>
30951 Richard Henderson <rth@redhat.com>
30953 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
30954 conditional on existance of a fast atomic load.
30955 * optabs-query.c (can_atomic_load_p): New function.
30956 * optabs-query.h (can_atomic_load_p): Declare it.
30957 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
30958 no fast atomic load is available for the particular size of access.
30959 (expand_atomic_compare_and_swap): Likewise.
30960 (expand_atomic_load): Likewise.
30961 (expand_atomic_store): Likewise.
30962 (expand_atomic_fetch_op): Likewise.
30963 * testsuite/lib/target-supports.exp
30964 (check_effective_target_sync_int_128): Remove x86 because it provides
30965 no fast atomic load.
30966 (check_effective_target_sync_int_128_runtime): Likewise.
30968 2017-02-01 Richard Biener <rguenther@suse.de>
30970 * graphite.c: Include tree-vectorizer.h for find_loop_location.
30971 (graphite_transform_loops): Provide opt-info for optimized nests.
30972 * tree-parloop.c (parallelize_loops): Provide opt-info for
30973 parallelized loops.
30975 2017-02-01 Richard Biener <rguenther@suse.de>
30977 PR middle-end/79315
30978 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
30979 was not set before.
30981 2017-02-01 Richard Biener <rguenther@suse.de>
30983 PR tree-optimization/71824
30984 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30985 Verify the loops are valid in the merged SESE region.
30986 (scop_detection::can_represent_loop_1): Check analyzing the
30987 evolution of the number of iterations in the region succeeds.
30989 2017-01-31 Ian Lance Taylor <iant@golang.org>
30991 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
30992 REG_ARGS_SIZE note to 32-bit push insns and call insn.
30994 2017-01-31 David Malcolm <dmalcolm@redhat.com>
30996 PR preprocessor/79210
30997 * input.c (get_substring_ranges_for_loc): Replace line_width
30998 assertion with error-handling.
31000 2017-01-31 Richard Biener <rguenther@suse.de>
31002 PR tree-optimization/77318
31003 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31004 (create_pw_aff_from_tree): Take loop parameter.
31005 (add_condition_to_pbb): Pass loop of the condition to
31006 create_pw_aff_from_tree.
31008 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31010 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31011 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31013 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31017 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31019 (convert_int_to_float128): Likewise.
31020 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31021 (convert_int_to_float128): Likewise.
31022 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31023 (UNSPEC_IEEE128_CONVERT): Likewise.
31024 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31025 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31026 Use local variables for IBM extended format.
31027 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31028 (fix_trunc<mode>si2_fprs): Likewise.
31029 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31030 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31031 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31032 to know that we can now have integers of all sizes in vector
31034 (fix<uns>_<mode>di2_hw): Likewise.
31035 (float<uns>_<mode>si2_hw): Likewise.
31036 (fix_<mode>si2_hw): Likewise.
31037 (fixuns_<mode>si2_hw): Likewise.
31038 (float<uns>_<mode>di2_hw): Likewise.
31039 (float_<mode>di2_hw): Likewise.
31040 (float_<mode>si2_hw): Likewise.
31041 (floatuns_<mode>di2_hw): Likewise.
31042 (floatuns_<mode>si2_hw): Likewise.
31043 (xscvqp<su>wz_<mode>): Delete, no longer used.
31044 (xscvqp<su>dz_<mode>): Likewise.
31045 (xscv<su>dqp_<mode>): Likewise.
31046 (ieee128_mfvsrd_64bit): Likewise.
31047 (ieee128_mfvsrd_32bit): Likewise.
31048 (ieee128_mfvsrwz): Likewise.
31049 (ieee128_mtvsrw): Likewise.
31050 (ieee128_mtvsrd_64bit): Likewise.
31051 (ieee128_mtvsrd_32bit): Likewise.
31053 2017-01-31 Martin Liska <mliska@suse.cz>
31056 * ipa-prop.c (ipa_free_all_node_params): Call release method
31057 instead of ~sumbol_summary to not to trigger double times
31060 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31062 PR tree-optimization/71691
31063 * bitmap.h (class auto_bitmap): New.
31064 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31065 is_maybe_undefined instead of ssa_undefined_value_p.
31067 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31069 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31070 __S390_ARCH_LEVEL__ to __ARCH__.
31072 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31074 PR tree-optimization/79267
31075 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31076 if should_remove_lhs_p is true.
31078 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31081 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31082 (add_alignment_attribute): New.
31083 (base_type_die): Add alignment attribute.
31084 (subrange_type_die): Likewise.
31085 (modified_type_die): Likewise.
31086 (gen_array_type_die): Likewise.
31087 (gen_descr_array_type_die: Likewise.
31088 (gen_enumeration_type_die): Likewise.
31089 (gen_subprogram_die): Likewise.
31090 (gen_variable_die): Likewise.
31091 (gen_field_die): Likewise.
31092 (gen_ptr_to_mbr_type_die): Likewise.
31093 (gen_struct_or_union_type_die): Likewise.
31094 (gen_subroutine_type_die): Likewise.
31095 (gen_typedef_die): Likewise.
31096 (base_type_cmp): Compare alignment attribute.
31098 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31101 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
31102 (setb_unsigned) New pattern for setb with CCUNS.
31103 * config/rs6000/rs6000.c (expand_block_compare): Use a different
31104 subfc./subfe sequence to avoid overflow problems. Generate a
31105 shorter sequence with cmpld/setb for power9.
31106 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
31107 for generating subfc. instruction.
31108 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
31109 now uses this instruction.
31111 2017-01-30 Ian Lance Taylor <iant@google.com>
31114 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
31115 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
31117 2017-01-30 Martin Sebor <msebor@redhat.com>
31119 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
31120 Move constant to the right of a relational operator.
31121 (get_mpfr_format_length, format_character, format_string): Ditto.
31122 (should_warn_p, maybe_warn): Same.
31124 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
31126 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
31129 * asan.c (get_translation_unit_decl): Remove function.
31130 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
31132 2017-01-30 Martin Liska <mliska@suse.cz>
31134 PR gcov-profile/79259
31135 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
31136 -fprofile-generate.
31138 2017-01-30 Martin Liska <mliska@suse.cz>
31141 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
31142 Initialize variables with NULL value.
31144 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
31147 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
31149 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
31151 2017-01-30 Richard Biener <rguenther@suse.de>
31153 PR tree-optimization/79276
31154 * tree-vrp.c (process_assert_insertions): Properly adjust common
31155 when removing a duplicate.
31157 2017-01-30 Richard Biener <rguenther@suse.de>
31159 PR tree-optimization/79256
31160 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31161 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31163 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31165 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31168 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31169 ("*r<noxa>sbg_<mode>_sll_bitmask")
31170 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31171 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31172 Use contiguous_bitmask_nowrap_operand.
31174 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31177 * config/rs6000/altivec.h (vec_xl): Revise #define.
31178 (vec_xst): Likewise.
31180 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31182 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31184 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31186 PR rtl-optimization/79194
31187 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31188 traps before call to bypass_conditional_jumps.
31190 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31192 PR tree-optimization/71374
31193 * lra-constraints.c (check_conflict_input_operands): New.
31194 (match_reload): Use it.
31196 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31199 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31200 account to calculate conflict_set.
31202 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31204 PR rtl-optimization/78559
31205 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31206 other_insn in combine.
31208 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31210 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31211 uint16_type_node for BT_UINT16.
31213 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31215 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31216 "RTL Tests" to menu.
31217 (GIMPLE Tests): New node.
31218 (RTL Tests): New node.
31220 2017-01-27 Richard Biener <rguenther@suse.de>
31222 PR tree-optimization/79245
31223 * tree-loop-distribution.c (distribute_loop): Apply cost
31224 modeling also to detected patterns.
31226 2017-01-27 Richard Biener <rguenther@suse.de>
31228 PR tree-optimization/71433
31229 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31230 (compare_assert_loc): New function.
31231 (process_assert_insertions): Sort and optimize assert locations
31232 to remove duplicates and push down identical assertions on
31233 edges to their destination block.
31235 2017-01-27 Richard Biener <rguenther@suse.de>
31237 PR tree-optimization/79244
31238 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31239 out SSA names even if abnormal.
31241 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31243 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31244 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31245 instead of MPFR_RNDN.
31247 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31250 * arm.c (arm_option_override): Don't call build_target_option_node
31251 until after doing all option overrides.
31252 (arm_valid_target_attribute_tree): Likewise.
31254 2017-01-27 Martin Liska <mliska@suse.cz>
31256 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31257 for {cd}tors and C++ {cd}tors.
31259 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31261 * config/s390/s390.md ("*setmem_long_and")
31262 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31264 2017-01-26 Martin Sebor <msebor@redhat.com>
31266 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31269 2017-01-26 Martin Sebor <msebor@redhat.com>
31271 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31272 HAVE_DFmode before using XFmode or DFmode.
31273 (parse_directive): Avoid using the z length modifier to avoid
31274 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31276 PR middle-end/78703
31277 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31278 to accept adjustment as an array.
31279 (get_int_range): New function.
31280 (struct directive): Make width and prec arrays.
31281 (directive::set_width, directive::set_precision): Call get_int_range.
31282 (format_integer, format_floating): Handle width and precision ranges.
31283 (format_string, parse_directive): Same.
31285 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31288 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31289 comdat_type_p set, just clone them, but keep the children in the
31293 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31294 which have direct callers with -fvar-tracking-assignments enabled
31296 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31297 inside of type units.
31299 2017-01-26 Martin Sebor <msebor@redhat.com>
31301 PR middle-end/78703
31302 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31304 (struct format_result): Replace number_chars, number_chars_min,
31305 and number_chars_max with a single member of struct result_range.
31307 (format_result::operator+=): Adjust.
31308 (struct fmtresult): Remove bounded. Handle likely and unlikely
31310 (fmtresult::adjust_for_width_or_precision): New function.
31311 (fmtresult:type_max_digits): New function.
31312 (bytes_remaining): Handle likely and unlikely counters.
31313 (min_bytes_remaining): Remove.
31314 (format_percent): Simplify.
31315 (format_integer, format_floating): Set likely and unlikely counters.
31316 (get_string_length, format_character, format_string): Same.
31317 (format_plain, should_warn_p): New function.
31318 (maybe_warn): Call should_warn_p. Update diagnostic messages
31319 and handle those for all directives, including plain strings.
31320 (format_directive): Handle likely and unlikely counters.
31321 Remove unnecessary quoting from diagnostics. Add an informational
31323 (add_bytes): Remove.
31324 (pass_sprintf_length::compute_format_length): Simplify.
31325 (try_substitute_return_value): Handle likely and unlikely counters.
31327 2017-01-26 Carl Love <cel@us.ibm.com>
31329 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31330 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31332 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31335 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31336 endianess for subregs into account.
31337 * lra-constraints.c (lra_constraints): Do risky transformations
31338 always on the first iteration.
31339 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31340 last_call_used_reg_set.
31341 (process_bb_lives): Define and use last_call_used_reg_set.
31342 * lra.c (lra): Always continue after lra_constraints on the first
31345 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31347 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31349 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31351 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31353 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31354 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31355 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31356 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31357 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31358 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31359 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31360 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31361 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31363 2017-01-26 Marek Polacek <polacek@redhat.com>
31366 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31367 for the third operand.
31369 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31371 PR middle-end/79236
31372 * omp-low.c (struct omp_context): Add simt_stmt field.
31373 (scan_omp_for): Return omp_context *.
31374 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31375 context to the _simt_ SIMD stmt.
31376 (lower_omp_for): For combined SIMD with sibling _simt_
31377 SIMD, make sure to use the same decls in _looptemp_
31378 clauses as in the sibling.
31380 2017-01-26 David Sherwood <david.sherwood@arm.com>
31382 PR middle-end/79212
31383 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31386 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31389 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31390 emit fld b; fld a; if possible.
31392 * brig-builtins.def: Update copyright years.
31393 * config/arm/arm_acle_builtins.def: Update copyright years.
31395 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31398 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31399 constraint instead of o for the stxsd instruction.
31401 2017-01-25 Carl Love <cel@us.ibm.com>
31403 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31404 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31406 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31408 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31410 2017-01-25 Richard Biener <rguenther@suse.de>
31412 PR tree-optimization/69264
31413 * target.def (vector_alignment_reachable): Improve documentation.
31414 * doc/tm.texi: Regenerate.
31415 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31417 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31418 earlier changes with respect to TYPE_USER_ALIGN.
31419 (vector_alignment_reachable_p): Likewise. Improve dumping.
31421 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31424 * config/arm/arm.md (xordi3): Force constant operand into a register
31427 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31429 * doc/invoke.texi (-fstore-merging): Correct default optimization
31430 levels at which it is enabled.
31431 (-O): Move -fstore-merging from list to...
31434 2017-01-25 Richard Biener <rguenther@suse.de>
31437 * omp-expand.c: Include debug.h.
31438 (expand_omp_taskreg): Make sure to generate early debug before
31439 outlining anything from a function.
31440 (expand_omp_target): Likewise.
31441 (grid_expand_target_grid_body): Likewise.
31443 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31446 * asan.c (get_translation_unit_decl): New function.
31447 (asan_add_global): Extract modules file name from globals
31448 TRANSLATION_UNIT_DECL name.
31450 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31453 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31454 for long calls with APCS frame and VFP.
31456 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31458 * cfg.c (original_copy_tables_initialized_p): New function.
31459 * cfg.h (original_copy_tables_initialized_p): New decl.
31460 * cfgrtl.c (relink_block_chain): Guard the call to
31461 free_original_copy_tables with a call to
31462 original_copy_tables_initialized_p.
31463 * cgraph.h (symtab_node::native_rtl_p): New decl.
31464 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31465 (symtab_node::needed_p): Don't assert for early assembly output
31466 for __RTL functions.
31467 (cgraph_node::finalize_function): Set "force_output" for __RTL
31469 (cgraph_node::analyze): Bail out early for __RTL functions.
31470 (analyze_functions): Update assertion to support __RTL functions.
31471 (cgraph_node::expand): Bail out early for __RTL functions.
31472 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31474 * function.h (struct function): Update comment for field
31476 * gimple-expr.c: Include "tree-pass.h".
31477 (gimple_has_body_p): Return false for __RTL functions.
31478 * Makefile.in (OBJS): Add run-rtl-passes.o.
31479 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31481 (gcc::pass_manager::get_clean_slate): New accessor.
31482 * passes.c: Include "insn-addr.h".
31483 (should_skip_pass_p): Add logging. Update logic for running
31484 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31485 property-provider override so it is only done for gimple passes.
31487 (skip_pass): New function.
31488 (execute_one_pass): Call skip_pass when skipping passes.
31489 * read-md.c (md_reader::read_char): Support filtering
31490 the input to a subset of line numbers.
31491 (md_reader::md_reader): Initialize fields
31492 m_first_line and m_last_line.
31493 (md_reader::read_file_fragment): New function.
31494 * read-md.h (md_reader::read_file_fragment): New decl.
31495 (md_reader::m_first_line): New field.
31496 (md_reader::m_last_line): New field.
31497 * read-rtl-function.c (function_reader::create_function): Only
31498 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31499 curr_properties. Set DECL_INITIAL to a dummy block.
31500 (read_rtl_function_body_from_file_range): New function.
31501 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31503 * run-rtl-passes.c: New file.
31504 * run-rtl-passes.h: New file.
31506 2017-01-24 Jeff Law <law@redhat.com>
31508 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31511 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31513 PR tree-optimization/79159
31514 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31515 (record_nonwrapping_iv): Improve boundary using above function if no
31516 value range information.
31518 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31519 Martin Jambor <mjambor@suse.cz>
31521 * brig-builtins.def: New file.
31522 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31523 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31524 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31525 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31526 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31527 * builtin-types.def (BT_INT8): New.
31528 (BT_INT16): Likewise.
31529 (BT_UINT8): Likewise.
31530 (BT_UINT16): Likewise.
31531 (BT_FN_ULONG): Likewise.
31532 (BT_FN_UINT_INT): Likewise.
31533 (BT_FN_UINT_ULONG): Likewise.
31534 (BT_FN_UINT_LONG): Likewise.
31535 (BT_FN_UINT_PTR): Likewise.
31536 (BT_FN_ULONG_PTR): Likewise.
31537 (BT_FN_INT8_FLOAT): Likewise.
31538 (BT_FN_INT16_FLOAT): Likewise.
31539 (BT_FN_UINT32_FLOAT): Likewise.
31540 (BT_FN_UINT16_FLOAT): Likewise.
31541 (BT_FN_UINT8_FLOAT): Likewise.
31542 (BT_FN_UINT64_FLOAT): Likewise.
31543 (BT_FN_UINT16_UINT32): Likewise.
31544 (BT_FN_UINT32_UINT16): Likewise.
31545 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31546 (BT_FN_INT_PTR_INT): Likewise.
31547 (BT_FN_UINT_PTR_UINT): Likewise.
31548 (BT_FN_LONG_PTR_LONG): Likewise.
31549 (BT_FN_ULONG_PTR_ULONG): Likewise.
31550 (BT_FN_VOID_UINT64_UINT64): Likewise.
31551 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31552 (BT_FN_INT8_INT8_INT8): Likewise.
31553 (BT_FN_INT16_INT16_INT16): Likewise.
31554 (BT_FN_INT_INT_INT): Likewise.
31555 (BT_FN_UINT_FLOAT_UINT): Likewise.
31556 (BT_FN_FLOAT_UINT_UINT): Likewise.
31557 (BT_FN_ULONG_UINT_UINT): Likewise.
31558 (BT_FN_ULONG_UINT_PTR): Likewise.
31559 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31560 (BT_FN_UINT_UINT_UINT): Likewise.
31561 (BT_FN_VOID_UINT_PTR): Likewise.
31562 (BT_FN_UINT_UINT_PTR: Likewise.
31563 (BT_FN_UINT32_UINT64_PTR): Likewise.
31564 (BT_FN_INT_INT_UINT_UINT): Likewise.
31565 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31566 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31567 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31568 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31569 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31570 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31571 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31572 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31573 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31574 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31575 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31576 * doc/frontends.texi: List BRIG FE.
31577 * doc/install.texi (Testing): Add BRIG tesring requirements.
31578 * doc/invoke.texi (Overall Options): Mention BRIG.
31579 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31581 2017-01-24 Richard Biener <rguenther@suse.de>
31583 PR translation/79208
31584 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31586 2017-01-24 Martin Jambor <mjambor@suse.cz>
31589 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31590 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31591 and known_contexts.
31593 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31595 PR middle-end/79123
31596 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31597 casts from signed to unsigned really don't have a range.
31599 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
31601 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
31602 GMP_RNDx for compatiblity.
31604 2017-01-24 Martin Liska <mliska@suse.cz>
31607 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
31608 that would prevent us to call alloca with -1 as argument.
31610 2017-01-24 Jakub Jelinek <jakub@redhat.com>
31612 * dwarf2out.c (output_compilation_unit_header, output_file_names):
31613 Avoid -Wformat-security warning.
31615 2017-01-23 Andrew Pinski <apinski@cavium.com>
31617 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31620 2017-01-23 Martin Sebor <msebor@redhat.com>
31622 PR middle-end/78703
31623 * gimple-ssa-sprintf.c (warn_level): New global.
31624 (format_integer): Use it here and throughout the rest of the file.
31625 Use the same switch to compute sign as base.
31626 (maybe_warn): New function.
31627 (format_directive): Factor out warnings into maybe_warn.
31628 Add debugging output. Use warn_level.
31629 (add_bytes): Use warn_level.
31630 (pass_sprintf_length::compute_format_length): Add debugging output.
31631 (try_substitute_return_value): Same.
31632 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31634 PR middle-end/78703
31635 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31636 (struct fmtresult, format_integer, format_floating): Adjust.
31637 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31638 (get_string_length, format_string,format_directive): Same.
31639 (pass_sprintf_length::compute_format_length): Same.
31640 (try_substitute_return_value): Simplify slightly.
31642 PR middle-end/78703
31643 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31644 (fmtresult::operator+=): Outlined.
31645 (struct fmtresult): Add ctors.
31646 (struct conversion_spec): Rename...
31647 (struct directive): ...to this. Add and remove data members.
31648 (directive::set_width, directive::set_precision): New functions.
31649 (format_percent): Use fmtresult ctor.
31650 (get_width_and_precision): Remove.
31651 (format_integer): Make naming changes. Avoid computing width and
31653 (format_floating): Same. Adjust indentation.
31654 (format_character, format_none): New functions.
31655 (format_string): Moved character handling to format_character.
31656 (format_directive): Remove arguments, change return type.
31657 (parse_directive): New function.
31658 (pass_sprintf_length::compute_format_length): Move directive
31659 parsing to parse_directive.
31661 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31663 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31664 (assign_assembler_name_if_needed): ... this.
31665 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31666 (assign_assembler_name_if_needed): ... this.
31667 (free_lang_data_in_cgraph): Adjust callers.
31668 * cgraphunit.c (cgraph_node::analyze): Likewise.
31669 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31672 2017-01-23 Richard Biener <rguenther@suse.de>
31674 PR tree-optimization/79088
31675 PR tree-optimization/79188
31676 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31677 resetting loop bounds after last path deletion. Reset loop
31678 bounds of the target loop, make code match the comments.
31679 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31680 Make sure loops need no fixups.
31682 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31684 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31685 exponent support with double type for first argument.
31686 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31687 type returned by __builtin_vec_extract_sig,
31688 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31689 functions from "vector int" to "vector unsigned int" or from
31690 "vector long long int" to "vector unsigned long long int".
31691 Changed type returned by __builtin_vec_extract_exp,
31692 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31693 functions from "vector int" to "vector unsigned int" or from
31694 "vector long long int" to "vector unsigned long long int".
31695 Changed return type of __builtin_vec_test_data_class,
31696 __builtin_vec_test_data_class_sp, and
31697 __builtin_vec_test_data_class_dp from "vector int" to
31698 "vector bool int" or from "vector long long int" to "vector bool
31699 long long int" and changed second argument type from "unsigned
31700 int" to "int". Added new overloaded function forms "vector float
31701 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31702 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31703 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31704 double, vector unsigned long long int)" and "vector double
31705 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31706 long int)". Changed return type of
31707 __builtin_scalar_test_data_class and
31708 __builtin_scalar_test_data_class_sp and
31709 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31710 int" and changed second argument from "unsigned int" to "int".
31711 Changed type returned by __builtin_scalar_test_neg,
31712 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31713 from "int" to "bool int". Added new overloaded function form
31714 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31715 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31716 exponent double-precision with floating point first argument.
31717 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31718 documentation of scalar_test_data_class, scalar_test_neg,
31719 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31720 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31721 vec_test_data_class built-in functions to reflect refinements in
31722 their type signatures.
31724 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31726 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31728 (aarch64_elf_asm_destructor): Likewise.
31730 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31732 PR rtl-optimization/78634
31733 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31734 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31735 * ifcvt.c (noce_try_cmove): Add missing cost check.
31737 PR rtl-optimization/71724
31738 * combine.c (if_then_else_cond): Look for situations where it is
31739 beneficial to undo the work of one of the recursive calls.
31741 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31743 PR tree-optimization/70754
31744 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31745 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31746 combined stmt before it if not NULL.
31747 (combine_chains): Process refs reversely and compute dominance point
31750 2017-01-23 Martin Liska <mliska@suse.cz>
31752 PR tree-optimization/79196
31753 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31754 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31756 (strlen_optimize_stmt): Call the renamed function.
31758 2017-01-23 Michael Matz <matz@suse.de>
31760 PR tree-optimization/78384
31761 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31763 2017-01-23 Richard Biener <rguenther@suse.de>
31765 PR tree-optimization/79186
31766 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31767 both incoming edges before moving an assert.
31769 2017-01-23 Martin Jambor <mjambor@suse.cz>
31771 * ipa-prop.c (load_from_param_1): Removed.
31772 (load_from_unmodified_param): Bits from load_from_param_1 put back
31774 (load_from_param): Removed.
31775 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31776 with stmt. Reverted back to use of load_from_unmodified_param.
31778 2017-01-23 Martin Jambor <mjambor@suse.cz>
31781 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31782 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31783 field a pointer to garbage collected vector, mark lattices and
31784 ipcp_orig_node with GTY((skip)).
31785 (ipa_get_param_count): Adjust to descriptors being a pointer.
31786 (ipa_get_param): Likewise.
31787 (ipa_get_type): Likewise.
31788 (ipa_get_param_move_cost): Likewise.
31789 (ipa_set_param_used): Likewise.
31790 (ipa_get_controlled_uses): Likewise.
31791 (ipa_set_controlled_uses): Likewise.
31792 (ipa_is_param_used): Likewise.
31793 (ipa_node_params_t): Move into garbage collector. New methods insert
31795 (ipa_node_params_sum): Annotate wth GTY(()).
31796 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31798 (ipa_load_from_parm_agg): Adjust declaration.
31799 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31800 * ipa-profile.c (ipa_profile): Likewise.
31801 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31802 (ipa_populate_param_decls): Make descriptors parameter garbage
31804 (ipa_dump_param): Adjust to descriptors being a pointer.
31805 (ipa_alloc_node_params): Likewise.
31806 (ipa_initialize_node_params): Likewise.
31807 (load_from_param_1): Make descriptors parameter garbage collected.
31808 (load_from_unmodified_param): Likewise.
31809 (load_from_param): Likewise.
31810 (ipa_load_from_parm_agg): Likewise.
31811 (ipa_node_params::~ipa_node_params): Removed.
31812 (ipa_free_all_node_params): Remove call to delete operator.
31813 (ipa_node_params_t::insert): New.
31814 (ipa_node_params_t::remove): Likewise.
31815 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31816 copy known_csts and known_contexts vectors.
31817 (ipa_read_node_info): Adjust to descriptors being a pointer.
31818 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31820 (ipcp_transform_function): Make descriptors variable garbage
31823 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31825 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31826 * config/i386/avx512dqintrin.h: Ditto.
31827 * config/i386/avx512fintrin.h: Ditto.
31828 * config/i386/i386.c: Handle new builtins.
31829 * config/i386/i386-builtin.def: Add new builtins.
31830 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31831 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31833 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31834 Martin Liska <mliska@suse.cz>
31836 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31837 * asan.c (asan_expand_poison_ifn): Support stores and use
31838 appropriate ASAN report function.
31839 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31840 * internal-fn.def (ASAN_POISON_USE): Declare.
31841 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31842 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31843 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31844 ASAN_POISON calls w/o LHS.
31845 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31846 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31847 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31848 * gimplify.c (asan_poison_variables): Add attribute
31849 use_after_scope_memory to variables that really needs to live
31851 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31852 having the attribute.
31854 2017-01-23 Martin Liska <mliska@suse.cz>
31856 * asan.c (create_asan_shadow_var): New function.
31857 (asan_expand_poison_ifn): Likewise.
31858 * asan.h (asan_expand_poison_ifn): New declaration.
31859 * internal-fn.c (expand_ASAN_POISON): Likewise.
31860 * internal-fn.def (ASAN_POISON): New builtin.
31861 * sanopt.c (pass_sanopt::execute): Expand
31862 asan_expand_poison_ifn.
31863 * tree-inline.c (copy_decl_for_dup_finish): Make function
31865 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31866 * tree-ssa.c (is_asan_mark_p): New function.
31867 (execute_update_addresses_taken): Rewrite local variables
31868 (identified just by use-after-scope as addressable) into SSA.
31870 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31872 * doc/install.texi (Specific): opensource.apple.com uses https
31873 now. Remove trailing slash.
31875 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31877 * README.Portability: Remove note on an Irix compatibility issue.
31879 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31881 * gcov.c (INCLUDE_ALGORITHM): Define.
31882 (INCLUDE_VECTOR): Define.
31883 No longer include <vector> and <algorithm> directly.
31885 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31887 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31889 * doc/invoke.texi (Code Gen Options): Ditto.
31891 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31894 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31896 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31898 rtl-optimization/79125
31899 * cprop.c (local_cprop_pass): Handle cases where we make an
31900 unconditional trap.
31902 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31906 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31907 read from, for big endian.
31909 2017-01-20 Jiong Wang <jiong.wang@arm.com>
31911 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
31912 register pauth builtins for LP64 only.
31914 2017-01-20 Marek Polacek <polacek@redhat.com>
31917 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
31920 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31922 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
31924 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
31925 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
31926 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
31928 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31931 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
31932 in big-endian mode when they are not a single duplicated value.
31934 2017-01-20 Richard Biener <rguenther@suse.de>
31936 * BASE-VER: Bump to 7.0.1.
31938 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31940 * omp-low.c (omplow_simd_context): New struct. Use it...
31941 (lower_rec_simd_input_clauses): ...here and...
31942 (lower_rec_input_clauses): ...here to hold common data. Adjust all
31943 references to idx, lane, max_vf, is_simt.
31945 2017-01-20 Graham Markall <graham.markall@embecosm.com>
31947 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
31950 2017-01-20 Martin Jambor <mjambor@suse.cz>
31952 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
31953 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
31955 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
31956 (GTFILES): Rename hsa.c to hsa-common.c.
31957 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
31958 * hsa-dump.c: Likewise.
31959 * hsa-gen.c: Likewise.
31960 * hsa-regalloc.c: Likewise.
31961 * ipa-hsa.c: Likewise.
31962 * omp-expand.c: Likewise.
31963 * omp-low.c: Likewise.
31964 * toplev.c: Likewise.
31966 2017-01-20 Marek Polacek <polacek@redhat.com>
31969 * doc/invoke.texi: Document -Wduplicated-branches.
31970 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
31971 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
31972 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
31973 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
31974 return 0 only when not OEP_LEXICOGRAPHIC.
31975 (fold_build_cleanup_point_expr): Use the expression
31976 location when building CLEANUP_POINT_EXPR.
31977 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
31978 * tree.c (add_expr): Handle error_mark_node.
31980 2017-01-20 Martin Liska <mliska@suse.cz>
31983 * tree-profile.c (init_ic_make_global_vars): Do not call
31985 (gimple_init_gcov_profiler): Likewise.
31987 2017-01-20 Martin Liska <mliska@suse.cz>
31990 * cgraph.h (maybe_create_reference): Remove argument and
31992 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
31994 * ipa-cp.c (create_specialized_node): Likewise.
31995 * symtab.c (symtab_node::maybe_create_reference): Handle
31996 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
31998 2017-01-20 Martin Liska <mliska@suse.cz>
32000 * read-rtl-function.c (function_reader::create_function): Use
32001 build_decl instread of build_decl_stat.
32003 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32005 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32006 * config/i386/avx512dqintrin.h: Ditto.
32007 * config/i386/avx512fintrin.h: Ditto.
32008 * config/i386/i386-builtin-types.def: Add new types.
32009 * config/i386/i386.c: Handle new types.
32010 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32011 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32012 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32013 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32014 (__builtin_ia32_kshiftridi): New.
32015 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32017 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32021 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32022 define to rs6000_init_stack_protect_guard.
32023 (rs6000_init_stack_protect_guard): New function.
32025 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32026 Yunqiang Su <yunqiang.su@imgtec.com>
32028 * config.gcc (supported_defaults): Add madd4.
32029 (with_madd4): Add validation.
32030 (all_defaults): Add madd4.
32031 * config/mips/mips.opt (mmadd4): New option.
32032 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32034 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32036 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32037 (ISA_HAS_FUSED_MADD4): Likewise.
32038 * doc/invoke.texi (-mmadd4): Document the new option.
32039 * doc/install.texi (--with-madd4): Document the new option.
32041 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32043 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32044 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32045 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32046 (aarch64_init_pauth_hint_builtins): New.
32047 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32048 (aarch64_expand_builtin): Expand new builtins.
32050 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32052 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32053 * combine-stack-adj.c (no_unhandled_cfa): Handle
32054 REG_CFA_TOGGLE_RA_MANGLE.
32055 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32056 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32057 info for return address signing.
32058 (aarch64_expand_epilogue): Likewise.
32060 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32062 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32063 * config/aarch64/aarch64-protos.h
32064 (aarch64_return_address_signing_enabled): New declaration.
32065 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32067 (aarch64_expand_prologue): Sign return address before it's pushed onto
32069 (aarch64_expand_epilogue): Authenticate return address fetched from
32071 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32072 (aarch64_attributes): New function attributes for "sign-return-address".
32073 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32074 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32075 ("*do_return"): Generate combined instructions according to key index.
32076 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32077 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32079 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32080 * config/aarch64/aarch64.opt (msign-return-address=): New.
32081 * doc/extend.texi (AArch64 Function Attributes): Documents
32082 "sign-return-address=".
32083 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32085 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32087 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32088 overall option summary.
32090 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32092 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32093 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32094 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32095 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32097 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
32099 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
32100 -mpower9-minmax by default for -mcpu=power9.
32101 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
32102 128-bit floating point.
32104 2017-01-20 Alan Modra <amodra@gmail.com>
32106 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
32107 optimizing for size.
32109 2017-01-20 Alan Modra <amodra@gmail.com>
32112 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
32113 for strcmp and strncmp from corresponding builtin decl.
32115 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32117 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
32118 instead of i386/rtems-64.h.
32119 * config/i386/rtems-64.h: Remove.
32121 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32125 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
32127 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
32129 2017-01-19 Tamar Christina <tamar.christina@arm.com>
32131 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
32132 Change int to HOST_WIDE_INT.
32133 * config/aarch64/aarch64-protos.h
32134 (aarch64_simd_gen_const_vector_dup): Likewise.
32135 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
32137 2017-01-19 David Malcolm <dmalcolm@redhat.com>
32139 * langhooks-def.h (lhd_type_for_size): New decl.
32140 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
32141 * langhooks.c (lhd_type_for_size): New function, taken from
32144 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
32146 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
32147 define_bypass for CR latency.
32148 (power9-cracked-alu): Update bypass latency and remove power9-branch.
32149 (power9-alu2): Add define_bypass for CR latency.
32151 (power9-mul): Update insn latency.
32152 (power9-mul-compare): Update insn latency, bypass latency and remove
32155 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32157 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
32159 * config/aarch64/aarch64.md
32160 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
32161 aarch64_nopcrelative_literal_loads.
32162 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32164 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32166 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32167 TARGET_LOONGSON_3A.
32168 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32170 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32173 * config.gcc (supported_defaults): Add lxc1-sxc1.
32174 (with_lxc1_sxc1): Add validation.
32175 (all_defaults): Add lxc1-sxc1.
32176 * config/mips/mips.opt (mlxc1-sxc1): New option.
32177 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32179 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32180 __mips_no_lxc1_sxc1.
32181 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32182 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32183 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32185 2017-01-19 Richard Biener <rguenther@suse.de>
32187 PR tree-optimization/72488
32188 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32189 sure to restore SSA info.
32190 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32192 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32194 PR rtl-optimization/79121
32195 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32196 of the inner type when shifting an extended value.
32198 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32201 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32202 interposable aliases.
32204 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32207 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32208 Use the evmergelohi instruction.
32209 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32210 (mov_sitf_e500_subreg8_2_be): Likewise.
32211 (mov_sitf_e500_subreg12_2_le): Likewise.
32212 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32213 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32214 (mov_sitf_e500_subreg8_2_le): Likewise.
32215 (mov_sitf_e500_subreg12_2_be): Likewise.
32217 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32219 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32220 attribute from vecsimple to vecperm.
32221 (altivec_vbpermq2): Likewise.
32223 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32226 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32228 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32229 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32230 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32231 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32232 case where N arg is SIZE_MAX.
32233 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32234 (cmpstrsi): Add pattern.
32236 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32238 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32239 __builtin_vec_revb builtins.
32240 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32241 built-in functions to support generation of the ISA 3.0 XXBR<x>
32242 vector byte reverse instructions.
32243 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32244 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32245 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32246 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32247 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32248 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32249 (P9V_BUILTIN_VEC_REVB): Likewise.
32250 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32251 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32252 (p9_xxbrq_v16qi): Likewise.
32253 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32254 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32255 (p9_xxbrh_v8hi): Likewise.
32256 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32257 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32258 vec_revb built-in functions.
32260 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32262 PR rtl-optimization/78952
32263 * config/i386/i386.md (any_extract): New code iterator.
32264 (*insvqi_2): Use any_extract for source operand.
32265 (*insvqi_3): Use any_shiftrt for source operand.
32267 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32269 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32271 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32273 2017-01-18 Matthias Klose <doko@ubuntu.com>
32275 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32277 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32279 * config/rs6000/altivec.h (vec_bperm): Change #define.
32280 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32281 (altivec_vbpermq2): New define_insn.
32282 (altivec_vbpermd): Likewise.
32283 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32284 function interface.
32285 (VBPERMD): Likewise.
32286 (VBPERM): New polymorphic function interface.
32287 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32288 Add entries for P9V_BUILTIN_VEC_VBPERM.
32289 * doc/extend.texi: Add interfaces for vec_bperm.
32291 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32293 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32294 first letter of error messages.
32295 (s390_resolve_overloaded_builtin): Likewise.
32296 * config/s390/s390.c (s390_expand_builtin): Likewise.
32297 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32298 (s390_valid_target_attribute_inner_p): Likewise.
32299 * config/s390/s390.md ("tabort"): Likewise.
32301 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32303 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32304 (ISA_AVOID_DIV_HILO): New macro.
32305 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32306 (ISA_HAS_DDIV): Likewise.
32308 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32310 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32312 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32314 * doc/invoke.texi (fabi-version): Spelling fix.
32316 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32319 * doc/invoke.texi (fabi-version): Mention mangling fix for
32322 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32325 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32327 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32330 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32331 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32333 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32335 (stack_protect_test): Ditto.
32336 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32337 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32339 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32340 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32341 -mstack-protector-guard-offset=.
32342 (RS/6000 and PowerPC Options): Ditto.
32344 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32346 * config/i386/i386.h (MASK_CLASS_P): New define.
32347 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32348 there are no registers from different register sets also when
32349 mask registers are used. Update function comment.
32350 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32351 to (*k/*r) and (*k/*km) alternatives.
32353 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32355 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32356 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32357 (EH_RETURN_HANDLER_RTX): New define.
32358 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32359 Force frame pointer in EH return functions.
32360 (aarch64_expand_epilogue): Add barrier for eh_return.
32361 (aarch64_final_eh_return_addr): Remove.
32362 (aarch64_eh_return_handler_rtx): New function.
32363 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32365 (aarch64_eh_return_handler_rtx): New prototype.
32367 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32369 * config/rs6000/altivec.h (vec_rlmi): New #define.
32370 (vec_vrlnm): Likewise.
32371 (vec_rlnm): Likewise.
32372 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32373 (UNSPEC_VRLNM): Likewise.
32374 (VIlong): New mode iterator.
32375 (altivec_vrl<VI_char>mi): New define_insn.
32376 (altivec_vrl<VI_char>nm): Likewise.
32377 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32379 (VRLDNM): Likewise.
32380 (RLNM): New polymorphic function entry.
32381 (VRLWMI): New monomorphic function entry.
32382 (VRLDMI): Likewise.
32383 (RLMI): New polymorphic function entry.
32384 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32385 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32386 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32389 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32392 * dwarf2out.c (field_byte_offset): Restore the
32393 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32394 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32395 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32398 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32401 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32403 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32406 * dwarf2out.c (add_data_member_location_attribute): For constant
32407 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32408 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32409 DW_AT_byte_size attributes.
32411 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32413 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32414 after forcing to constant memory when the code model is medium.
32416 2017-01-17 Julia Koval <julia.koval@intel.com>
32419 * config/i386/avx512fintrin.h
32420 (_mm512_i32gather_ps): Change __addr type to void const*.
32421 (_mm512_mask_i32gather_ps): Ditto.
32422 (_mm512_i32gather_pd): Ditto.
32423 (_mm512_mask_i32gather_pd): Ditto.
32424 (_mm512_i64gather_ps): Ditto.
32425 (_mm512_mask_i64gather_ps): Ditto.
32426 (_mm512_i64gather_pd): Ditto.
32427 (_mm512_mask_i64gather_pd): Ditto.
32428 (_mm512_i32gather_epi32): Ditto.
32429 (_mm512_mask_i32gather_epi32): Ditto.
32430 (_mm512_i32gather_epi64): Ditto.
32431 (_mm512_mask_i32gather_epi64): Ditto.
32432 (_mm512_i64gather_epi32): Ditto.
32433 (_mm512_mask_i64gather_epi32): Ditto.
32434 (_mm512_i64gather_epi64): Ditto.
32435 (_mm512_mask_i64gather_epi64): Ditto.
32436 (_mm512_i32scatter_ps): Change __addr type to void*.
32437 (_mm512_mask_i32scatter_ps): Ditto.
32438 (_mm512_i32scatter_pd): Ditto.
32439 (_mm512_mask_i32scatter_pd): Ditto.
32440 (_mm512_i64scatter_ps): Ditto.
32441 (_mm512_mask_i64scatter_ps): Ditto.
32442 (_mm512_i64scatter_pd): Ditto.
32443 (_mm512_mask_i64scatter_pd): Ditto.
32444 (_mm512_i32scatter_epi32): Ditto.
32445 (_mm512_mask_i32scatter_epi32): Ditto.
32446 (_mm512_i32scatter_epi64): Ditto.
32447 (_mm512_mask_i32scatter_epi64): Ditto.
32448 (_mm512_i64scatter_epi32): Ditto.
32449 (_mm512_mask_i64scatter_epi32): Ditto.
32450 (_mm512_i64scatter_epi64): Ditto.
32451 (_mm512_mask_i64scatter_epi64): Ditto.
32452 * config/i386/avx512pfintrin.h
32453 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32454 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32455 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32456 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32457 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32458 (_mm512_prefetch_i32scatter_ps): Ditto.
32459 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32460 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32461 (_mm512_prefetch_i64scatter_pd): Ditto.
32462 (_mm512_prefetch_i64scatter_ps): Ditto.
32463 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32464 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32465 * config/i386/avx512vlintrin.h
32466 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32467 (_mm_mmask_i32gather_ps): Ditto.
32468 (_mm256_mmask_i32gather_pd): Ditto.
32469 (_mm_mmask_i32gather_pd): Ditto.
32470 (_mm256_mmask_i64gather_ps): Ditto.
32471 (_mm_mmask_i64gather_ps): Ditto.
32472 (_mm256_mmask_i64gather_pd): Ditto.
32473 (_mm_mmask_i64gather_pd): Ditto.
32474 (_mm256_mmask_i32gather_epi32): Ditto.
32475 (_mm_mmask_i32gather_epi32): Ditto.
32476 (_mm256_mmask_i32gather_epi64): Ditto.
32477 (_mm_mmask_i32gather_epi64): Ditto.
32478 (_mm256_mmask_i64gather_epi32): Ditto.
32479 (_mm_mmask_i64gather_epi32): Ditto.
32480 (_mm256_mmask_i64gather_epi64): Ditto.
32481 (_mm_mmask_i64gather_epi64): Ditto.
32482 (_mm256_i32scatter_ps): Change __addr type to void*.
32483 (_mm256_mask_i32scatter_ps): Ditto.
32484 (_mm_i32scatter_ps): Ditto.
32485 (_mm_mask_i32scatter_ps): Ditto.
32486 (_mm256_i32scatter_pd): Ditto.
32487 (_mm256_mask_i32scatter_pd): Ditto.
32488 (_mm_i32scatter_pd): Ditto.
32489 (_mm_mask_i32scatter_pd): Ditto.
32490 (_mm256_i64scatter_ps): Ditto.
32491 (_mm256_mask_i64scatter_ps): Ditto.
32492 (_mm_i64scatter_ps): Ditto.
32493 (_mm_mask_i64scatter_ps): Ditto.
32494 (_mm256_i64scatter_pd): Ditto.
32495 (_mm256_mask_i64scatter_pd): Ditto.
32496 (_mm_i64scatter_pd): Ditto.
32497 (_mm_mask_i64scatter_pd): Ditto.
32498 (_mm256_i32scatter_epi32): Ditto.
32499 (_mm256_mask_i32scatter_epi32): Ditto.
32500 (_mm_i32scatter_epi32): Ditto.
32501 (_mm_mask_i32scatter_epi32): Ditto.
32502 (_mm256_i32scatter_epi64): Ditto.
32503 (_mm256_mask_i32scatter_epi64): Ditto.
32504 (_mm_i32scatter_epi64): Ditto.
32505 (_mm_mask_i32scatter_epi64): Ditto.
32506 (_mm256_i64scatter_epi32): Ditto.
32507 (_mm256_mask_i64scatter_epi32): Ditto.
32508 (_mm_i64scatter_epi32): Ditto.
32509 (_mm_mask_i64scatter_epi32): Ditto.
32510 (_mm256_i64scatter_epi64): Ditto.
32511 (_mm256_mask_i64scatter_epi64): Ditto.
32512 (_mm_i64scatter_epi64): Ditto.
32513 (_mm_mask_i64scatter_epi64): Ditto.
32514 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32515 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32516 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32517 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32518 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32519 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32520 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32521 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32522 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32523 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32524 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32525 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32526 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32527 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32528 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32529 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32530 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32531 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32532 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32533 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32534 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32535 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32536 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32537 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32538 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32539 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32540 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32541 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32542 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32543 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32544 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32545 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32546 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32547 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32548 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32549 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32550 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32551 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32552 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32553 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32554 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32555 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32556 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32557 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32558 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32559 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32560 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32561 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32562 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32563 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32564 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32565 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32566 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32567 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32568 definitions accordingly.
32570 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32571 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32574 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32577 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32580 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32581 hard regs for inner regno.
32583 2017-01-17 Martin Liska <mliska@suse.cz>
32586 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32587 assumption and add comment.
32589 2017-01-17 Nathan Sidwell <nathan@acm.org>
32591 * ipa-visibility.c (localize_node): New function, broken out of ...
32592 (function_and_variable_visibility): ... here. Call it.
32594 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32596 PR middle-end/77445
32597 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32598 correctly set frequency of oudgoing edge.
32599 (duplicate_thread_path): Fix profile updating.
32601 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32604 * configure.ac: Add GCC_BASE_VER.
32605 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
32606 version from BASE-VER file.
32607 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
32608 (gcc.o): Depend on $(BASEVER).
32609 * common.opt (dumpfullversion): New option.
32610 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
32611 * doc/invoke.texi: Document -dumpfullversion.
32612 * doc/install.texi: Document --with-gcc-major-version-only.
32613 * configure: Regenerated.
32615 2017-01-17 Richard Biener <rguenther@suse.de>
32617 PR tree-optimization/71433
32618 * tree-vrp.c (register_new_assert_for): Merge same asserts
32619 on all incoming edges.
32620 (process_assert_insertions_for): Handle insertions at the
32623 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32625 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32626 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32628 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32631 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32634 2017-01-17 Alan Modra <amodra@gmail.com>
32637 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32638 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32639 symbolic stack limit when pic.
32641 2017-01-16 Martin Sebor <msebor@redhat.com>
32643 PR tree-optimization/78608
32644 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32646 2017-01-16 Jeff Law <law@redhat.com>
32649 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32650 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32651 for several include directories that may be relative to sysroot.
32652 * config/i386/x-mingw32 (gplus_includedir): Define.
32653 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32654 (native_system_includedir): Likewise.
32655 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32656 override if TARGET_SYSTEM_ROOT is defined.
32657 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32659 PR tree-optimization/79090
32660 PR tree-optimization/33562
32661 PR tree-optimization/61912
32662 PR tree-optimization/77485
32663 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32664 and computed trims into the dump file.
32666 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32668 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32670 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32673 * gimplify.c (gimplify_init_constructor): If want_value and
32674 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32678 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32679 sequence. Formatting fixes.
32680 (doloop_optimize): Formatting fixes.
32683 * gcc.c (debug_level_greater_than_spec_func): New function.
32684 (static_spec_functions): Add debug-level-gt spec function.
32685 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32687 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32688 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32689 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32690 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32691 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32692 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32694 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32696 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32697 QImode fixups to general and mask registers only.
32699 2017-01-16 Carl Love <cel@us.ibm.com>
32701 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32702 for built-in functions
32703 vector signed char vec_nabs (vector signed char)
32704 vector signed short vec_nabs (vector signed short)
32705 vector signed int vec_nabs (vector signed int)
32706 vector signed long long vec_nabs (vector signed long long)
32707 vector float vec_nabs (vector float)
32708 vector double vec_nabs (vector double)
32709 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32711 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32712 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32713 * doc/extend.texi: Update the documentation file for the new built-in
32716 2017-01-16 Martin Sebor <msebor@redhat.com>
32718 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32721 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32723 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32724 UNSPEC_VSX__XXSPLTD to require special splat handling.
32726 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32729 * system.h: Poison strndup.
32731 2017-01-16 Alan Modra <amodra@gmail.com>
32734 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32737 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32739 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32741 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32743 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32744 call recog here. Assert that INSN_CODE (insn) is non-negative.
32746 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32749 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32751 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32752 in the currently scheduled RTL fragment.
32754 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32756 PR rtl-optimization/78751
32757 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32760 2017-01-14 Jeff Law <law@redhat.com>
32762 PR tree-optimization/79090
32763 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32764 variable length stores.
32765 (compute_trims): Delete dead assignment to *trim_tail.
32766 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32769 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32771 PR rtl-optimization/78626
32772 PR rtl-optimization/78727
32773 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32774 of a block, and split such blocks after everything else is finished.
32776 2017-01-14 Alan Modra <amodra@gmail.com>
32779 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32780 target legitimate_combined_insn.
32781 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32782 (rs6000_legitimate_combined_insn): New function.
32783 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32785 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32786 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32787 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32789 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32791 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32793 2017-01-13 Jeff Law <law@redhat.com>
32795 PR tree-optimization/33562
32796 PR tree-optimization/61912
32797 PR tree-optimization/77485
32798 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32800 (delete_dead_assignment): Likewise.
32801 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32802 statement to delete_dead_call and delete_dead_assignment.
32804 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32807 * substring-locations.c (format_warning_va): Strengthen case 1 so
32808 that both endpoints of the substring must be within the format
32809 range for just the substring to be printed.
32811 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32813 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32814 * config/i386/i386.c (ix86_target_string): Add missing options
32815 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32816 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32817 flags_other and ix86_target_other to flags2_other. Display unknown
32819 (ix86_valid_target_attribute_inner_p): Add missing options and
32820 reorder options by implied ISAs, as in ix86_target_string.
32822 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32824 * hash-table.h (hash_table::too_empty_p): New function.
32825 (hash_table::expand): Use it.
32826 (hash_table::traverse): Likewise.
32827 (hash_table::empty_slot): Use sizeof (value_type) instead of
32828 sizeof (PTR) to convert bytes to elements. Shrink the table
32829 if the current size is excessive for the current number of
32832 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32834 * ira-costs.c (record_reg_classes): Break from the inner loop
32835 early once alt_fail is known to be true. Update outer loop
32836 handling accordingly.
32838 2017-01-13 Jeff Law <law@redhat.com>
32840 * tree-ssa-dse.c (decrement_count): New function.
32841 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32842 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32843 when we know the partially dead statement is a mem* function.
32845 PR tree-optimization/61912
32846 PR tree-optimization/77485
32847 * tree-ssa-dse.c: Include expr.h.
32848 (maybe_trim_constructor_store): New function.
32849 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32851 PR tree-optimization/33562
32852 PR tree-optimization/61912
32853 PR tree-optimization/77485
32854 * doc/invoke.texi: Document new dse-max-object-size param.
32855 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32856 * tree-ssa-dse.c: Include params.h.
32857 (dse_store_status): New enum.
32858 (initialize_ao_ref_for_dse): New, partially extracted from
32860 (valid_ao_ref_for_dse, normalize_ref): New.
32861 (setup_live_bytes_from_ref, compute_trims): Likewise.
32862 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32863 (maybe_trim_partially_dead_store): Likewise.
32864 (maybe_trim_complex_store): Likewise.
32865 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32866 Track what bytes live from the original store. Return tri-state
32867 for dead, partially dead or live.
32868 (dse_dom_walker): Add constructor, destructor and new private members.
32869 (delete_dead_call, delete_dead_assignment): New extracted from
32871 (dse_optimize_stmt): Make a member of dse_dom_walker.
32872 Use initialize_ao_ref_for_dse.
32874 PR tree-optimization/33562
32875 PR tree-optimization/61912
32876 PR tree-optimization/77485
32877 * sbitmap.h (bitmap_count_bits): Prototype.
32878 (bitmap_clear_range, bitmap_set_range): Likewise.
32879 * sbitmap.c (bitmap_clear_range): New function.
32880 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32882 2017-01-13 Martin Liska <mliska@suse.cz>
32885 * function.c (set_cfun): Add new argument force.
32886 * function.h (set_cfun): Likewise.
32887 * ipa-inline-transform.c (inline_call): Use the function when
32888 strict alising from is dropped for function we inline to.
32890 2017-01-13 Richard Biener <rguenther@suse.de>
32892 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32893 for dumping GIMPLE INTEGER_CSTs.
32895 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32897 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32898 to 201112L since C++17.
32900 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32903 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32904 if -fsanitize=kernel-address is present.
32906 2017-01-13 Richard Biener <rguenther@suse.de>
32908 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32909 as _Literal ( type ) number in case usual suffixes do not
32910 preserve all information.
32912 2017-01-13 Richard Biener <rguenther@suse.de>
32914 PR tree-optimization/77283
32915 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
32916 and ssa-iterators.h.
32917 (is_feasible_trace): Implement a cost model based on joiner
32920 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
32923 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
32924 char or short to __float128/_Float128 directly.
32926 2017-01-12 Martin Sebor <msebor@redhat.com>
32928 to -Wformat-overflow.
32929 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
32930 (min_bytes_remaining): Same.
32931 (get_string_length): Same.
32932 (format_string): Same.
32933 (format_directive): Same.
32935 (pass_sprintf_length::handle_gimple_call): Same.
32937 2017-01-12 Jakub Jelinek <jakub@redhat.com>
32939 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
32940 info.nowrite calls with no lhs that can't throw. Return bool
32941 whether gsi_remove has been called or not.
32942 (pass_sprintf_length::handle_gimple_call): Return bool whether
32943 try_substitute_return_value called gsi_remove. Formatting fix.
32944 (pass_sprintf_length::execute): Don't use gsi_remove if
32945 handle_gimple_call returned true.
32948 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
32949 be removed due to side-effects, don't remove following barrier nor
32950 turn the successor edge into fallthru edge.
32952 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32955 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
32956 element-reversing loads and stores as not swappable.
32958 2017-01-12 Nathan Sidwell <nathan@acm.org>
32959 Nicolai Stange <nicstange@gmail.com>
32961 * combine.c (try_combine): Don't ignore result of overlap checking
32962 loop. Combine overlap & asm check into single loop.
32964 2017-01-12 Richard Biener <rguenther@suse.de>
32966 * tree-pretty-print.c (dump_generic_node): Provide -gimple
32967 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
32969 2017-01-12 Richard Biener <rguenther@suse.de>
32971 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
32972 and TS_TARGET_OPTION directly derive from TS_BASE.
32973 * tree-core.h (tree_optimization_option): Derive from tree_base.
32974 (tree_target_option): Likewise.
32976 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
32978 * config/i386/i386.c (memory_address_length): Increase len
32979 only when rip_relative_addr_p returns false.
32981 2017-01-11 Julia Koval <julia.koval@intel.com>
32983 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
32984 (OPTION_MASK_ISA_SGX_SET): New.
32985 (ix86_handle_option): Handle OPT_msgx.
32986 * config.gcc: Added sgxintrin.h.
32987 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
32988 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
32989 * config/i386/i386.c (ix86_target_string): Add -msgx.
32991 (ix86_option_override_internal): Handle new options.
32992 (ix86_valid_target_attribute_inner_p): Add sgx.
32993 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
32994 * config/i386/i386.opt: Add msgx.
32995 * config/i386/sgxintrin.h: New file.
32996 * config/i386/x86intrin.h: Add sgxintrin.h.
32998 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33001 * fold-const.c (maybe_nonzero_address): Return 1 for function
33003 (tree_single_nonzero_warnv_p): Don't handle function local objects
33007 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33010 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33013 * opts.c: Include "spellcheck.h"
33014 (struct string_fragment): New struct.
33015 (struct edit_distance_traits<const string_fragment &>): New
33017 (get_closest_sanitizer_option): New function.
33018 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33020 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33022 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33024 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33025 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33026 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33027 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33028 for initial die_offset if dwarf_split_debug_info.
33029 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33030 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33032 (output_skeleton_debug_sections): Formatting fix. Use
33033 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33034 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33036 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33038 * config/arm/cortex-a53.md: Add bypasses for
33039 cortex_a53_r2f_cvt.
33040 (cortex_a53_r2f): Only use for transfers.
33041 (cortex_a53_f2r): Likewise.
33042 (cortex_a53_r2f_cvt): Add reservation for conversions.
33043 (cortex_a53_f2r_cvt): Likewise.
33045 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33047 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33048 to all inlined functions, change static to extern.
33050 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33053 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33055 (arm_assemble_integer): Likewise.
33057 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33059 * config.gcc: Use new awk script to check CPU, FPU and architecture
33060 parameters for --with-... options.
33061 * config/arm/parsecpu.awk: New file
33062 * config/arm/arm-cpus.in: New file.
33063 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33065 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33067 * config/arm/t-arm: Update dependency rules.
33068 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33069 of processing .def files.
33070 * config/arm/genopt.sh: Deleted.
33071 * config/arm/gentune.sh: Deleted.
33072 * config/arm/arm-cores.def: Deleted.
33073 * config/arm/arm-arches.def: Deleted.
33074 * config/arm/arm-fpus.def: Deleted.
33075 * config/arm/arm-tune.md: Regenerated.
33076 * config/arm/arm-tables.opt: Regenerated.
33077 * config/arm/arm-cpu.h: New generated file.
33078 * config/arm/arm-cpu-data.h: New generated file.
33079 * config/arm/arm-cpu-cdata.h: New generated file.
33081 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33084 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33086 (input_varpool_node): Unpack dynamically_initialized bit.
33088 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33090 PR rtl-optimization/79032
33091 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33092 the alignment of the adjusted memory reference against that of MODE,
33093 instead of the alignment of the original memory reference.
33095 2017-01-11 Martin Jambor <mjambor@suse.cz>
33097 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
33099 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
33100 decorated functions.
33102 2017-01-11 Richard Biener <rguenther@suse.de>
33104 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
33105 set range/nonnull info for PHI results. Do not set it on
33106 stmts marked for removal.
33108 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
33110 * expr.c (store_field): In the bitfield case, fetch the return value
33111 from the registers before applying a single big-endian adjustment.
33112 Always do a final load for a BLKmode value not larger than a word.
33114 2017-01-10 David Malcolm <dmalcolm@redhat.com>
33117 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33118 that we correctly handle column numbers greater than
33119 LINE_MAP_MAX_COLUMN_NUMBER.
33121 2017-01-10 Martin Sebor <msebor@redhat.com>
33123 PR middle-end/78245
33124 * gimple-ssa-sprintf.c (get_destination_size): Call
33125 {init,fini}object_sizes.
33126 * tree-object-size.c (addr_object_size): Adjust.
33127 (pass_through_call): Adjust.
33128 (pass_object_sizes::execute): Adjust.
33129 * tree-object-size.h (fini_object_sizes): Declare.
33131 2017-01-10 Martin Sebor <msebor@redhat.com>
33133 PR tree-optimization/78775
33134 * builtins.c (get_size_range): Move...
33135 * calls.c: ...to here.
33136 (alloc_max_size): Accept zero argument.
33137 (operand_signed_p): Remove.
33138 (maybe_warn_alloc_args_overflow): Call get_size_range.
33139 * calls.h (get_size_range): Declare.
33141 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
33143 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
33144 from TI's devices.csv file as of September 2016.
33145 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
33147 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
33149 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
33150 * doc/invoke.texi: Likewise.
33151 * doc/md.texi: Likewise.
33152 * doc/objc.texi: Likewise.
33154 2017-01-10 Joshua Conner <joshconner@google.com>
33156 * config/arm/fuchsia-elf.h: New file.
33157 * config/fuchsia.h: New file.
33158 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
33159 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
33161 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33163 2016-01-10 Richard Biener <rguenther@suse.de>
33165 PR tree-optimization/79034
33166 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33167 Propagate out degenerate PHIs in the joiner.
33169 2017-01-10 Martin Liska <mliska@suse.cz>
33171 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33172 (sort_congruence_classes_by_decl_uid): Likewise.
33173 (sort_congruence_class_groups_by_decl_uid): Likewise.
33174 (sem_item_optimizer::merge_classes): Sort class, groups in these
33175 classes and members in the groups by DECL_UID of declarations.
33176 This would make merge operations stable.
33178 2017-01-10 Martin Liska <mliska@suse.cz>
33180 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33181 usage of m_classes_vec.
33182 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33183 (sem_item_optimizer::get_group_by_hash): Likewise.
33184 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33185 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33186 (sem_item_optimizer::verify_classes): Likewise.
33187 (sem_item_optimizer::process_cong_reduction): Likewise.
33188 (sem_item_optimizer::dump_cong_classes): Likewise.
33189 (sem_item_optimizer::merge_classes): Likewise.
33190 * ipa-icf.h (congruence_class_hash): Rename from
33191 congruence_class_group_hash. Remove declaration of m_classes_vec.
33193 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33195 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33196 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33197 * config.gcc: Add avx512vpopcntdqintrin.h.
33198 * config/i386/avx512vpopcntdqintrin.h: New.
33199 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33200 * config/i386/i386-builtin-types.def: Add new types.
33201 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33202 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33203 __builtin_ia32_vpopcountq_v8di_mask): New.
33204 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33205 __AVX512VPOPCNTDQ__.
33206 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33207 (PTA_AVX512VPOPCNTDQ): Define.
33208 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33209 TARGET_AVX512VPOPCNTDQ_P): Define.
33210 * config/i386/i386.opt: Add mavx512vpopcntdq.
33211 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33212 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33214 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33216 PR middle-end/77484
33217 * predict.def (PRED_CALL): Set to 67.
33219 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33221 * expr.c (store_field): In the bitfield case, if the value comes from
33222 a function call and is of an aggregate type returned in registers, do
33223 not modify the field mode; extract the value in all cases if the mode
33224 is BLKmode and the size is not larger than a word.
33226 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33229 * config/i386/cpuid.h: Fix undefined behavior.
33231 2017-01-04 Jeff Law <law@redhat.com>
33233 PR tree-optimization/79007
33234 PR tree-optimization/67955
33235 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33236 conservative for pt.null when flag_non_call_exceptions is on.
33238 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33240 PR translation/79019
33241 PR translation/79020
33242 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33243 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33244 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33246 * config/avr/avr.opt (maccumulate-args): Likewise.
33247 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33248 * common.opt (freport-bug): Likewise.
33249 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33250 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33251 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33252 translatable string.
33253 * config/i386/i386.c (function_value_32): Likewise.
33254 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33255 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33257 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33258 * common/config/msp430/msp430-common.c (msp430_handle_option):
33260 * symtab.c (symtab_node::verify_base): Likewise.
33261 * opts.c (set_debug_level): Likewise.
33262 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33263 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33264 missing whitespace to translatable strings.
33265 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33266 * config/sh/superh.h: Likewise.
33267 * config/i386/xopintrin.h: Likewise.
33268 * config/i386/znver1.md: Likewise.
33269 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33270 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33271 * double-int.h (struct double_int): Likewise.
33272 * double-int.c (div_and_round_double): Likewise.
33273 * wide-int.cc: Likewise.
33274 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33275 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33276 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33277 (crossjumps_occurred): ... this.
33278 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33281 PR tree-optimization/78899
33282 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33283 returning bool return struct loop *, NULL for failure and the new
33285 (versionable_outer_loop_p): Don't version outer loop if it has
33286 dont_vectorized bit set.
33287 (tree_if_conversion): When versioning outer loop, ensure
33288 tree_if_conversion is performed also on the inner loop of the
33289 non-vectorizable outer loop copy.
33290 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33291 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33292 prevent vectorization of it.
33293 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33294 the outer loop vectorization of the non-scalar version is attempted
33295 before vectorization of the inner loop in scalar version. If
33296 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33297 vectorization of its inner loop.
33298 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33299 has 2 inner loops, rename also on edges from bb whose single pred
33300 is outer_loop->header. Fix typo in function comment.
33302 2017-01-09 Martin Sebor <msebor@redhat.com>
33305 * asan.c (asan_emit_stack_protection): Increase local buffer size
33306 to avoid snprintf truncation warning.
33308 2017-01-09 Andrew Pinski <apinski@cavium.com>
33310 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33311 to reference thunderx2t99 for the tuning structure
33312 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33314 (thunderx2t99_extra_costs): This.
33315 * config/aarch64/aarch64-tune.md: Regenerate.
33316 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33317 (vulcan_addrcost_table): This.
33318 (vulcan_regmove_cost): Rename to ...
33319 (thunderx2t99_regmove_cost): This.
33320 (vulcan_vector_cost): Rename to ...
33321 (thunderx2t99_vector_cost): this.
33322 (vulcan_branch_cost): Rename to ...
33323 (thunderx2t99_branch_cost): This.
33324 (vulcan_tunings): Rename to ...
33325 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33326 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33328 2017-01-09 Martin Jambor <mjambor@suse.cz>
33332 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33333 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33334 (propagate_vr_accross_jump_function): Use the above function for all
33335 value range computations for pass-through jump functions and type
33336 converasion from explicit value range values.
33337 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33338 parameters from TYPE_ARG_TYPES.
33339 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33340 (ipa_write_node_info): Stream type of the actual argument.
33341 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33343 2017-01-09 Martin Liska <mliska@suse.cz>
33346 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33347 (lookup_compiler): Do not show error message with have_E.
33349 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33351 PR tree-optimization/78938
33352 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33353 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33354 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33355 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33358 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33360 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33363 2017-01-09 Richard Biener <rguenther@suse.de>
33365 PR tree-optimization/78997
33366 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33367 name condition properly.
33369 2017-01-09 Richard Biener <rguenther@suse.de>
33372 * dwarf2out.c (is_cxx): New overload with context.
33373 (is_naming_typedef_decl): Use it.
33375 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33377 * invoke.texi (Option Summary): Correct spacing in option lists
33378 and add line breaks to fix over-long lines.
33380 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33382 PR middle-end/17660
33384 * extend.texi (Common Variable Attributes): Add xref to GCC
33385 Internals manual to explain mode attribute keywords.
33387 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33390 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33391 and Preprocessor Options.
33392 (Options for Linking): Document -pthread here....
33393 (RS/6000 and PowerPC Options): ...not here.
33394 (Solaris 2 Options): ...or here.
33395 * doc/cppopts.texi: Document -pthread.
33397 2017-01-08 Martin Sebor <msebor@redhat.com>
33399 PR middle-end/77708
33400 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33401 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33402 New member functions.
33403 (format_directive): Used them.
33405 (pass_sprintf_length::handle_gimple_call): Same.
33406 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33407 to avoid truncation for any argument.
33408 (extract_affine_mul): Same.
33409 * tree.c (get_file_function_name): Same.
33411 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33413 PR middle-end/77484
33414 * predict.def (PRED_INDIR_CALL): Set to 86.
33416 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33418 PR preprocessor/54124
33419 * doc/cppopts.texi: Reformat -d subtable to list the full name
33420 of the options. Add cross-reference to the docs for the general
33421 compiler -d options.
33422 * doc/invoke.texi (Developer Options): Add cross-reference to the
33423 preprocessor-specific -d option documentation.
33425 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33427 PR preprocessor/13498
33428 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33429 redudant material, and reflect new command-line options.
33430 (System Headers): Likewise.
33432 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33434 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33435 -isystem, and -idirafter. Copy-edit.
33436 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33437 default for -ftrack-macro-expansion. Delete obsolete and
33438 badly-formatted implementation details about -fdebug-cpp output.
33439 * doc/cppwarnopts.texi: Copy-edit.
33441 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33444 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33445 that the transition from a max line width >= 1<<10 to narrower
33446 lines works correctly.
33448 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33450 * doc/options.texi (PerFunction): New.
33451 * opt-functions.awk (switch_flags): Map both Optimization and
33452 PerFunction to CL_OPTIMIZATION.
33453 * opth-gen.awk: Test for PerFunction flag along with
33455 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33456 it only when the latter is present. Skip those that don't in
33457 the hash function generator.
33458 * common.opt (fvar-tracking): Mark as PerFunction instead of
33460 (fvar-tracking-assignments): Likewise.
33461 (fvar-tracking-assignments-toggle): Likewise.
33462 (fvar-tracking-uninit): Likewise.
33464 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33466 PR translation/79018
33467 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33470 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33473 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33474 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33475 TARGET_LONG_JUMP_TABLE_OFFSETS.
33476 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33477 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33478 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33479 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33480 * config/m68k/m68k.md (tablejump expander): Likewise.
33481 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33482 TARGET_LONG_JUMP_TABLE_OFFSETS.
33483 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33484 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33486 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33487 David Holsgrove <david.holsgrove@xilinx.com>
33489 * common/config/microblaze/microblaze-common.c
33490 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33491 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33493 * config/microblaze/microblaze.c (microblaze_must_save_register)
33494 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33496 (microblaze_eh_return): New function.
33497 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33498 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33499 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33500 * config/microblaze/microblaze.md (eh_return): New pattern.
33502 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33504 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33505 GCC_DIAGNOSTIC_STRINGIFY): Define.
33507 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33509 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33511 * config/arm/arm.md (<mcrr>): New.
33513 * config/arm/arm.c (arm_arch5te): New.
33514 (arm_option_override): Set arm_arch5te.
33515 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33517 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33518 (arm_mcrr_qualifiers): ... this. New.
33519 (MRRC_QUALIFIERS): Define to...
33520 (arm_mrrc_qualifiers): ... this. New.
33521 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33523 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33524 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33525 (MRRCI, mrrc, MRRC): New.
33526 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33527 VUNSPEC_MRRC2): New.
33529 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33531 * config/arm/arm.md (<mcr>): New.
33533 * config/arm/arm.c (arm_coproc_builtin_available): Add
33534 support for mcr, mrc, mcr2 and mrc2.
33535 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33536 (arm_mcr_qualifiers): ... this. New.
33537 (MRC_QUALIFIERS): Define to ...
33538 (arm_mrc_qualifiers): ... this. New.
33539 (MCR_QUALIFIERS): Define to ...
33540 (arm_mcr_qualifiers): ... this. New.
33541 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33543 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33544 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33545 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33546 VUNSPEC_MRC2): New.
33548 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33550 * config/arm/arm.md (*ldc): New.
33554 * config/arm/arm.c (arm_coproc_builtin_available): Add
33555 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33556 (arm_coproc_ldc_stc_legitimate_address): New.
33557 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33558 'qualifier_const_pointer'.
33559 (LDC_QUALIFIERS): Define to...
33560 (arm_ldc_qualifiers): ... this. New.
33561 (STC_QUALIFIERS): Define to...
33562 (arm_stc_qualifiers): ... this. New.
33563 * config/arm/arm-protos.h
33564 (arm_coproc_ldc_stc_legitimate_address): New.
33565 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33566 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33567 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33568 stc2, stcl, stc2l): New.
33569 * config/arm/constraints.md (Uz): New.
33570 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33571 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33572 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33573 VUNSPEC_STC2L): New.
33575 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33577 * config/arm/arm.md (<cdp>): New.
33578 * config/arm/arm.c (neon_const_bounds): Rename this ...
33579 (arm_const_bounds): ... this.
33580 (arm_coproc_builtin_available): New.
33581 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33582 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33583 (CDP_QUALIFIERS): Define to...
33584 (arm_cdp_qualifiers): ... this. New.
33586 (arm_expand_builtin_args): Add case for 6 arguments.
33587 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33588 (arm_const_bounds): ... this.
33589 (arm_coproc_builtin_available): New.
33590 * config/arm/arm_acle.h (__arm_cdp): New.
33592 * config/arm/arm_acle_builtins.def (cdp): New.
33594 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33595 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33596 'arm_const_bounds'.
33597 * config/arm/types.md (coproc): New.
33598 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
33599 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
33600 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
33601 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
33603 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33605 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
33606 (UBINOP_QUALIFIERS): New.
33608 (acle_builtin_data): New. Change comment.
33609 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
33610 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
33611 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
33612 arm_acle_builtins.def.
33613 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
33614 (arm_init_acle_builtins): New.
33615 (CRC32_BUILTIN): Remove.
33616 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33617 crc32cb, crc32ch and crc32cw.
33618 (arm_init_crc32_builtins): Remove.
33619 (arm_init_builtins): Use arm_init_acle_builtins rather
33620 than arm_init_crc32_builtins.
33621 (arm_expand_acle_builtin): New.
33622 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33623 * config/arm/arm_acle_builtins.def: New.
33625 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33627 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33628 (arm_builtin_datum): ... this.
33629 (arm_init_neon_builtin): Rename to ...
33630 (arm_init_builtin): ... this. Add a new parameters PREFIX
33631 and USE_SIG_IN_NAME.
33632 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33633 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33634 'arm_builtin_datum'.
33635 (arm_init_vfp_builtins): Likewise.
33636 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33637 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33638 (arm_expand_neon_args): Rename to ...
33639 (arm_expand_builtin_args): ... this. Rename builtin_arg
33640 enum values and differentiate between ARG_BUILTIN_MEMORY
33641 and ARG_BUILTIN_NEON_MEMORY.
33642 (arm_expand_neon_builtin_1): Rename to ...
33643 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33644 values, arm_expand_builtin_args and add bool parameter NEON.
33645 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33646 (arm_expand_vfp_builtin): Likewise.
33647 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33649 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33651 PR middle-end/77484
33652 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33653 * predict.c (tree_estimate_probability_bb): Reverse direction of
33654 polymorphic call predictor.
33656 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33658 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33659 (determine_pass_name_match): ...this new function and...
33660 (should_skip_pass_p): ...this new function.
33662 2017-01-06 Nathan Sidwell <nathan@acm.org>
33664 * ipa-visibility.c (function_and_variable_visibility): Reformat
33665 comments and long lines. Remove extrneous if.
33666 * symtab.c (symtab_node::make_decl_local): Fix code format.
33667 (symtab_node::set_section_for_node): Fix comment typo.
33669 2017-01-06 Martin Liska <mliska@suse.cz>
33672 * lra-constraints.c: Rename invariant to lra_invariant.
33673 * predict.c (set_even_probabilities): Initialize e to NULL.
33675 2017-01-05 Martin Sebor <msebor@redhat.com>
33677 PR tree-optimization/78910
33678 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33679 (format_integer): Correct off-by-one error in the handling
33680 of precision with negative numbers in signed conversions..
33682 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33684 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33686 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33688 PR tree-optimization/71016
33689 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33690 factor_out_conditional_conversion. Formatting fix.
33691 (factor_out_conditional_conversion): Add cond_stmt argument.
33692 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33693 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33696 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33698 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33699 read-rtl-function.o, and selftest-rtl.o.
33700 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33701 (selftest::aarch64_test_loading_full_dump): New function.
33702 (selftest::aarch64_run_selftests): New function.
33703 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33704 selftest::aarch64_run_selftests.
33705 * config/i386/i386.c
33706 (selftest::ix86_test_loading_dump_fragment_1): New function.
33707 (selftest::ix86_test_loading_call_insn): New function.
33708 (selftest::ix86_test_loading_full_dump): New function.
33709 (selftest::ix86_test_loading_unspec): New function.
33710 (selftest::ix86_run_selftests): Call the new functions.
33711 * emit-rtl.c (maybe_set_max_label_num): New function.
33712 * emit-rtl.h (maybe_set_max_label_num): New decl.
33713 * function.c (instantiate_decls): Guard call to
33714 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33715 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33717 * gensupport.c (gen_reader::gen_reader): Pass "false"
33718 for new "compact" param of rtx_reader.
33719 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33720 rather than an empty string for NULL strings.
33721 * read-md.c: Potentially include config.h rather than bconfig.h.
33722 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33723 (have_error): New global, copied from errors.c.
33724 (md_reader::read_name): Rename to...
33725 (md_reader::read_name_1): ...this, adding "out_loc" param,
33726 and converting "missing name or number" to returning false, rather
33728 (md_reader::read_name): Reimplement in terms of read_name_1.
33729 (md_reader::read_name_or_nil): New function.
33730 (md_reader::read_string): Handle "(nil)" by returning NULL.
33731 (md_reader::md_reader): Add new param "compact".
33732 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33733 (md_reader::read_file): New method.
33734 * read-md.h (md_reader::md_reader): Add new param "compact".
33735 (md_reader::read_file): New method.
33736 (md_reader::is_compact): New accessor.
33737 (md_reader::read_name): Convert return type from void to file_location.
33738 (md_reader::read_name_or_nil): New decl.
33739 (md_reader::read_name_1): New decl.
33740 (md_reader::m_compact): New field.
33741 (noop_reader::noop_reader): Pass "false" for new "compact" param
33743 (rtx_reader::rtx_reader): Add new "compact" param.
33744 (rtx_reader::read_rtx_operand): Make virtual and convert return
33745 type from void to rtx.
33746 (rtx_reader::read_until): New decl.
33747 (rtx_reader::handle_any_trailing_information): New virtual function.
33748 (rtx_reader::postprocess): New virtual function.
33749 (rtx_reader::finalize_string): New virtual function.
33750 (rtx_reader::m_in_call_function_usage): New field.
33751 (rtx_reader::m_reuse_rtx_by_id): New field.
33752 * read-rtl-function.c: New file.
33753 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33754 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33755 (selftest::verify_three_block_rtl_cfg): New decl.
33756 * read-rtl-function.h: New file.
33757 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33758 For host, include function.h, memmodel.h, and emit-rtl.h.
33759 (one_time_initialization): New function.
33760 (struct compact_insn_name): New struct.
33761 (compact_insn_names): New array.
33762 (find_code): Handle insn codes in compact dumps.
33763 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33764 (bind_subst_iter_and_attr): Likewise.
33765 (add_condition_to_string): Likewise.
33766 (add_condition_to_rtx): Likewise.
33767 (apply_attribute_uses): Likewise.
33768 (add_current_iterators): Likewise.
33769 (apply_iterators): Likewise.
33770 (initialize_iterators): Guard usage of apply_subst_iterator with
33771 #ifdef GENERATOR_FILE.
33772 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33773 (md_reader::read_mapping): Likewise.
33774 (add_define_attr_for_define_subst): Likewise.
33775 (add_define_subst_attr): Likewise.
33776 (read_subst_mapping): Likewise.
33777 (check_code_iterator): Likewise.
33778 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33780 (one_time_initialization): New function.
33781 (rtx_reader::read_until): New method.
33782 (read_flags): New function.
33783 (parse_reg_note_name): New function.
33784 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33785 Handle reuse_rtx ids.
33786 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33787 Add parsing support for RTL dumps, mirroring the special-cases in
33788 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33789 values, and calling handle_any_trailing_information.
33790 (rtx_reader::read_rtx_operand): Convert return type from void
33791 to rtx, returning return_rtx. Handle case 'e'. Call
33792 finalize_string on XSTR and XTMPL fields.
33793 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33794 "(nil)" values were omitted. Call the postprocess vfunc on the
33796 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33797 class ctor. Initialize m_in_call_function_usage. Call
33798 one_time_initialization.
33799 * rtl-tests.c (selftest::test_uncond_jump): Call
33800 set_new_first_and_last_insn.
33801 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33802 * selftest-rtl.c: New file.
33803 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33804 (selftest::get_insn_by_uid): New decl.
33805 * selftest-run-tests.c (selftest::run_tests): Call
33806 read_rtl_function_c_tests.
33807 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33808 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33811 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33813 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33814 operands in a special way. Assert that pos+len <= mode precision.
33816 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33818 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33819 3 argument Alias with unlimited for the negative form.
33820 (fno-vect-cost-model): Removed.
33822 2017-01-05 Martin Liska <mliska@suse.cz>
33824 * hsa-gen.c (gen_hsa_divmod): New function.
33825 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33827 2017-01-05 Martin Liska <mliska@suse.cz>
33830 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33833 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33835 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33836 small constant length operands.
33838 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33840 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33841 between loop iterations.
33843 2017-01-05 Martin Liska <mliska@suse.cz>
33846 * gimplify.c (gimplify_decl_expr): Compare to
33847 asan_poisoned_variables instread of checking flags.
33848 (gimplify_target_expr): Likewise.
33849 (gimplify_expr): Likewise.
33850 (gimplify_function_tree): Conditionally initialize
33851 asan_poisoned_variables.
33853 2017-01-04 Jeff Law <law@redhat.com>
33855 PR tree-optimizatin/78812
33856 * rtl.h (contains_mem_rtx_p): Prototype.
33857 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33858 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33859 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33860 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33861 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33863 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33865 * input.c (assert_char_at_range): Default-initialize actual_range.
33867 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33869 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33870 to match the caller.
33872 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33874 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33875 insns after final jump in test to emit dummy move.
33877 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33879 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33880 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33882 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33884 * multiple_target.c (create_dispatcher_calls): Init e_next.
33885 * tree-ssa-loop-split.c (split_loop): Init border.
33886 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33889 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33894 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33895 (altivec_register_operand): Do not return true if the operand
33896 contains a SUBREG mixing SImode and SFmode.
33897 (vsx_register_operand): Likewise.
33898 (vsx_reg_sfsubreg_ok): New predicate.
33899 (vfloat_operand): Do not return true if the operand contains a
33900 SUBREG mixing SImode and SFmode.
33901 (vint_operand): Likewise.
33902 (vlogical_operand): Likewise.
33903 (gpc_reg_operand): Likewise.
33904 (int_reg_operand): Likewise.
33905 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33906 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33907 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33909 (rs6000_emit_move_si_sf_subreg): New helper function.
33910 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
33911 fixup SUBREGs involving SImode and SFmode.
33912 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
33913 numbers for the new peephole2 optimization.
33914 (peephole2 for SFmode unions): New peephole2 to optimize cases in
33915 the GLIBC math library that do AND/IOR/XOR operations on single
33916 precision floating point.
33917 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
33918 target macros to say whether we need to avoid SUBREGs mixing
33920 (TARGET_ALLOW_SF_SUBREG): Likewise.
33921 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
33922 (UNSPEC_SI_FROM_SF): Likewise.
33923 (iorxor): Change spacing.
33924 (and_ior_xor): New iterator for AND, IOR, and XOR.
33925 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
33926 (movdi_from_sf_zero_ext): Likewise.
33927 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
33928 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
33929 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
33930 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
33931 (fms<mode>4): Likewise.
33932 (fnma<mode>4): Likewise.
33933 (fnms<mode>4): Likewise.
33934 (nfma<mode>4): Likewise.
33935 (nfms<mode>4): Likewise.
33937 2017-01-04 Marek Polacek <polacek@redhat.com>
33940 * doc/invoke.texi: Document -Wpointer-compare.
33942 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33944 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
33947 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
33948 descriptions for -gdwarf-5 and emit them as uleb128 instead of
33951 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
33954 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
33955 documentation of the powerpc_popcntb_ok attribute.
33956 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33957 code to issue warning messages if a requested CPU configuration is
33958 not supported by the binary (assembler and loader) toolchain.
33959 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
33960 made to define a built-in function that has been disabled.
33961 (paired_init_builtins): Add assertion to prevent ICE if attempt is
33962 made to define a built-in function that has been disabled.
33963 (altivec_init_builtins): Add comment explaining why definition
33964 of the DST built-in functions is not preceded by an assertion
33965 check. Add assertions to prevent ICE if attempts are made to
33966 define an altivec predicate or an abs* built-in function that has
33968 (htm_init_builtins): Add comment explaining why definition of the
33969 htm built-in functions is not preceded by an assertion check.
33971 2017-01-04 Jeff Law <law@redhat.com>
33973 PR tree-optimizatin/67955
33974 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
33975 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
33976 the points-to solution does not include pt_null. Use DECL_PT_UID
33979 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
33981 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
33982 Use gen_int_mode instead of gen_lopwart for const_int operands.
33984 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33986 PR tree-optimization/71563
33987 * match.pd: Simplify X << Y into X if Y is known to be 0 or
33988 out of range value - has low bits known to be zero.
33990 2017-01-04 Alan Modra <amodra@gmail.com>
33992 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
33993 * configure: Regenerate.
33994 * config.in: Regenerate.
33996 2017-01-04 Jakub Jelinek <jakub@redhat.com>
33999 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34000 a substring of the message, but strcmp with the whole message. Ifdef
34001 ENABLE_NLS, translate the message first using dgettext.
34003 2017-01-03 Jeff Law <law@redhat.com>
34005 PR tree-optimizatin/78856
34006 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34007 (mark_threaded_blocks): Remove code to truncate thread paths that
34008 cross multiple loop headers. Instead invalidate the cached loop
34009 iteration information and handle case of a thread path walking
34010 into an irreducible region.
34012 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34015 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34016 assertions. Add support for doing the signbit if the IEEE 128-bit
34017 floating point value is in a GPR.
34018 * config/rs6000/rs6000.md (Fsignbit): Delete.
34019 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34020 Update the length attribute if the value is in a GPR.
34021 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34022 the sign or zero extension instruction, since the value is always 0/1.
34023 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34026 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34027 extracting SImode to a GPR register so that we can generate a
34028 store, limit the vector to be in a traditional Altivec register
34029 for the vextuwrx instruction.
34031 2017-01-03 Ian Lance Taylor <iant@google.com>
34033 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34035 2017-01-03 Martin Sebor <msebor@redhat.com>
34037 PR tree-optimization/78696
34038 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34039 precision. Use MPFR for %f for greater fidelity. Correct handling
34041 (pass_sprintf_length::compute_format_length): Set width and precision
34042 specified by asrerisk to void_node for vararg functions.
34043 (try_substitute_return_value): Adjust dump output.
34045 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34047 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34049 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34051 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34052 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34053 -mlra/-mno-lra was passed to the compiler.
34055 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34057 PR rtl-optimization/65618
34058 * emit-rtl.c (try_split): Move initialization of "before" and
34059 "after" to just before the call to emit_insn_after_setloc.
34061 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34063 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34065 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34067 * dwarf2out.c (gen_enumeration_type_die): When
34068 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34070 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34072 PR tree-optimization/78965
34073 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34074 Change first argument from const call_info & to call_info &. For %n
34075 set info.nowrite to false.
34077 PR middle-end/78901
34078 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34079 possibly throwing calls.
34081 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34082 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34083 and fns handling, rather than in a separate case SSA_NAME.
34085 2017-01-02 Jeff Law <law@redhat.com>
34087 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34088 fixes for first_period and second_period variables.
34090 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34093 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34094 (*insvqi_1): New insn pattern.
34095 (*insvqi_1_mem_rex64): Ditto.
34096 (*insvqi_2): Ditto.
34097 (*insvqi_3): Rename from *insvqi.
34099 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
34101 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
34103 * doc/cfg.texi (Edges): Remove reference to Java.
34104 (Maintaining the CFG): Ditto.
34106 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34108 PR middle-end/77674
34109 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
34110 transparent aliases.
34112 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34114 PR middle-end/77484
34115 * predict.def (PRED_CALL): Update hitrate.
34116 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
34117 * predict.c (tree_estimate_probability_bb): Split CALL predictor
34118 into direct/indirect/polymorphic variants.
34120 2017-01-01 Jakub Jelinek <jakub@redhat.com>
34122 Update copyright years.
34124 * gcc.c (process_command): Update copyright notice dates.
34125 * gcov-dump.c (print_version): Ditto.
34126 * gcov.c (print_version): Ditto.
34127 * gcov-tool.c (print_version): Ditto.
34128 * gengtype.c (create_file): Ditto.
34129 * doc/cpp.texi: Bump @copying's copyright year.
34130 * doc/cppinternals.texi: Ditto.
34131 * doc/gcc.texi: Ditto.
34132 * doc/gccint.texi: Ditto.
34133 * doc/gcov.texi: Ditto.
34134 * doc/install.texi: Ditto.
34135 * doc/invoke.texi: Ditto.
34137 Copyright (C) 2017 Free Software Foundation, Inc.
34139 Copying and distribution of this file, with or without modification,
34140 are permitted in any medium without royalty provided the copyright
34141 notice and this notice are preserved.