1 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
3 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
4 Alphabetize prototypes of built-in functions, separating out
5 built-in functions that are listed in this section but should be
8 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
11 * expmed.c (emit_store_flag): Do not emit setcc followed by a
12 conditional move when trapping comparison was split to a
13 non-trapping one (and vice versa).
15 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
17 * config/s390/s390.c (s390_function_profiler): Generate nops
18 instead of profiler call sequences.
19 * config/s390/s390.opt: Add the new option.
21 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
23 * config/s390/s390.c (s390_function_profiler): Generate
25 * config/s390/s390.opt: Add the new option.
27 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
29 * common.opt: Add the new warning.
30 * config/s390/s390.c (s390_function_profiler): Emit "brasl
31 %r0,__fentry__" when -mfentry is specified.
32 (s390_option_override_internal): Disallow -mfentry for 31-bit
34 * config/s390/s390.opt: Add the new option.
36 2018-07-16 Richard Biener <rguenther@suse.de>
39 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
40 for function-local FUNCTION_DECL and RESULT_DECL immediately.
42 2018-07-16 Martin Liska <mliska@suse.cz>
45 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
48 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
50 * config/arc/arcHS.md: Update ARCHS scheduling rules.
52 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
54 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
56 * config/arc/arc-cpus.def (hs4x): New cpu.
58 * config/arc/arc-tables.opt: Regenerate.
59 * config/arc/arc.c (arc_sched_issue_rate): New function.
60 (TARGET_SCHED_ISSUE_RATE): Define.
61 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
62 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
64 (attr tune): Add ARCHS4x tune values.
65 (attr tune_dspmpy): Define.
66 (*tst): Correct instruction type.
67 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
68 * config/arc/arcHS4x.md: New file.
69 * config/arc/fpu.md: Update instruction type attributes.
70 * config/arc/t-multilib: Regenerate.
72 2018-07-16 Tom de Vries <tdevries@suse.de>
75 * var-tracking.c (vt_initialize): Fix pre_dec handling.
77 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
79 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
82 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
84 PR tree-optimization/86514
85 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
86 conversion to a boolean type from a type with greater precision.
88 2018-07-16 Tom de Vries <tdevries@suse.de>
90 * var-tracking.c (vt_initialize): Print adjusted insn slim if
91 dump_flags request TDF_SLIM.
93 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
95 * fold-const.c (int_const_binop_1): Abstract...
96 (wide_int_binop): ...wide int code here.
97 (poly_int_binop): ...poly int code here.
98 Abstract the rest of int_const_binop_1 into int_const_binop.
99 * fold-const.h (wide_int_binop): New.
100 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
101 Remove useless PLUS/MINUS_EXPR case.
102 (zero_nonzero_bits_from_vr): Move wide int code...
103 (zero_nonzero_bits_from_bounds): ...here.
104 (extract_range_from_binary_expr_1): Move mask optimization code...
105 (range_easy_mask_min_max): ...here.
106 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
107 (range_easy_mask_min_max): New.
109 2018-07-15 Jeff Law <law@redhat.com>
112 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
115 2018-07-14 Jim Wilson <jimw@sifive.com>
117 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
119 2018-07-14 Paul Koning <ni1d@arrl.net>
121 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
123 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
125 * lto-streamer-out.c (copy_function_or_variable): Dump info about
128 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
129 Steve Munroe <munroesj52@gmail.com>
131 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
132 (_mm_andnot_si128): Likewise.
133 (_mm_or_si128): Likewise.
134 (_mm_xor_si128): Likewise.
136 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
139 * builtins.c (expand_builtin_memcmp): Inline the calls first
140 when result_eq is false.
141 (expand_builtin_strcmp): Inline the calls first.
142 (expand_builtin_strncmp): Likewise.
143 (inline_string_cmp): New routine. Expand a string compare
144 call by using a sequence of char comparison.
145 (inline_expand_builtin_string_cmp): New routine. Inline expansion
146 a call to str(n)cmp/memcmp.
147 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
149 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
151 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
153 * config/arm/driver-arm.c: Include arm-native.h.
154 (host_detect_local_cpu): Use auto-generated data tables.
155 (vendors, arm_cpu_table): Delete. Move part information to ...
156 * config/arm/arm-cpus.in: ... here.
157 * config/arm/parsecpu.awk (gen_native): New function.
158 (vendor, part): New CPU fields.
159 (END): Add support for building the native CPU detection tables.
160 * config/arm/t-arm (arm-native.h): Add build rule.
161 (driver-arm.o): Add dependency on arm-native.h.
163 2018-07-13 Richard Biener <rguenther@suse.de>
166 * match.pd (addr1 - addr2): Allow either of the operand to
169 2018-07-13 Tom de Vries <tdevries@suse.de>
171 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
174 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
176 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
177 arrays instead of numbered variables.
179 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
181 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
182 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
184 2018-07-13 Richard Biener <rguenther@suse.de>
187 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
188 instead of get_context_die.
190 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
191 Richard Biener <rguenther@suse.de>
194 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
195 that the loop latch destination where phi is defined.
197 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
199 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
200 (riscv_expand_epilogue): Add assertion to check interrupt mode.
201 (riscv_set_current_function): Extract getting interrupt type to new
203 (riscv_get_interrupt_type): New function.
204 (riscv_merge_decl_attributes): New function, checking interrupt type is
206 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
208 2018-07-12 Paul Koning <ni1d@arrl.net>
210 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
213 2018-07-12 Paul Koning <ni1d@arrl.net>
215 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
216 zero reference, add doloop_end instead.
217 * doc/md.texi (decrement_and_branch_until_zero): Remove.
218 (Looping patterns): Remove decrement_and_branch_until_zero. Add
219 detail for doloop_end.
221 2018-07-12 Martin Sebor <msebor@redhat.com>
224 * attribs.c (decl_attributes): Reject conflicting attributes before
225 calling attribute handlers.
227 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
229 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
231 (gcc::dump_manager::get_dump_file_name): likewise.
232 (dump_begin): Likewise.
233 * dumpfile.h (dump_begin): Update prototype.
234 (gcc::dump_manager::get_dump_file_name,
235 gcc::dump_manager::get_dump_file_name): Update prototype.
237 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
239 * internal-fn.h (vectorizable_internal_fn_p): New function.
240 * tree-vect-slp.c (compatible_calls_p): Likewise.
241 (vect_build_slp_tree_1): Remove nops argument. Handle calls
242 to internal functions.
243 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
245 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
247 * fold-const.h (inverse_conditions_p): Declare.
248 * fold-const.c (inverse_conditions_p): New function.
249 * match.pd: Use inverse_conditions_p. Add folds of view_converts
250 that test the inverse condition of a conditional internal function.
251 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
252 * internal-fn.c (internal_fn_mask_index): Handle conditional
254 (vectorized_internal_fn_supported_p): New function.
255 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
256 (any_pred_load_store): Replace with...
257 (need_to_predicate): ...this new variable.
258 (redundant_ssa_names): New variable.
259 (ifcvt_can_use_mask_load_store): Move initial checks to...
260 (ifcvt_can_predicate): ...this new function. Handle tree codes
261 for which a conditional internal function exists.
262 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
263 instead of ifcvt_can_use_mask_load_store. Update after variable
265 (predicate_load_or_store): New function, split out from
266 predicate_mem_writes.
267 (check_redundant_cond_expr): New function.
268 (value_available_p): Likewise.
269 (predicate_rhs_code): Likewise.
270 (predicate_mem_writes): Rename to...
271 (predicate_statements): ...this. Use predicate_load_or_store
272 and predicate_rhs_code.
273 (combine_blocks, tree_if_conversion): Update after above name changes.
274 (ifcvt_local_dce): Handle redundant_ssa_names.
275 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
276 general conditional functions.
277 * tree-vect-stmts.c (vectorizable_call): Likewise.
279 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
280 Alan Hayward <alan.hayward@arm.com>
281 David Sherwood <david.sherwood@arm.com>
283 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
284 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
285 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
286 plus and minus and convert them into IFN_COND_FMA-based sequences.
287 (convert_mult_to_fma): Handle conditional plus and minus.
289 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
291 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
292 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
293 (cond_fnms_optab): New optabs.
294 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
296 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
297 * internal-fn.h (get_conditional_internal_fn): Declare.
298 (get_unconditional_internal_fn): Likewise.
299 * internal-fn.c (cond_ternary_direct): New macro.
300 (expand_cond_ternary_optab_fn): Likewise.
301 (direct_cond_ternary_optab_supported_p): Likewise.
302 (FOR_EACH_COND_FN_PAIR): Likewise.
303 (get_conditional_internal_fn): New function.
304 (get_unconditional_internal_fn): Likewise.
305 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
306 (gimple_match_op::gimple_match_op): Add a new overload for 5
308 (gimple_match_op::set_op): Likewise.
309 (gimple_resimplify5): Declare.
310 * genmatch.c (decision_tree::gen): Generate simplifications for
312 * gimple-match-head.c (gimple_simplify): Define an overload for
313 5 operands. Handle calls with 5 arguments in the top-level overload.
314 (convert_conditional_op): Handle conversions from unconditional
315 internal functions to conditional ones.
316 (gimple_resimplify5): New function.
317 (build_call_internal): Pass a fifth operand.
318 (maybe_push_res_to_seq): Likewise.
319 (try_conditional_simplification): Try converting conditional
320 internal functions to unconditional internal functions.
321 Handle 3-operand unconditional forms.
322 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
323 Define ternary equivalents of the current rules for binary conditional
325 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
327 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
328 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
329 (optab): Handle them.
330 (SVE_COND_FP_TERNARY): New int iterator.
331 (sve_fmla_op, sve_fmad_op): New int attributes.
332 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
333 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
334 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
336 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
338 * target.def (preferred_else_value): New target hook.
339 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
340 * doc/tm.texi: Regenerate.
341 * targhooks.h (default_preferred_else_value): Declare.
342 * targhooks.c (default_preferred_else_value): New function.
343 * internal-fn.h (conditional_internal_fn_code): Declare.
344 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
345 (get_conditional_internal_fn): Use it.
346 (conditional_internal_fn_code): New function.
347 * gimple-match.h (gimple_match_cond): New struct.
348 (gimple_match_op): Add a cond member function.
349 (gimple_match_op::gimple_match_op): Update all forms to take a
351 * genmatch.c (expr::gen_transform): Use the same condition as res_op
352 for the suboperation, but don't specify a particular else_value.
353 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
354 (visit_nary_op, visit_reference_op_load): Pass
355 gimple_match_cond::UNCOND to the gimple_match_op constructor.
356 * gimple-match-head.c: Include tree-eh.h
357 (convert_conditional_op): New function.
358 (maybe_resimplify_conditional_op): Likewise.
359 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
360 (gimple_resimplify2): Likewise.
361 (gimple_resimplify3): Likewise.
362 (gimple_resimplify4): Likewise.
363 (maybe_push_res_to_seq): Return null for conditional operations.
364 (try_conditional_simplification): New function.
365 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
367 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
369 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
371 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
373 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
375 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
377 (hash_tree): Do not hash DECL_FCONTEXT
378 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
379 Do not stream DECL_FCONTEXT.
380 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
381 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
383 2018-07-12 Richard Biener <rguenther@suse.de>
386 * dwarf2out.c (gen_block_die): Only output blocks when they have
387 at least one !DECL_IGNORED_P variable.
389 2018-07-12 Richard Biener <rguenther@suse.de>
392 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
393 Remove -mieee-fp handling.
395 2018-07-12 Richard Biener <rguenther@suse.de>
397 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
398 left-over from last patch.
400 2018-07-12 Jakub Jelinek <jakub@redhat.com>
402 PR tree-optimization/86492
403 * gimple-ssa-store-merging.c
404 (imm_store_chain_info::coalesce_immediate_stores): Call
405 check_no_overlap even for the merge_overlapping case. Formatting fix.
407 2018-07-12 Richard Biener <rguenther@suse.de>
410 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
411 move possibly trapping operations into the conditional.
413 2018-07-12 Richard Biener <rguenther@suse.de>
415 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
416 (vn_lookup_simplify_result): Remove recursion limit applied
418 (vn_nary_build_or_lookup_1): Adjust.
419 (try_to_simplify): Likewise.
420 * gimple-match-head.c (gimple_resimplify1): Instead apply one
422 (gimple_resimplify2): Likewise.
423 (gimple_resimplify3): Likewise.
424 (gimple_resimplify4): Likewise.
426 2018-07-11 Jakub Jelinek <jakub@redhat.com>
428 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
429 Use __mmask64 type instead of __mmask8 for __M argument.
430 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
431 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
433 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
434 __mmask16 for __M argument.
435 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
436 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
437 to __mmask16 instead of __mmask8.
438 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
439 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
440 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
441 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
442 instead of __mmask16 for __U argument.
443 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
444 __mmask16 instead of __mmask8 for __U argument.
445 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
447 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
449 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
451 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
453 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
455 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
457 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
459 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
461 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
463 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
465 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
466 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
467 return type as well as __M argument type and all casts from __mmask8
469 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
470 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
471 return type as well as __M argument type and all casts from __mmask8
473 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
474 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
475 return type as well as __M argument type and all casts from __mmask8
477 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
478 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
479 return type as well as __M argument type and all casts from __mmask8
481 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
482 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
485 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
487 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
488 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
491 2018-07-11 Paul Koning <ni1d@arrl.net>
493 * doc/md.texi (define_subst): Document how multiple occurrences of
494 the same argument in the replacement pattern are handled.
496 2018-07-11 Paul Koning <ni1d@arrl.net>
498 * doc/extend.texi (Common Variable Attributes): Move "mode" into
500 (Common Type Attributes): Add "mode" attribute.
502 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
504 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
505 stream DECL_ORIGINAL_TYPE.
506 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
507 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
508 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
509 Do not walk original type.
510 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
512 (write_ts_decl_non_common_tree_pointers): Do not stream
514 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
515 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
517 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
519 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
520 thread twice from the same starting edge.
522 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
524 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
525 * gimple.c (gimple_call_nonnull_result_p): ...here...
526 (gimple_call_nonnull_arg): ...and here.
527 * gimple.h (gimple_call_nonnull_result_p): New.
528 (gimple_call_nonnull_arg): New.
530 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
532 * config/arm/arm-cpus.in: Move information from fpu field of each
533 cpu definition to the isa field.
534 * config/arm/parsecpu.awk (fpu): Delete match rule.
535 (gen_comm_data): Don't add bits from the CPU's FPU entry.
537 2018-07-11 Richard Biener <rguenther@suse.de>
540 * dwarf2out.c (init_sections_and_labels): Use
541 output_asm_line_debug_info consistently.
542 (dwarf2out_early_finish): Likewise.
543 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
546 2018-07-11 Richard Biener <rguenther@suse.de>
548 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
549 Rework father_bb setting in a way to avoid propagating constants
550 multiple times on a loop body.
552 2018-07-10 Mark Wielaard <mark@klomp.org>
555 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
557 2018-07-10 Richard Biener <rguenther@suse.de>
559 * hash-map.h (hash_map::iterator::operator*): Return
560 references to key and value.
562 2018-07-10 Jakub Jelinek <jakub@redhat.com>
565 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
566 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
567 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
568 (gimplify_omp_for): For composite loops, move outer
569 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
570 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
571 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
572 TREE_LIST for both the original class iterator and the "last" helper
573 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
574 loop, remember has_decl_expr from outer composite loops for the
575 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
577 2018-07-09 Martin Sebor <msebor@redhat.com>
581 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
582 accessing implicitly initialized array elements.
583 * expr.c (string_constant): Handle string initializers of
584 character arrays within aggregates.
585 * gimple-fold.c (fold_array_ctor_reference): Add argument.
586 Store element offset. As a special case, handle zero size.
587 (fold_nonarray_ctor_reference): Same.
588 (fold_ctor_reference): Add argument. Store subobject offset.
589 * gimple-fold.h (fold_ctor_reference): Add argument.
591 2018-07-09 Paul Koning <ni1d@arrl.net>
593 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
594 (pdp11_insn_cost): New function.
595 (pdp11_md_asm_adjust): New function.
596 (TARGET_INVALID_WITHIN_DOLOOP): Define.
597 (pdp11_rtx_costs): Update to match machine better.
598 (output_addr_const_pdp11): Correct format mismatch warnings.
599 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
600 * config/pdp11/pdp11.md: General change to add base_cost and/or
601 length attributes for use by new pdp11_insn_cost function.
602 (MIN_BRANCH): Correct definition.
604 (doloop_end): Use standard pattern name for looping pattern.
605 (doloop_end_nocc): New.
606 (movsf): Add another constraint alternative.
607 (zero_extendqihi2): Add constraint alternatives for not in place
609 (zero_extendhisi2): Remove.
610 (shift patterns): Add CC handling variants.
614 (define_peephole2): New peephole to recognize mov that sets CC for
617 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
619 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
620 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
623 2018-07-09 Jakub Jelinek <jakub@redhat.com>
626 * real.c (real_nextafter): Return true if result is denormal.
628 2018-07-09 Martin Liska <mliska@suse.cz>
630 * common.opt: Add back wrongly removed attribute.
632 2018-07-09 Richard Biener <rguenther@suse.de>
635 * dwarf2out.c (gen_block_die): For an early generated DIE
636 always output high/low PC attributes.
638 2018-07-09 Tom de Vries <tdevries@suse.de>
640 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
641 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
642 onto VAR_DECL with abstract origin.
644 2018-07-07 Jim Wilson <jimw@sifive.com>
646 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
648 2018-07-07 Tom de Vries <tdevries@suse.de>
650 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
652 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
654 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
656 (combine_bound): Use wide-int overflow calculation instead of
658 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
660 * fold-const.c (int_const_binop_2): Same.
661 (extract_muldiv_1): Same.
662 (fold_div_compare): Same.
663 (fold_abs_const): Same.
665 * poly-int.h (add): Same.
669 * predict.c (predict_iv_comparison): Same.
670 * profile-count.c (slow_safe_scale_64bit): Same.
671 * simplify-rtx.c (simplify_const_binary_operation): Same.
672 * tree-chrec.c (tree_fold_binomial): Same.
673 * tree-data-ref.c (split_constant_offset_1): Same.
674 * tree-if-conv.c (idx_within_array_bound): Same.
675 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
676 * tree-ssa-phiopt.c (minmax_replacement): Same.
677 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
678 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
679 * vr-values.c (vr_values::adjust_range_with_scev): Same.
680 * wide-int.cc (wi::add_large): Same.
681 (wi::mul_internal): Same.
682 (wi::sub_large): Same.
683 (wi::divmod_internal): Same.
684 * wide-int.h: Change overflow type to overflow_type for neg, add,
685 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
686 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
687 mul_internal, divmod_internal.
688 (overflow_type): New enum.
689 (accumulate_overflow): New.
691 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
693 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
694 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
696 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
698 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
699 argument is checked for zero before entering loop, avoid checking again.
701 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
703 * gimplify.h (generic_expr_could_trap_p): Set as global function.
704 * gimplify.h (generic_expr_could_trap_p): Likwise.
705 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
707 2018-07-06 Jakub Jelinek <jakub@redhat.com>
709 PR tree-optimization/86401
710 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
711 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
712 (fold_bit_and_mask): ... here. New helper function for match.pd.
713 * fold-const.h (fold_bit_and_mask): Declare.
714 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
716 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
719 * target.def (translate_mode_attribute): New hook.
720 * targhooks.h (default_translate_mode_attribute): Declare.
721 * targhooks.c (default_translate_mode_attribute): New function.
722 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
723 * doc/tm.texi: Regenerate.
724 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
725 (rs6000_translate_mode_attribute): New function.
727 2018-07-06 Paul Koning <ni1d@arrl.net>
729 * doc/md.texi (define_split): Document DONE and FAIL.
730 (define_peephole2): Ditto.
732 2018-07-05 Jeff Law <law@redhat.com>
734 PR tree-optimization/86010
735 * tree-ssa-dse.c (compute_trims): More aggressively trim at
736 both the head and tail of mem* and str* calls.
738 2018-07-05 Jim Wilson <jimw@sifive.com>
740 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
742 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
744 * config/aarch64/aarch64-simd.md: correct flags text for
745 MIN_EXPR replacement.
747 2018-07-05 James Clarke <jrtc27@jrtc27.com>
749 * configure: Regenerated.
751 2018-07-05 Carl Love <cel@us.ibm.com>
753 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
754 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
755 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
756 VSX_BUILTIN_DOUBLEL_V4SF.
758 2018-07-05 Martin Sebor <msebor@redhat.com>
761 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
762 than its domain to compute its the upper bound of a char array.
764 2018-07-05 Nathan Sidwell <nathan@acm.org>
766 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
767 * doc/cpp.texi: Update comment.
768 * doc/tm.texi: Rebuilt.
769 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
770 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
771 * doc/extend.texi (Backwards Compatibility): Clarify it is system
772 headers affected by extern "C".
773 * system.h: Poison NO_IMPLICIT_EXTERN_C.
774 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
775 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
776 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
777 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
778 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
779 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
780 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
781 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
782 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
783 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
784 config/sparc/sp64-elf.h, config/spu/spu.h,
785 config/stormy16/stormy16.h, config/v850/v850.h,
786 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
787 define NO_IMPLICIT_EXTERN_C.
788 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
790 2018-07-05 Tamar Christina <tamar.christina@arm.com>
793 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
794 instead of GET_MODE_SIZE when comparing Units.
796 2018-07-05 Tamar Christina <tamar.christina@arm.com>
799 * rtlanal.c (set_noop_p): Constrain on mode change,
800 include hard-reg-set.h
802 2018-07-05 Tamar Christina <tamar.christina@arm.com>
804 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
806 2018-07-05 Jakub Jelinek <jakub@redhat.com>
809 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
812 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
814 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
816 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
819 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
821 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
823 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
826 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
828 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
830 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
833 2018-07-04 Martin Liska <mliska@suse.cz>
835 * tree-switch-conversion.c: Define
836 max_ratio_for_speed and max_ratio_for_size constants.
838 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
839 Martin Liska <mliska@suse.cz>
844 * common.opt: Rename align options with 'str_' prefix.
845 * common/config/i386/i386-common.c (set_malign_value): New
847 (ix86_handle_option): Use it to set -falign-* options/
848 * config/aarch64/aarch64-protos.h (struct tune_params): Change
849 type from int to string.
850 * config/aarch64/aarch64.c: Update default values from int
852 * config/alpha/alpha.c (alpha_override_options_after_change):
854 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
855 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
856 max skip conditionally.
857 * config/i386/freebsd.h (SUBALIGN_LOG): New.
858 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
859 max skip conditionally.
860 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
861 max skip conditionally.
862 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
863 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
864 max skip conditionally.
865 * config/i386/i386.c (struct ptt): Change type from int to
867 (ix86_default_align): Set default values.
868 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
869 max skip conditionally.
870 * config/i386/iamcu.h (SUBALIGN_LOG): New.
871 (ASM_OUTPUT_MAX_SKIP_ALIGN):
872 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
873 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
874 max skip conditionally.
875 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
876 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
877 * config/i386/x86-64.h (SUBALIGN_LOG): New.
878 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
879 max skip conditionally.
880 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
881 * config/ia64/ia64.c (ia64_option_override): Set default values
882 for alignment options.
883 * config/m68k/m68k.c: Handle new str_align_* options.
884 * config/mips/mips.c (mips_set_compression_mode): Change
886 (mips_option_override): Set default values for options.
887 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
889 * config/rs6000/rs6000.c (rs6000_option_override_internal):
891 * config/rx/rx.c (rx_option_override): Likewise.
892 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
893 (LABEL_ALIGN): Use align_labels_log.
894 (LOOP_ALIGN): Use align_loops_align.
895 * config/s390/s390.c (s390_asm_output_function_label): Use new
897 * config/sh/sh.c (sh_override_options_after_change):
898 Change type of constants.
899 * config/spu/spu.c (spu_sched_init): Likewise.
900 * config/sparc/sparc.c (sparc_option_override): Set default
902 * config/visium/visium.c (visium_option_override): Likewise.
903 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
904 emit p2align format with last argument if it's not needed.
905 * doc/invoke.texi: Document extended format of -falign-*.
906 * final.c: Use align_labels alignment.
907 * flags.h (struct target_flag_state): Change type to use
909 (struct align_flags_tuple): New.
910 (struct align_flags): Likewise.
911 (align_loops_log): Redefine macro to use new types.
912 (align_loops_max_skip): Redefine macro to use new types.
913 (align_jumps_log): Redefine macro to use new types.
914 (align_jumps_max_skip): Redefine macro to use new types.
915 (align_labels_log): Redefine macro to use new types.
916 (align_labels_max_skip): Redefine macro to use new types.
917 (align_functions_log): Redefine macro to use new types.
918 (align_loops): Redefine macro to use new types.
919 (align_jumps): Redefine macro to use new types.
920 (align_labels): Redefine macro to use new types.
921 (align_functions): Redefine macro to use new types.
922 (align_functions_max_skip): Redefine macro to use new types.
923 (align_loops_value): New macro.
924 (align_jumps_value): New macro.
925 (align_labels_value): New macro.
926 (align_functions_value): New macro.
927 * function.c (invoke_set_current_function_hook): Propagate
928 alignment values from flags to global variables default in
930 * ipa-icf.c (sem_function::equals_wpa): Use
931 cl_optimization_option_eq instead of memcmp.
932 * lto-streamer.h (cl_optimization_stream_out): Support streaming
934 (cl_optimization_stream_in): Likewise.
935 * optc-save-gen.awk: Support strings in cl_optimization.
936 * opth-gen.awk: Likewise.
937 * opts.c (finish_options): Remove error checking of invalid
939 (MAX_CODE_ALIGN): Remove.
940 (MAX_CODE_ALIGN_VALUE): Likewise.
941 (parse_and_check_align_values): New function.
942 (check_alignment_argument): Likewise.
943 (common_handle_option): Use check_alignment_argument.
944 * opts.h (parse_and_check_align_values): Declare.
945 * toplev.c (init_alignments): Remove.
946 (read_log_maxskip): New.
947 (parse_N_M): Likewise.
948 (parse_alignment_opts): Likewise.
949 (backend_init_target): Remove usage of init_alignments.
950 * toplev.h (parse_alignment_opts): Declare.
951 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
953 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
954 * tree.c (cl_option_hasher::equal): New.
955 * varasm.c: Use new global macros.
957 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
959 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
960 Use a simpler align directive also if MAXSKIP = ALIGN-1.
961 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
962 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
963 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
964 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
965 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
966 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
967 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
968 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
969 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
970 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
971 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
973 2018-07-04 Martin Liska <mliska@suse.cz>
974 Jonathan Wakely <jwakely@redhat.com>
976 * coverage.c: Use correct type.
977 * doc/invoke.texi: Language correction.
979 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
982 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
983 ENDBRANCH for non-tail call which may return via indirect branch.
984 * doc/extend.texi: Document indirect_return attribute.
986 2018-07-03 Martin Sebor <msebor@redhat.com>
988 PR tree-optimization/86274
989 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
991 (format_floating): Correct handling of infinities and NaNs.
993 2018-07-03 Martin Sebor <msebor@redhat.com>
995 * print-tree.c (print_real_cst): New function.
996 (print_node_brief): Call it.
999 2018-07-03 Jeff Law <law@redhat.com>
1001 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
1002 into a single pattern.
1004 * config/h8300/h8300.md (ors code_iterator): New.
1005 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
1006 a single pattern and single splitter.
1007 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
1008 (iorqi3_1, xorqi3_1): Likewise.
1009 (iorqi3, xorqi3 expanders): Similarly.
1011 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
1012 (movmd_internal) into a single pattern using the P mode iterator.
1013 (movmd splitters): Similarly.
1014 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
1015 (movsd splitters): Similarly.
1017 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
1018 ADDB, ADDW and ADDL into a single ADD attribute which selects the
1019 right table based on the size of the operand.
1020 * config/h8300/h8300.md (length_table): Corresponding changes. All
1021 references to "addb", "addw" and "addl" changed to "add".
1022 (btst patterns): Merge two variants into a single pattern.
1023 (tstqi, tsthi): Likewise.
1024 (addhi3_incdec, addsi3_incdec): Likewise.
1025 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
1026 (mulhi3, mulsi3): Likewise.
1027 (udivhi3, udivsi3): Likewise.
1028 (divhi3, divsi3): Likewise.
1029 (andorqi3, andorhi3, andorsi3): Likewise.
1031 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
1034 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
1035 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
1037 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
1039 PR tree-optimization/85694
1040 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
1041 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
1043 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
1044 (<u>avg<mode>3_ceil): New patterns.
1046 2018-07-03 David Malcolm <dmalcolm@redhat.com>
1048 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
1049 scan-tree-dump directive.
1050 * gcc.dg/vect/slp-perm-2.c: Likewise.
1051 * gcc.dg/vect/slp-perm-3.c: Likewise.
1052 * gcc.dg/vect/slp-perm-5.c: Likewise.
1053 * gcc.dg/vect/slp-perm-6.c: Likewise.
1054 * gcc.dg/vect/slp-perm-7.c: Likewise.
1055 * gcc.dg/vect/slp-perm-8.c: Likewise.
1057 2018-07-03 Marek Polacek <polacek@redhat.com>
1060 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
1062 2018-07-03 Richard Biener <rguenther@suse.de>
1065 * tree-ssa-structalias.c (find_func_clobbers): Properly
1066 handle indirect calls.
1068 2018-07-03 Jeff Law <law@redhat.com>
1070 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
1071 (shifts): New code iterator.
1072 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
1073 expander. Fix HImode handling on H8/SX.
1074 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
1075 (subqi3, subhi3, subsi3 expanders): Likewise.
1076 (andqi3, andhi3, andsi3 expanders): Likewise.
1077 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
1078 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
1079 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
1080 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
1081 (zero_extendqihi2, zero_extendqisi2): Likewise.
1082 (extendqihi2, extendqisi2): Likewise.
1083 (rotlqi3, rotlhi3, rotlsi3): Likewise.
1084 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
1085 (rotlqi3_1, rotlhi3_1): Likewise.
1086 (logicalhi3_sn, logicalsi3_sn): Likewise.
1087 (logicalhi3, logicalsi3): Likewise.
1089 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
1091 * tree-vect-patterns.c (vect_recog_rotate_pattern)
1092 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
1093 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
1094 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
1095 type to append_pattern_def_seq instead of creating a stmt_vec_info
1097 (build_mask_conversion): Likewise. Remove vinfo argument.
1098 (vect_add_conversion_to_patterm): Likewise, renaming to...
1099 (vect_add_conversion_to_pattern): ...this.
1100 (vect_recog_mask_conversion_pattern): Update call to
1101 build_mask_conversion. Pass the vector type to
1102 append_pattern_def_seq here too.
1103 (vect_recog_gather_scatter_pattern): Update call to
1104 vect_add_conversion_to_pattern.
1106 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
1108 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
1109 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
1110 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
1111 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
1112 STMT_VINFO_PATTERN_DEF_SEQ to null here.
1113 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
1114 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
1115 append_pattern_def_seq instead of new_pattern_def_seq.
1116 (vect_recog_divmod_pattern): Do both of the above.
1117 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
1120 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
1122 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
1123 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
1124 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
1125 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
1126 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
1127 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
1128 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
1129 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
1130 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
1131 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
1132 parameter with a single stmt_vec_info.
1133 (vect_recog_func_ptr): Likewise.
1134 (vect_recog_gather_scatter_pattern): Likewise, folding in...
1135 (vect_try_gather_scatter_pattern): ...this.
1136 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
1137 the stmt_vec_info of the statement to be matched. Don't clear
1138 STMT_VINFO_RELATED_STMT.
1139 (vect_pattern_recog): Update call accordingly.
1141 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
1143 PR tree-optimization/85694
1144 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
1145 (uavgM3_ceil): Document new optabs.
1146 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
1147 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
1149 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
1150 (savg_ceil_optab): New optabs.
1151 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
1152 (vect_vect_recog_func_ptrs): Add it.
1153 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
1154 constant directly from the associated lhs.
1156 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
1158 * tree-vect-patterns.c (vect_split_statement): New function.
1159 (vect_convert_input): Use it to try to split an existing cast.
1161 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
1163 * poly-int.h (print_hex): New function.
1164 * dumpfile.h (dump_dec, dump_hex): Declare.
1165 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
1166 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
1167 min_input_precision, operation_precision and operation_sign.
1168 * tree-vect-patterns.c (vect_get_range_info): New function.
1169 (vect_same_loop_or_bb_p, vect_single_imm_use)
1170 (vect_operation_fits_smaller_type): Delete.
1171 (vect_look_through_possible_promotion): Add an optional
1172 single_use_p parameter.
1173 (vect_recog_over_widening_pattern): Rewrite to use new
1174 stmt_vec_info infomration. Handle one operation at a time.
1175 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
1176 (vect_truncatable_operation_p, vect_set_operation_type)
1177 (vect_set_min_input_precision): New functions.
1178 (vect_determine_min_output_precision_1): Likewise.
1179 (vect_determine_min_output_precision): Likewise.
1180 (vect_determine_precisions_from_range): Likewise.
1181 (vect_determine_precisions_from_users): Likewise.
1182 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
1183 (vect_vect_recog_func_ptrs): Put over_widening first.
1185 (vect_pattern_recog): Call vect_determine_precisions.
1187 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
1189 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
1190 statements that have been replaced by further pattern statements.
1191 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
1193 2018-07-03 Richard Biener <rguenther@suse.de>
1195 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
1196 always set *dt. Dump vectype in vectype overload.
1197 * dumpfile.h (dump_gimple_expr): New function.
1198 (dump_gimple_expr_loc): Likewise.
1199 * dumpfile.c (dump_gimple_expr): New function.
1200 (dump_gimple_expr_loc): Likewise.
1202 2018-07-02 Jeff Law <law@redhat.com>
1204 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
1205 the H8/300, H8/300H and H8/S variants into a single pattern.
1206 (movhi_h8300, movqi_h8300hs): Similarly.
1207 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
1208 (QHI mode iterator): New.
1210 * config/h8300/h8300.md: Remove trailing whitespace.
1212 2018-07-02 Jim Wilson <jimw@sifive.com>
1214 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
1215 instead of emit_insn for interrupt returns.
1216 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
1217 (riscv_sret, riscv_uret): Likewise.
1219 2018-07-02 David Malcolm <dmalcolm@redhat.com>
1221 * pretty-print.c (selftest::test_pp_format): Move save and restore
1222 of quotes to class auto_fix_quotes, and add an instance.
1223 * selftest.c: Include "intl.h".
1224 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
1225 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
1226 * selftest.h (selftest::auto_fix_quotes): New class.
1228 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
1230 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
1231 (aarch64_sve_prepare_conditional_op): Remove.
1232 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
1233 Allow aarch64_simd_reg_or_zero as select operand; remove
1234 the aarch64_sve_prepare_conditional_op call.
1235 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
1236 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
1237 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
1238 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
1239 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
1240 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
1241 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
1242 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
1243 and a splitters to match all of the *_any patterns.
1244 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
1246 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
1247 (SVE_COND_FP_BINARY_REV): Remove.
1248 (sve_int_op_rev, sve_fp_op_rev): New.
1249 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
1250 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
1251 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
1252 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
1253 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
1254 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
1255 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
1256 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
1257 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
1259 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
1260 Remove match_dup 1 from the inner unspec.
1261 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
1263 * config/aarch64/aarch64.md (movprfx): New attr.
1264 (length): Default movprfx to 8.
1265 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
1266 (*madd<SVE_I>, *msub<SVE_I): Likewise.
1267 (*<su>mul<SVE_I>3_highpart): Likewise.
1268 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
1269 (*v<ASHIFT><SVE_I>3): Likewise.
1270 (*<su><MAXMIN><SVE_I>3): Likewise.
1271 (*<su><MAXMIN><SVE_F>3): Likewise.
1272 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
1273 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
1274 (*div<SVE_F>4): Likewise.
1276 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
1278 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
1281 2018-07-02 Richard Biener <rguenther@suse.de>
1283 PR tree-optimization/86363
1284 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
1285 memset argument refers to a non-variable address.
1287 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
1289 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
1290 {PLUS,MINUS}_EXPR code to...
1291 (adjust_symbolic_bound): ...here,
1292 (combine_bound): ...here,
1293 (set_value_range_with_overflow): ...and here.
1295 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
1297 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
1299 (extract_range_from_abs_expr): ...here.
1301 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
1303 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
1304 -fno-omit-frame-pointer when not optimizing.
1306 2018-07-02 Martin Liska <mliska@suse.cz>
1309 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
1310 (propagate_nothrow): Likewise.
1312 2018-07-02 Martin Liska <mliska@suse.cz>
1315 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
1317 2018-07-02 David Malcolm <dmalcolm@redhat.com>
1319 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
1320 function in r262149, changing "loc" param from source_location to
1321 const dump_location_t &.
1322 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
1323 declaration, as above.
1325 2018-07-01 Paul Koning <ni1d@arrl.net>
1327 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
1328 -munit-asm, -mgnu-asm, -mdec-asm.
1329 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
1330 (pdp11_output_labelref): New.
1331 (pdp11_output_def): New.
1332 (pdp11_output_addr_vec_elt): New.
1333 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
1334 %# and %@ format codes.
1335 (pdp11_option_override): New.
1336 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
1337 (pdp11_output_ident): New.
1338 (pdp11_asm_named_section): New.
1339 (pdp11_asm_init_sections): New.
1340 (pdp11_file_start): New.
1341 (pdp11_file_end): New.
1342 (output_ascii): Use .ascii/.asciz for -mdec-asm.
1343 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
1344 %o, like %c but octal.
1345 (pdp11_option_override): New.
1346 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
1348 (DATA_SECTION_ASM_OP): Ditto.
1349 (READONLY_DATA_SECTION_ASM_OP): New.
1350 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
1351 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
1352 (ASM_OUTPUT_LABELREF): Ditto.
1353 (ASM_OUTPUT_DEF): Ditto.
1354 (ASM_OUTPUT_EXTERNAL): New.
1355 (ASM_OUTPUT_SOURCE_FILENAME): New.
1356 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
1357 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
1358 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
1359 %# and %@ format codes.
1360 * config/pdp11/pdp11.opt (mgnu-asm): New.
1361 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
1362 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
1363 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
1365 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
1367 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
1368 dereferencing path[] beyond its length.
1370 (debug_all_paths): New.
1371 (rewire_first_differing_edge): New.
1372 (adjust_paths_after_duplication): New.
1373 (duplicate_thread_path): Call adjust_paths_after_duplication.
1375 (thread_through_all_blocks): Add new argument to
1376 duplicate_thread_path.
1378 2018-06-30 Jim Wilson <jimw@sifive.com>
1380 * config/riscv/predicates.md (p2m1_shift_operand): New.
1381 (high_mask_shift_operand): New.
1382 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
1383 pattern using p2m1_shift_operand.
1384 (lshsi3_zero_extend_3+2): New combiner pattern using
1385 high_mask_shift_operand.
1387 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
1389 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
1391 (vect_recog_rotate_pattern): ...here.
1392 (vect_convert_input): Try to insert casts of invariants in the
1394 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
1395 preheader to be empty.
1397 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
1399 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
1400 vector type. If given, install it in the new statement's
1402 (vect_element_precision): New function.
1403 (vect_unpromoted_value): New struct.
1404 (vect_unpromoted_value::vect_unpromoted_value): New function.
1405 (vect_unpromoted_value::set_op): Likewise.
1406 (vect_look_through_possible_promotion): Likewise.
1407 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
1408 (vect_widened_op_tree, vect_convert_input): Likewise.
1409 (vect_convert_inputs, vect_convert_output): Likewise.
1410 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
1411 to handle the optional cast of the multiplication result and
1412 vect_widened_op_tree to detect the widened multiplication itself.
1413 Do not require the input and output of promotion casts to have
1414 the same sign, but base the signedness of the operation on the
1415 input rather than the result. If the pattern includes two
1416 promotions, check that those promotions have the same sign.
1417 Do not restrict the MULT_EXPR handling to a double-width result;
1418 handle quadruple-width results and wider. Use vect_convert_inputs
1419 to convert the inputs to the common type.
1420 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
1421 to handle the optional cast of the ABS result. Also allow a sign
1422 change or a sign extension between the ABS and MINUS.
1423 Use vect_widened_op_tree to detect the widened subtraction and use
1424 vect_convert_inputs to convert the inputs to the common type.
1425 (vect_handle_widen_op_by_const): Delete.
1426 (vect_recog_widen_op_pattern): New function.
1427 (vect_recog_widen_mult_pattern): Use it.
1428 (vect_recog_widen_shift_pattern): Likewise.
1429 (vect_recog_widen_sum_pattern): Use
1430 vect_look_through_possible_promotion to handle the promoted
1433 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
1435 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
1436 the containing gimple_seq *.
1437 * gimple-iterator.h (gsi_for_stmt): Declare it.
1438 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
1439 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
1440 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
1441 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
1442 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
1444 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
1446 (vect_mark_pattern_stmts): ...here. Handle cases in which the
1447 statement being replaced is part of an existing pattern
1448 definition sequence, inserting the new pattern statements before
1450 (vect_pattern_recog_1): Don't return a bool. If the statement
1451 is already part of a pattern, instead apply pattern matching
1452 to the pattern definition statements. Don't clear the
1453 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
1454 (vect_pattern_recog): Don't break after the first match;
1455 continue processing the pattern definition statements instead.
1456 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
1458 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
1460 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
1461 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
1462 (vect_recog_widen_sum_pattern): Use it.
1464 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
1466 * tree-vect-loop.c (vectorizable_reduction): Assert that the
1467 phi is not a pattern statement and has not been replaced by
1468 a pattern statement.
1469 * tree-vect-patterns.c (type_conversion_p): Don't check
1470 STMT_VINFO_IN_PATTERN_P.
1471 (vect_recog_vector_vector_shift_pattern): Likewise.
1472 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
1473 the pattern statement rather than the original statement; check
1474 directly for a WIDEN_MULT_EXPR here.
1475 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
1476 vect_is_simple_use to return the pattern statement rather
1477 than the original statement; use is_pattern_stmt_p to check
1478 for such a pattern statement.
1479 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
1480 to return the pattern statement rather than the original statement;
1481 don't do the same transformation here.
1482 (vect_is_simple_use): If the defining statement has been replaced
1483 by a pattern statement, return the pattern statement instead.
1484 Remove the corresponding (local) transformation from the vectype
1487 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
1489 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
1490 end and default to null.
1491 * tree-vect-loop.c (vect_create_epilog_for_reduction)
1492 (vectorizable_reduction): Update calls accordingly, dropping the
1493 gimple ** argument if the passed-back statement isn't needed.
1494 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
1495 (vect_recog_rotate_pattern): Likewise.
1496 (vect_recog_mask_conversion_pattern): Likewise.
1497 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
1498 (vect_mask_constant_operand_p): Likewise.
1499 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
1500 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
1501 (get_group_load_store_type, get_load_store_type): Likewise.
1502 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
1503 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
1504 (vectorizable_conversion, vectorizable_assignment): Likewise.
1505 (vectorizable_shift, vectorizable_operation): Likewise.
1506 (vectorizable_store, vect_is_simple_cond): Likewise.
1507 (vectorizable_condition, vectorizable_comparison): Likewise.
1508 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
1509 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
1510 and move it to the end. Cope with null def_stmt_outs.
1512 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
1514 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
1516 2018-06-29 Jeff Law <law@redhat.com>
1518 * config/v850/v850.c (v850_legitimate_address_p): Handle large
1519 displacements for TARGET_V850E2V3 and newer.
1520 (TARGET_LRA_P): Remove. Defaults to LRA now.
1521 * config/v850/v850.md (sign23byte_load): Remove.
1522 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
1523 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
1525 2018-06-29 Martin Liska <mliska@suse.cz>
1528 * coverage.c (coverage_init): Mangle full path name.
1529 * doc/invoke.texi: Document the change.
1530 * gcov-io.c (mangle_path): New.
1531 * gcov-io.h (mangle_path): Likewise.
1532 * gcov.c (mangle_name): Use mangle_path for path mangling.
1534 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1536 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
1537 if starting source register is not even.
1539 2018-06-29 Martin Liska <mliska@suse.cz>
1541 PR tree-optimization/86263
1542 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
1543 Make edge redirection.
1545 2018-06-29 David Malcolm <dmalcolm@redhat.com>
1547 * dumpfile.c (dump_loc): Add indentation based on scope depth.
1548 (dump_scope_depth): New variable.
1549 (get_dump_scope_depth): New function.
1550 (dump_begin_scope): New function.
1551 (dump_end_scope): New function.
1552 * dumpfile.h (get_dump_scope_depth): New declaration.
1553 (dump_begin_scope): New declaration.
1554 (dump_end_scope): New declaration.
1555 (class auto_dump_scope): New class.
1556 (AUTO_DUMP_SCOPE): New macro.
1557 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
1560 2018-06-29 Richard Biener <rguenther@suse.de>
1562 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
1563 compute_all_dependences succeeds.
1564 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
1565 exceed --param loop-max-datarefs-for-datadeps.
1567 2018-06-29 Jakub Jelinek <jakub@redhat.com>
1569 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
1571 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
1574 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
1575 alternative 0 in preferred_for_speed attribute.
1577 2018-06-28 Paul Koning <ni1d@arrl.net>
1579 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
1580 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
1581 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
1582 * config/pdp11/pdp11.md: Correct "length" attribute calculation
1583 for shift insn patterns.
1585 2018-06-28 David Malcolm <dmalcolm@redhat.com>
1587 * cgraph.c (cgraph_node::get_body): Replace assignments to
1588 "dump_file" with calls to set_dump_file.
1589 * dumpfile.c (alt_dump_file): Make static, and group with...
1590 (alt_flags): ...this definition.
1591 (dumps_are_enabled): New variable.
1592 (refresh_dumps_are_enabled): New function.
1593 (set_dump_file): New function.
1594 (set_alt_dump_file): New function.
1595 (gcc::dump_manager::dump_start): Replace assignments to
1596 "dump_file" and "alt_dump_file" with calls to set_dump_file and
1598 (gcc::dump_manager::dump_finish): Likewise.
1599 * dumpfile.h (alt_dump_file): Delete decl.
1600 (dumps_are_enabled): New variable decl.
1601 (set_dump_file): New function decl.
1602 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
1604 * tree-nested.c (lower_nested_functions): Replace assignments to
1605 "dump_file" with calls to set_dump_file.
1607 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
1609 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
1610 goto_locus of each outgoing edge of each basic block.
1612 2018-06-28 Richard Biener <rguenther@suse.de>
1614 * dwarf2out.c (decl_scope_table): Remove.
1615 (push_decl_scope): Likewise.
1616 (pop_decl_scope): Likewise.
1617 (gen_type_die_for_member): Do not call push/pop_decl_scope.
1618 (gen_struct_or_union_type_die): Likewise.
1619 (gen_tagged_type_die): Likewise.
1620 (dwarf2out_init): Do not initialize decl_scope_table.
1621 (dwarf2out_c_finalize): Do not free it.
1623 2018-06-28 Richard Biener <rguenther@suse.de>
1625 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
1626 deciding whether to not re-use a DIE.
1628 2018-06-28 Richard Biener <rguenther@suse.de>
1630 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
1631 DW_AT_abstract_origin attribute.
1633 2018-06-28 Martin Liska <mliska@suse.cz>
1635 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
1636 Use newly introduced constants.
1637 * tree-switch-conversion.h (struct jump_table_cluster):
1638 Define max_ratio_for_size and max_ratio_for_speed.
1640 2018-06-28 Martin Liska <mliska@suse.cz>
1642 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
1643 Add new checking assert to catch invalid state.
1644 (jump_table_cluster::can_be_handled): Handle single case
1646 (jump_table_cluster::is_beneficial): Bail out for such case.
1647 (bit_test_cluster::find_bit_tests):
1648 Add new checking assert to catch invalid state.
1649 (bit_test_cluster::can_be_handled): Handle single case
1651 (bit_test_cluster::is_beneficial): Bail out for such case.
1652 (switch_decision_tree::analyze_switch_statement):
1655 2018-06-28 Martin Liska <mliska@suse.cz>
1657 * common.opt: Introduce -completion option.
1658 * gcc.c (driver_handle_option): Handle it.
1659 (driver::main): Print completions if completion
1661 * opt-suggestions.c (option_proposer::get_completions):
1663 (option_proposer::suggest_completion): Likewise.
1664 (option_proposer::find_param_completions): Likewise.
1665 (verify_autocompletions): Likewise.
1666 (test_completion_valid_options): Likewise.
1667 (test_completion_valid_params): Likewise.
1668 (in_completion_p): Likewise.
1669 (empty_completion_p): Likewise.
1670 (test_completion_partial_match): Likewise.
1671 (test_completion_garbage): Likewise.
1672 (opt_proposer_c_tests): Likewise.
1673 * opt-suggestions.h: Declare new functions.
1674 * opts.c (common_handle_option): Handle OPT__completion_.
1675 * selftest-run-tests.c (selftest::run_tests): Add
1676 opt_proposer_c_tests.
1677 * selftest.c (assert_str_startswith): New.
1678 * selftest.h (assert_str_startswith): Likewise.
1679 (opt_proposer_c_tests): New.
1680 (ASSERT_STR_STARTSWITH): Likewise.
1682 2018-06-28 Martin Liska <mliska@suse.cz>
1684 * Makefile.in: Add opt-suggestions.o.
1685 * gcc-main.c: Include opt-suggestions.h.
1686 * gcc.c (driver::driver): Likewise.
1687 (driver::~driver): Remove m_option_suggestions.
1688 (driver::build_option_suggestions): Moved to option_proposer.
1689 (driver::suggest_option): Likewise.
1690 (driver::handle_unrecognized_options): Use option_proposer.
1691 * gcc.h (class driver): Add new memver m_option_proposer.
1692 * opt-suggestions.c: New file.
1693 * opt-suggestions.h: New file.
1695 2018-06-28 Martin Liska <mliska@suse.cz>
1697 * vec.h (class auto_string_vec): New (moved from auto_argvec).
1698 (auto_string_vec::~auto_string_vec): Likewise.
1700 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
1702 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
1703 prevent_decl_creation_for_types fields up and add reset_location field.
1704 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
1705 statement if id->reset_location is true.
1706 (copy_edges_for_bb): Do not set goto_locus on the new edges if
1707 id->reset_location is true.
1708 (copy_phis_for_bb): Force input_location on the arguments if
1709 id->reset_location is true.
1710 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
1711 is set on the function to be inlined.
1713 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
1715 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
1717 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
1719 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
1720 registers for Pmode.
1721 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
1722 hard registers for the clobbered pseudo.
1724 2018-06-27 Paul Koning <ni1d@arrl.net>
1726 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
1727 mutually exclusive options.
1728 * config/pdp11/constraints.md (h): New constraint.
1729 (O): Update definition to match shift code generation.
1730 (D): New constraint.
1731 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
1733 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
1735 (output_jump): Change arguments.
1736 (pdp11_fixed_cc_regs): New function.
1737 (pdp11_cc_mode): Ditto.
1738 (pdp11_expand_shift): Ditto.
1739 (pdp11_assemble_shift): Ditto.
1740 (pdp11_small_shift): Ditto.
1741 (pdp11_branch_cost): Remove.
1742 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
1744 (pdp11_register_move_cost): Update for CC registers.
1745 (pdp11_rtx_costs): Add case for LSHIFTRT.
1746 (pdp11_output_jump): Add CCNZ mode conditional branches.
1747 (notice_update_cc_on_set): Remove.
1748 (pdp11_cc_mode): New function.
1749 (simple_memory_operand): Correct pre/post decrement case.
1750 (no_side_effect_operand): New function.
1751 (pdp11_regno_reg_class): Add CC_REGS class.
1752 (pdp11_fixed_cc_regs): New function.
1753 (pdp11_small_shift): New function.
1754 (pdp11_expand_shift): New function to expand shift insns.
1755 (pdp11_assemble_shift): New function to output shifts.
1756 (pdp11_branch_cost): Remove.
1757 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
1758 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
1759 (WCHAR_TYPE): Ditto.
1760 (PTRDIFF_TYPE): Ditto.
1761 (ADJUST_INSN_LENGTH): New macro.
1762 (FIXED_REGISTERS): Add CC registers.
1763 (CALL_USED_REGISTERS): Ditto.
1765 (REG_CLASS_NAMES): Ditto.
1766 (REG_CLASS_CONTENTS): Ditto.
1767 (SELECT_CC_MODE): Use new function.
1768 (TARGET_FLAGS_REGNUM): New macro.
1769 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
1770 (cc0_reg_rtx): Remove.
1771 (CC_STATUS_MDEP): Remove.
1772 (CC_STATUS_MDEFP_INIT): Remove.
1773 (CC_IN_FPU): Remove.
1774 (NOTICE_UPDATE_CC): Remove.
1775 (REGISTER_NAMES): Add CC registers.
1776 (BRANCH_COST): Change to constant 1.
1777 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
1779 * config/pdp11/pdp11.opt (mbcopy): Remove.
1780 (mbcopy-builtin): Remove.
1781 (mbranch-cheap): Remove.
1782 (mbranch-expensive): Remove.
1783 * config/pdp11/predicates.md (expand_shift_operand): Update to
1784 match shift code generation.
1785 (ccnz_operator): New predicate.
1786 * doc/invoke.texi (PDP-11 Options): Remove deleted options
1787 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
1788 Remove non-existent option -mabshi, -mno-abshi. Document mutually
1790 * doc/md.texi (PDP-11): Document new D and h constraints. Update
1791 description of O constraint.
1793 2018-06-27 Jeff Law <law@redhat.com>
1794 Austin Law <austinklaw@gmail.com>
1796 * config/v850/v850.md (addsi3_set_flags): New pattern.
1797 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
1798 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
1799 (zero_extendhisi2_v850_set_flags): Likewise.
1800 (zero_extendqisi2_v850_set_flags): Likewise.
1801 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
1802 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
1803 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
1805 * config/v850/v850-protos.h (notice_update_cc): Remove.
1806 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
1807 (v850_print_operand): Handle 'D' and "d".
1808 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
1809 Add handling of arithmetic/logical operations compared against zero.
1810 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
1811 Do not look at v850_compare_op, instead get mode from last argument.
1812 (v850_gen_compare): Remove
1813 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
1814 after reload for prologue insns.
1815 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
1817 (construct_save_jarl): Likewise.
1818 (TARGET_FLAGS_REGNUM): Define.
1819 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
1820 (NOTICE_UPDATE_CC): Remove.
1821 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
1822 than cc0. Conditionalize on reload_completed.
1823 (cmpsi_insn, setfcc_insn): Likewise.
1824 (tst1 splitter): Turn into define_and_split which sets the flags
1826 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
1827 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
1828 (cstoresf4, cstoredf4): Clobber the flags.
1829 (cmpsi, cmpsf, cmpdf): Remove expanders.
1830 (setf_insn): Remove pattern.
1831 (addsi3): Turn into define_and_split which clobbers the flags after
1832 reload and a suitable pattern (addsi3_clobber_flags) for use after
1834 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
1835 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
1836 (ashrsi3, ashrsi3_v850e2): Likewise.
1837 (bins): Clobber the flags.
1838 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
1839 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
1840 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
1841 (call_value_internal_short, call_value_internal_long): Likewise.
1842 (callt_save_interrupt, callt_return_interrupt): Likewise.
1843 (save_interrupt, return_interrupt): Likewise.
1844 (callt_save_all_interrupt, save_all_interrupt): Likewise.
1845 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
1846 (restore_all_interrupt, _restore_all_interrupt): Likewise.
1847 (All FP comparisons): Only allow after reload has completed.
1849 (divh, divhu): Tweak output template.
1850 (branch_z_normal, branch_z_invert): Remove
1851 (branch_nz_normal, branch_nz_invert): Likewise.
1852 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
1854 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
1855 * config/v850/v850.c (notice_update_cc): Remove.
1856 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
1857 (CC_NO_CARRY): Likewise.
1858 (NOTICE_UPDATE_CC): Define to nothing.
1859 * config/v850/v850.md: Remove block comment on cc0 handling
1860 Remove "cc" attribute from all patterns. Remove cc_status handling
1861 from all patterns. Minor formatting fixes.
1863 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1865 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
1866 (cortex-a76.cortex-a55): Likewise.
1867 * config/aarch64/aarch64-tune.md: Regenerate.
1868 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
1869 cortex-a76.cortex-a55.
1871 2018-06-27 Jeff Law <law@redhat.com>
1873 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
1874 (MULTILIB_DIRNAMES): Similarly.
1876 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
1878 * gimple.h (gimple_return_retbnd): Delete.
1879 (gimple_return_set_retbnd): Likewise.
1880 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
1881 gimple_return_set_retbnd.
1882 * gimple-pretty-print.c (dump_gimple_return): Remove call to
1883 gimple_return_retbnd and adjust.
1884 * tree-inline.h (struct copy_body_data): Remove retbnd field.
1885 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
1886 Explicitly return NULL in a couple more cases. Move assertion
1887 on debug statements and remove unreachable code.
1888 (reset_debug_binding): Do not test id->retbnd.
1889 (expand_call_inline): Do not set it.
1891 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
1893 * configure.ac: Add --disable-gcov option.
1894 * configure: Regenerate.
1895 * Makefile.in: Honour @enable_gcov@.
1896 * doc/install.texi: Document --disable-gcov.
1898 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1900 * config/arm/arm-cpus.in (cortex-a76): New entry.
1901 (cortex-a76.cortex-a55): Likewise.
1902 * config/arm/arm-tables.opt: Regenerate.
1903 * config/arm/arm-tune.md: Likewise.
1904 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
1905 * doc/invoke.texi (ARM Options): Document cortex-a76 and
1906 cortex-a76.cortex-a55.
1908 2018-06-27 Tamar Christina <tamar.christina@arm.com>
1911 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
1913 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
1915 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
1917 (EPILOGUE_USES): Likewise.
1919 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
1921 * tree-inline.c (remap_location): New function extracted from...
1922 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
1923 (copy_phis_for_bb): ...here. Call remap_location.
1924 (copy_cfg_body): Adjust call to copy_edges_for_bb.
1926 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
1928 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
1929 unaligned vsx for 16B memset.
1931 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
1934 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
1935 ieee128_float_type_node to long_double_type_node unless
1936 TARGET_LONG_DOUBLE_128 is set.
1938 2018-06-26 David Malcolm <dmalcolm@redhat.com>
1940 * cfgloop.c (get_loop_location): Convert return type from
1941 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
1942 by implicit construction from rtx_insn *, and using
1943 dump_user_location_t::from_function_decl for the fallback case.
1944 * cfgloop.h (get_loop_location): Convert return type from
1945 location_t to dump_user_location_t.
1946 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
1947 dump_printf_loc to pass in a dump_location_t rather than a
1948 location_t, via the gimple stmt.
1949 * coverage.c (get_coverage_counts): Update calls to
1950 dump_printf_loc to pass in dump_location_t rather than a
1952 * doc/optinfo.texi (Dump types): Convert example of
1953 dump_printf_loc from taking "locus" to taking "insn". Update
1954 description of the "_loc" calls to cover dump_location_t.
1955 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
1957 (dump_user_location_t::dump_user_location_t): New constructors,
1958 from gimple *stmt and rtx_insn *.
1959 (dump_user_location_t::from_function_decl): New function.
1960 (dump_loc): Make static.
1961 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
1962 const dump_location_t &.
1963 (dump_generic_expr_loc): Delete.
1964 (dump_printf_loc): Convert param "loc" from location_t to
1965 const dump_location_t &.
1966 (selftest::test_impl_location): New function.
1967 (selftest::dumpfile_c_tests): New function.
1968 * dumpfile.h: Include "profile-count.h".
1969 (class dump_user_location_t): New class.
1970 (struct dump_impl_location_t): New struct.
1971 (class dump_location_t): New class.
1972 (dump_printf_loc): Convert 2nd param from source_location to
1973 const dump_location_t &.
1974 (dump_generic_expr_loc): Delete.
1975 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
1976 const dump_location_t &.
1977 * gimple-fold.c (fold_gimple_assign): Update call to
1978 dump_printf_loc to pass in a dump_location_t rather than a
1979 location_t, via the gimple stmt.
1980 (gimple_fold_call): Likewise.
1981 * gimple-loop-interchange.cc
1982 (loop_cand::analyze_iloop_reduction_var): Update for change to
1983 check_reduction_path.
1984 (tree_loop_interchange::interchange): Update for change to
1986 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
1987 change in return-type of find_loop_location.
1988 (graphite_regenerate_ast_isl): Likewise.
1989 * graphite-optimize-isl.c (optimize_isl): Likewise.
1990 * graphite.c (graphite_transform_loops): Likewise.
1991 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
1992 pass in a dump_location_t rather than a location_t, via the
1994 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
1995 * ipa.c (walk_polymorphic_call_targets): Likewise.
1996 * loop-unroll.c (report_unroll): Convert "locus" param from
1997 location_t to dump_location_t.
1998 (decide_unrolling): Update for change to get_loop_location's
2000 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
2001 location_t to dump_user_location_t.
2002 (grid_find_single_omp_among_assignments_1): Updates calls to
2003 dump_printf_loc to pass in a dump_location_t rather than a
2004 location_t, via the gimple stmt.
2005 (grid_parallel_clauses_gridifiable): Convert "tloc" from
2006 location_t to dump_location_t. Updates calls to dump_printf_loc
2007 to pass in a dump_location_t rather than a location_t, via the
2009 (grid_inner_loop_gridifiable_p): Likewise.
2010 (grid_dist_follows_simple_pattern): Likewise.
2011 (grid_gfor_follows_tiling_pattern): Likewise.
2012 (grid_target_follows_gridifiable_pattern): Likewise.
2013 (grid_attempt_target_gridification): Convert initialization
2014 of local "grid" from memset to zero-initialization; FIXME: does
2015 this require C++11? Update call to dump_printf_loc to pass in a
2016 optinfo_location rather than a location_t, via the gimple stmt.
2017 * profile.c (read_profile_edge_counts): Updates call to
2018 dump_printf_loc to pass in a dump_location_t rather than a
2020 (compute_branch_probabilities): Likewise.
2021 * selftest-run-tests.c (selftest::run_tests): Call
2023 * selftest.h (dumpfile_c_tests): New decl.
2024 * tree-loop-distribution.c (pass_loop_distribution::execute):
2025 Update for change in return type of find_loop_location.
2026 * tree-parloops.c (parallelize_loops): Likewise.
2027 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
2028 "locus" from location_t to dump_user_location_t.
2029 (canonicalize_loop_induction_variables): Likewise.
2030 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
2031 for change in return type of find_loop_location.
2032 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
2033 to dump_printf_loc to pass in a dump_location_t rather than a
2034 location_t, via the stmt.
2035 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
2037 * tree-vect-loop-manip.c (find_loop_location): Convert return
2038 type from source_location to dump_user_location_t.
2039 (vect_do_peeling): Update for above change.
2040 (vect_loop_versioning): Update for change in type of
2042 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
2043 from location_t to dump_user_location_t.
2044 (vect_estimate_min_profitable_iters): Update for change in type
2046 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
2047 location_t to dump_location_t.
2048 (vect_slp_bb): Update for change in type of vect_location.
2049 * tree-vectorizer.c (vect_location): Convert from source_location
2050 to dump_user_location_t.
2051 (try_vectorize_loop_1): Update for change in vect_location's type.
2052 (vectorize_loops): Likewise.
2053 (increase_alignment): Likewise.
2054 * tree-vectorizer.h (vect_location): Convert from source_location
2055 to dump_user_location_t.
2056 (find_loop_location): Convert return type from source_location to
2057 dump_user_location_t.
2058 (check_reduction_path): Convert 1st param from location_t to
2059 dump_user_location_t.
2060 * value-prof.c (check_counter): Update call to dump_printf_loc to
2061 pass in a dump_user_location_t rather than a location_t; update
2062 call to error_at for change in type of "locus".
2063 (check_ic_target): Update call to dump_printf_loc to
2064 pass in a dump_user_location_t rather than a location_t, via the
2067 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
2069 * config/s390/s390.h (enum processor_flags): Do not use
2070 default tune parameter when -march was specified.
2072 2018-06-26 Jakub Jelinek <jakub@redhat.com>
2075 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
2076 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
2079 2018-06-26 Richard Biener <rguenther@suse.de>
2081 PR tree-optimization/86287
2083 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
2084 (vect_analyze_loop): Initialize n_stmts.
2086 2018-06-26 Richard Biener <rguenther@suse.de>
2089 * fold-const.c (fold_convertible_p): Pointer extension
2092 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
2095 * dwarf2out.c (loc_list_has_views): Adjust comments.
2096 (dw_loc_list): Split single cross-partition range with
2099 2018-06-25 Jeff Law <law@redhat.com>
2101 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
2102 on -mbig-switch by default.
2104 * config/v850/predicates.md (const_float_1_operand): Fix match_code
2106 (const_float_0_operand): Remove unused predicate.
2107 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
2108 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
2109 (recipsf2): New expander. Original pattern now called
2111 (recipdf2, recipdf2_insn): Similarly.
2112 (rsqrtsf2, rsqrtsf2_insn): Similarly
2113 (rsqrtdf2, rsqrtdf2_insn): Similarly
2115 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
2117 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
2118 Simplify logic for FreeBSD (twice).
2120 2018-06-25 Martin Sebor <msebor@redhat.com>
2122 PR tree-optimization/86204
2123 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
2124 a strnlen result if it's less than the length of the string.
2126 2018-06-25 Martin Sebor <msebor@redhat.com>
2128 PR tree-optimization/85700
2129 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
2130 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
2131 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
2133 2018-06-25 Martin Sebor <msebor@redhat.com>
2135 * doc/extend.texi (Zero-length arrays): Update and clarify.
2137 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
2139 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
2140 added IEEE/IBM long double multilib support on PowerPC little
2141 endian Linux systems.
2142 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
2143 (MULTILIB_DEFAULTS): Likewise.
2144 * config/rs6000/rs6000.c (rs6000_option_override_internal):
2146 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
2147 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
2148 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
2150 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
2153 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
2154 (REORDER_45): Likewise.
2156 2018-06-25 Jeff Law <law@redhat.com>
2158 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
2159 dividend to 32 bits. Adjust length.
2160 (udivmodhi4): Cleanup output template. Fix length.
2162 2018-06-25 Carl Love <cel@us.ibm.com>
2164 * config/rs6000/vsx.md: Change word selector to prefered location.
2166 2018-06-25 Richard Biener <rguenther@suse.de>
2168 PR tree-optimization/86304
2169 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
2170 epilogue-if-converted loops as well.
2172 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
2174 * lto-section-out.c (lto_begin_section): Do not print section
2175 name for noaddr and unnumbered dumps.
2177 2018-06-25 Richard Biener <rguenther@suse.de>
2179 * tree-vectorizer.h (struct vec_info_shared): New structure
2180 with parts split out from struct vec_info and loop_nest from
2181 struct _loop_vec_info.
2182 (struct vec_info): Adjust accordingly.
2183 (struct _loop_vec_info): Likewise.
2184 (LOOP_VINFO_LOOP_NEST): Adjust.
2185 (LOOP_VINFO_DATAREFS): Likewise.
2186 (LOOP_VINFO_DDRS): Likewise.
2187 (struct _bb_vec_info): Likewise.
2188 (BB_VINFO_DATAREFS): Likewise.
2189 (BB_VINFO_DDRS): Likewise.
2190 (struct _stmt_vec_info): Add dr_aux member.
2191 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
2192 (DR_MISALIGNMENT_UNINITIALIZED): New.
2193 (set_dr_misalignment): Adjust.
2194 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
2195 (vect_analyze_loop): Adjust prototype.
2196 (vect_analyze_loop_form): Likewise.
2197 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
2198 Compute dependences lazily.
2199 (vect_record_base_alignments): Use shared datarefs/ddrs.
2200 (vect_verify_datarefs_alignment): Likewise.
2201 (vect_analyze_data_refs_alignment): Likewise.
2202 (vect_analyze_data_ref_accesses): Likewise.
2203 (vect_analyze_data_refs): Likewise.
2204 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
2205 constructor parameter for shared part.
2206 (vect_analyze_loop_form): Pass in shared part and adjust.
2207 (vect_analyze_loop_2): Pass in storage for the number of
2208 stmts. Move loop nest finding to the caller. Compute
2210 (vect_analyze_loop): Pass in shared part.
2211 (vect_transform_loop): Verify shared datarefs are unchanged.
2212 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
2213 constructor parameter for shared part.
2214 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
2215 (vect_slp_bb): Verify shared datarefs are unchanged before
2217 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
2219 (new_stmt_vec_info): Initialize DR_AUX misalignment to
2220 DR_MISALIGNMENT_UNINITIALIZED.
2221 * tree-vectorizer.c (vec_info::vec_info): Add constructor
2222 parameter for shared part.
2223 (vec_info::~vec_info): Adjust.
2224 (vec_info_shared::vec_info_shared): New.
2225 (vec_info_shared::~vec_info_shared): Likewise.
2226 (vec_info_shared::save_datarefs): Likewise.
2227 (vec_info_shared::check_datarefs): Likewise.
2228 (try_vectorize_loop_1): Construct shared part live for analyses
2229 of a single loop for multiple vector sizes.
2230 * tree-parloops.c (gather_scalar_reductions): Adjust.
2232 2018-06-25 Richard Biener <rguenther@suse.de>
2234 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
2235 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
2236 (vect_analyze_data_refs): Remove similar code from here and
2237 simplify accordingly.
2239 2018-06-25 Richard Biener <rguenther@suse.de>
2241 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
2242 for reverse storage order accesses rather than asserting
2243 they cannot happen here.
2245 2018-06-25 Tom de Vries <tdevries@suse.de>
2248 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
2249 Use data16 instead of .byte for insn prefix.
2251 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
2254 * parser.c (make_char_string_pack): Pass this literal chars
2255 through cpp_interpret_string.
2256 (cp_parser_userdef_numeric_literal): Check the result of
2257 make_char_string_pack.
2259 2018-06-24 Maya Rashish <coypu@sdf.org>
2261 * ginclude/stddef.h: Simplify conditions around avoiding
2262 re-definition of __size_t.
2264 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
2266 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
2267 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
2269 2018-06-22 Maya Rashish <coypu@sdf.org>
2271 * doc/invoke.texi (mno-fancy-math-387): Update for changes
2272 made to OpenBSD and NetBSD through the years.
2274 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
2276 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
2277 behavior of vec_pack (vector double, vector double) to match
2278 behavior of vec_float2 (vector double, vector double).
2280 2018-06-22 Olivier Hainque <hainque@adacore.com>
2282 * gimplify.c (gimplify_function_tree): Prevent creation
2283 of a trampoline for the address of the current function
2284 passed to entry/exit instrumentation hooks.
2286 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
2289 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
2292 2018-06-22 Martin Liska <mliska@suse.cz>
2294 PR tree-optimization/86263
2295 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
2296 Bail out if is_enabled is false.
2297 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
2299 (jump_table_cluster::is_enabled): New function.
2301 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
2303 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
2304 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
2305 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
2306 (lto_input_ts_binfo_tree_pointers): Likewise.
2307 * tree-streamer-out.c (streamer_write_tree_bitfields,
2308 write_ts_binfo_tree_pointers): Likewise.
2309 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
2311 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
2313 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
2315 2018-06-22 Martin Liska <mliska@suse.cz>
2317 * symbol-summary.h (get): Make it pure and inline move
2318 functionality from ::get function.
2319 (get): Remove and inline into ::get and ::get_create.
2320 (get_create): Move code from ::get function.
2322 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2325 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
2326 -x assembler-with-cpp.
2328 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2330 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
2331 _FILE_OFFSET_BITS=64 for C++.
2333 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
2335 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
2336 conversion insn that shows up when pr85657-3.c is compiled using
2337 IEEE 128-bit long double.
2338 (neg<mode>2_internal): Use the correct mode to check whether the
2339 mode is IBM extended.
2340 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
2341 multiply and divide external functions from being created more
2344 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
2346 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
2348 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
2349 the edge can be forwarded.
2350 (cfg_layout_merge_blocks): Likewise.
2352 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
2354 * except.c (finish_eh_generation): Commit edge insertions only after
2355 the EH edges have been redirected from post-landing to landing pads.
2357 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
2359 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
2360 create_tmp_var_for to create the FRAME decl.
2361 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
2363 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
2365 * tree-inline.c (copy_edges_for_bb): Minor tweak.
2366 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
2367 debug statement when resetting its value.
2368 (expand_call_inline): Copy the locus of the call onto the assignment
2369 of the return value, if any. Use local variable in more cases.
2371 2018-06-21 Martin Liska <mliska@suse.cz>
2373 * ipa-pure-const.c (propagate_nothrow): Use
2374 funct_state_summaries->get.
2375 (dump_malloc_lattice): Likewise.
2376 (propagate_malloc): Likewise.
2378 2018-06-21 Richard Biener <rguenther@suse.de>
2380 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
2381 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
2382 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
2384 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
2385 BLOCK_ABSTRACT_ORIGIN unconditionally.
2387 2018-06-21 David Malcolm <dmalcolm@redhat.com>
2389 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
2391 * ipa-reference.c (ipa_reference_c_finalize): Delete
2392 ipa_ref_opt_sum_summaries and set it to NULL.
2394 2018-06-21 Tom de Vries <tdevries@suse.de>
2396 PR tree-optimization/85859
2397 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
2398 test with comment from bb_no_side_effects_p.
2400 2018-06-21 Richard Biener <rguenther@suse.de>
2402 PR tree-optimization/86232
2403 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
2404 max for constant niter.
2406 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
2408 * config/aarch64/aarch64-simd.md
2409 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
2411 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
2413 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
2414 Make opernads of the unspec commutative.
2416 2018-06-21 Richard Biener <rguenther@suse.de>
2418 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
2419 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
2420 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
2421 (vect_analyze_data_ref_dependence): Re-order checks to deal with
2423 (vect_record_base_alignments): Do not record base alignment
2424 for gathers or scatters.
2425 (vect_compute_data_ref_alignment): Drop return value that is always
2426 true. Bail out early for gathers or scatters.
2427 (vect_enhance_data_refs_alignment): Bail out early for gathers
2429 (vect_find_same_alignment_drs): Likewise.
2430 (vect_analyze_data_refs_alignment): Remove dead code.
2431 (vect_slp_analyze_and_verify_node_alignment): Likewise.
2432 (vect_analyze_data_refs): For possible gathers or scatters do
2433 not create an alternate DR, just check their possible validity
2434 and mark them. Adjust DECL_NONALIASED handling to not rely
2436 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
2437 update inits of gathers or scatters.
2438 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
2439 Also copy gather/scatter flag to pattern vinfo.
2441 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
2443 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
2444 behavior of vec_packsu (vector unsigned long long, vector unsigned
2445 long long) to match behavior of vec_packs with same signature.
2447 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
2448 Thomas Schwinge <thomas@codesourcery.com>
2449 Cesar Philippidis <cesar@codesourcery.com>
2451 * gimplify.c (gimplify_scan_omp_clauses): Add support for
2452 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
2453 (gimplify_adjust_omp_clauses): Likewise.
2454 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
2455 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
2456 (gimplify_omp_target_update): Update handling of acc update and
2458 * omp-low.c (install_var_field): Remove unused parameter
2459 base_pointers_restrict.
2460 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
2461 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
2463 (omp_target_base_pointers_restrict_p): Delete.
2464 (scan_omp_target): Update call to scan_sharing_clauses.
2465 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
2467 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
2468 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
2469 (convert_local_omp_clauses): Likewise.
2470 * tree-pretty-print.c (dump_omp_clause): Likewise.
2471 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
2473 (omp_clause_code_name): Likewise.
2475 2018-06-20 Jakub Jelinek <jakub@redhat.com>
2478 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
2481 PR tree-optimization/86231
2482 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
2483 anti-range don't overwrite *vr0min before using it to compute *vr0max.
2485 2018-06-20 Tom de Vries <tdevries@suse.de>
2487 PR tree-optimization/86097
2488 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
2489 iv type if signedness of iv type is not the same as that of *nit.
2491 2018-06-20 Jakub Jelinek <jakub@redhat.com>
2493 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
2494 EDGE_EH edges, verify they are all EDGE_EH.
2496 2018-06-20 Maya Rashish <coypu@sdf.org>
2498 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
2500 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2502 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
2503 * config/aarch64/aarch64.c (xgene1_tunings): Add
2504 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
2505 (aarch64_mode_valid_for_sched_fusion_p):
2506 Allow 16-byte modes.
2507 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
2508 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
2510 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
2512 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
2513 * config/aarch64/iterators.md (VQ2): New mode iterator.
2515 2018-06-20 Martin Liska <mliska@suse.cz>
2517 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
2518 Change default ratio from 10 to 8.
2520 2018-06-20 Martin Liska <mliska@suse.cz>
2522 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
2524 (bit_test_cluster::find_bit_tests): Likewise.
2525 (switch_decision_tree::analyze_switch_statement): Find clusters.
2526 * tree-switch-conversion.h (struct jump_table_cluster): Document
2529 2018-06-20 Martin Liska <mliska@suse.cz>
2531 * tree-switch-conversion.c (switch_conversion::collect):
2532 Record m_uniq property.
2533 (switch_conversion::expand): Bail out for special conditions.
2534 (group_cluster::~group_cluster): New.
2535 (group_cluster::group_cluster): Likewise.
2536 (group_cluster::dump): Likewise.
2537 (jump_table_cluster::emit): New.
2538 (switch_decision_tree::fix_phi_operands_for_edges): New.
2539 (struct case_node): Remove struct.
2540 (jump_table_cluster::can_be_handled): New.
2541 (case_values_threshold): Moved to header.
2542 (reset_out_edges_aux): Likewise.
2543 (jump_table_cluster::is_beneficial): New.
2544 (bit_test_cluster::can_be_handled): Likewise.
2545 (add_case_node): Remove.
2546 (bit_test_cluster::is_beneficial): New.
2547 (case_bit_test::cmp): New.
2548 (bit_test_cluster::emit): New.
2549 (expand_switch_as_decision_tree_p): Remove.
2550 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
2551 (fix_phi_operands_for_edge): Likewise.
2552 (switch_decision_tree::analyze_switch_statement): New.
2553 (compute_cases_per_edge): Move ...
2554 (switch_decision_tree::compute_cases_per_edge): ... here.
2555 (try_switch_expansion): Likewise.
2556 (switch_decision_tree::try_switch_expansion): Likewise.
2557 (record_phi_operand_mapping): Likewise.
2558 (switch_decision_tree::record_phi_operand_mapping): Likewise.
2559 (emit_case_decision_tree): Likewise.
2560 (switch_decision_tree::emit): Likewise.
2561 (balance_case_nodes): Likewise.
2562 (switch_decision_tree::balance_case_nodes): Likewise.
2563 (dump_case_nodes): Likewise.
2564 (switch_decision_tree::dump_case_nodes): Likewise.
2565 (emit_jump): Likewise.
2566 (switch_decision_tree::emit_jump): Likewise.
2567 (emit_cmp_and_jump_insns): Likewise.
2568 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
2569 (emit_case_nodes): Likewise.
2570 (switch_decision_tree::emit_case_nodes): Likewise.
2571 (conditional_probability): Remove.
2572 * tree-switch-conversion.h (enum cluster_type): New.
2574 (struct cluster): Likewise.
2575 (cluster::cluster): Likewise.
2576 (struct simple_cluster): Likewise.
2577 (simple_cluster::simple_cluster): Likewise.
2578 (struct group_cluster): Likewise.
2579 (struct jump_table_cluster): Likewise.
2580 (struct bit_test_cluster): Likewise.
2581 (struct min_cluster_item): Likewise.
2582 (struct case_tree_node): Likewise.
2583 (case_tree_node::case_tree_node): Likewise.
2584 (jump_table_cluster::case_values_threshold): Likewise.
2585 (struct case_bit_test): Likewise.
2586 (struct switch_decision_tree): Likewise.
2587 (struct switch_conversion): Likewise.
2588 (switch_decision_tree::reset_out_edges_aux): Likewise.
2590 2018-06-20 Martin Liska <mliska@suse.cz>
2592 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
2593 (hoist_edge_and_branch_if_true): Likewise.
2594 (expand_switch_using_bit_tests_p): Likewise.
2595 (struct case_bit_test): Likewise.
2596 (case_bit_test_cmp): Likewise.
2597 (emit_case_bit_tests): Likewise.
2598 (switch_conversion::switch_conversion): New class.
2599 (struct switch_conv_info): Remove old struct.
2600 (collect_switch_conv_info): More to ...
2601 (switch_conversion::collect): ... this.
2602 (check_range): Likewise.
2603 (switch_conversion::check_range): Likewise.
2604 (check_all_empty_except_final): Likewise.
2605 (switch_conversion::check_all_empty_except_final): Likewise.
2606 (check_final_bb): Likewise.
2607 (switch_conversion::check_final_bb): Likewise.
2608 (create_temp_arrays): Likewise.
2609 (switch_conversion::create_temp_arrays): Likewise.
2610 (free_temp_arrays): Likewise.
2611 (gather_default_values): Likewise.
2612 (switch_conversion::gather_default_values): Likewise.
2613 (build_constructors): Likewise.
2614 (switch_conversion::build_constructors): Likewise.
2615 (constructor_contains_same_values_p): Likewise.
2616 (switch_conversion::contains_same_values_p): Likewise.
2617 (array_value_type): Likewise.
2618 (switch_conversion::array_value_type): Likewise.
2619 (build_one_array): Likewise.
2620 (switch_conversion::build_one_array): Likewise.
2621 (build_arrays): Likewise.
2622 (switch_conversion::build_arrays): Likewise.
2623 (gen_def_assigns): Likewise.
2624 (switch_conversion::gen_def_assigns): Likewise.
2625 (prune_bbs): Likewise.
2626 (switch_conversion::prune_bbs): Likewise.
2627 (fix_phi_nodes): Likewise.
2628 (switch_conversion::fix_phi_nodes): Likewise.
2629 (gen_inbound_check): Likewise.
2630 (switch_conversion::gen_inbound_check): Likewise.
2631 (process_switch): Use the newly created class.
2632 (switch_conversion::expand): New.
2633 (switch_conversion::~switch_conversion): New.
2634 * tree-switch-conversion.h: New file.
2636 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
2638 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
2639 tree-vect-patterns.c.
2640 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
2641 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
2642 (vect_recog_sad_pattern): Likewise.
2643 (vect_recog_widen_sum_pattern): Likewise.
2644 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
2645 (vect_recog_widen_shift_pattern): Remove the type_in argument.
2646 (vect_recog_rotate_pattern): Likewise.
2647 (vect_recog_mult_pattern): Likewise.
2648 (vect_recog_vector_vector_shift_pattern): Likewise.
2649 (vect_recog_divmod_pattern): Likewise.
2650 (vect_recog_mixed_size_cond_pattern): Likewise.
2651 (vect_recog_bool_pattern): Likewise.
2652 (vect_recog_mask_conversion_pattern): Likewise.
2653 (vect_try_gather_scatter_pattern): Likewise.
2654 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
2655 (vect_recog_over_widening_pattern): Likewise.
2656 (vect_recog_gather_scatter_pattern): Likewise.
2657 (vect_recog_func_ptr): Move from tree-vectorizer.h
2658 (vect_vect_recog_func_ptrs): Move further down the file.
2659 (vect_recog_func): Likewise. Remove the third argument.
2660 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
2661 (vect_pattern_recog_1): Expect the pattern function to do any
2662 necessary target tests. Also expect it to provide a vector type.
2663 Remove the type_in handling.
2665 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
2667 * tree-vect-patterns.c (vect_pattern_detected): New function.
2668 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
2669 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
2670 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
2671 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
2672 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
2673 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
2674 (vect_recog_mask_conversion_pattern)
2675 (vect_try_gather_scatter_pattern): Likewise.
2677 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
2679 * tree-vect-patterns.c (vect_get_internal_def): New function.
2680 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
2681 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
2682 (search_type_for_mask_1): Use it.
2684 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
2686 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
2687 redundant WIDEN_SUM_EXPR handling.
2688 (vect_recog_sad_pattern): Likewise.
2690 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
2692 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
2693 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
2694 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
2695 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
2696 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
2698 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
2700 * tree-vect-stmts.c (vectorizable_call): Make sure that we
2701 use the stmt_vec_info of the original bb statement for the
2702 new zero assignment, even if the call is part of a pattern.
2704 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
2706 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
2707 that the sequence is attached to the original statement rather
2708 than the pattern statement.
2709 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
2710 PATTERN_DEF_SEQ from the original statement rather than
2711 the main pattern statement.
2712 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
2713 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
2714 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
2716 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
2718 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
2719 definition statements before the early exit for statements that aren't
2721 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
2723 (vect_transform_loop): ...here. Process pattern definition
2724 statements without first checking whether the main pattern
2725 statement is live or relevant.
2727 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
2729 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
2730 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
2732 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
2734 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
2735 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
2736 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
2737 (expand_block_compare): Change select_block_compare_mode call.
2738 (expand_strncmp_align_check): Use new functions, fix comment.
2739 (emit_final_str_compare_gpr): New function.
2740 (expand_strn_compare): Refactor and clean up code.
2741 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
2743 2018-06-19 Tony Reix <tony.reix@atos.com>
2744 Damien Bergamini <damien.bergamini@atos.com>
2745 David Edelsohn <dje.gcc@gmail.com>
2747 * collect2.c (static_obj): New variable.
2748 (static_libs): New variable.
2749 (is_in_list): Uncomment declaration.
2750 (main): Track AIX libraries linked statically.
2751 (is_in_list): Uncomment definition.
2752 (scan_prog_file): Don't add AIX shared libraries initializer
2753 to constructor list if linking statically.
2755 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
2757 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
2759 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
2761 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
2763 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
2766 2018-06-19 Martin Liska <mliska@suse.cz>
2768 * config/i386/i386.c (ix86_can_inline_p): Do not use
2769 ipa_fn_summaries::get_create.
2770 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
2772 (devirtualization_time_bonus): Likewise.
2773 (ipcp_propagate_stage): Likewise.
2774 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
2775 (edge_set_predicate): Likewise.
2776 (evaluate_conditions_for_known_args): Likewise.
2777 (evaluate_properties_for_edge): Likewise.
2778 (ipa_call_summary::reset): Tranform to ...
2779 (ipa_call_summary::~ipa_call_summary): ... this.
2780 (ipa_fn_summary::reset): Transform to ...
2781 (ipa_fn_summary::~ipa_fn_summary): ... this.
2782 (ipa_fn_summary_t::remove): Rename to ...
2783 (ipa_fn_summary_t::remove_callees): ... this.
2784 (ipa_fn_summary_t::duplicate): Use placement new
2785 instead of memory copy.
2786 (ipa_call_summary_t::duplicate): Likewise.
2787 (ipa_call_summary_t::remove): Remove.
2788 (dump_ipa_call_summary): Change get_create to get.
2789 (ipa_dump_fn_summary): Dump only when summary exists.
2790 (analyze_function_body): Use symbol_summary::get instead
2792 (compute_fn_summary): Likewise.
2793 (estimate_edge_devirt_benefit): Likewise.
2794 (estimate_edge_size_and_time): Likewise.
2795 (inline_update_callee_summaries): Likewise.
2796 (remap_edge_change_prob): Likewise.
2797 (remap_edge_summaries): Likewise.
2798 (ipa_merge_fn_summary_after_inlining): Likewise.
2799 (write_ipa_call_summary): Likewise.
2800 (ipa_fn_summary_write): Likewise.
2801 (ipa_free_fn_summary): Likewise.
2802 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
2803 (struct ipa_call_summary): Likewise.
2804 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
2806 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
2807 (estimate_size_after_inlining): Likewise.
2808 (estimate_growth): Likewise.
2809 (growth_likely_positive): Likewise.
2810 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
2811 (inline_call): Likewise.
2812 * ipa-inline.c (caller_growth_limits): Likewise.
2813 (can_inline_edge_p): Likewise.
2814 (can_inline_edge_by_limits_p): Likewise.
2815 (compute_uninlined_call_time): Likewise.
2816 (compute_inlined_call_time): Likewise.
2817 (want_inline_small_function_p): Likewise.
2818 (edge_badness): Likewise.
2819 (update_caller_keys): Likewise.
2820 (update_callee_keys): Likewise.
2821 (inline_small_functions): Likewise.
2822 (inline_to_all_callers_1): Likewise.
2823 (dump_overall_stats): Likewise.
2824 (early_inline_small_functions): Likewise.
2825 (early_inliner): Likewise.
2826 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
2827 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
2828 * ipa-pure-const.c (malloc_candidate_p): Likewise.
2829 * ipa-split.c (execute_split_functions): Likewise.
2830 * symbol-summary.h: Likewise.
2831 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
2833 2018-06-19 Richard Biener <rguenther@suse.de>
2835 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
2836 (vectorize_loops): ... here. Fix dbgcnt handling.
2837 (try_vectorize_loop): Wrap try_vectorize_loop_1.
2839 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
2842 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
2843 ieee128 argument takes up only one (vector) register, not two (floating
2846 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
2848 * gimplify.c (gimplify_init_constructor): Really never clear for an
2849 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
2851 2018-06-19 Richard Biener <rguenther@suse.de>
2853 PR tree-optimization/86179
2854 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
2855 after failed recognition.
2857 2018-06-18 Martin Sebor <msebor@redhat.com>
2860 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
2861 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
2862 Handle integer subtraction.
2863 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
2864 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
2866 2018-06-18 David Malcolm <dmalcolm@redhat.com>
2868 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
2869 param from rtx to rtx_insn *.
2870 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
2872 (frv_ifcvt_modify_insn): Likwise.
2873 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
2874 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
2875 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
2876 as_a <rtx_insn *> cast to local "unprotected_region" once
2877 it's been established that it's not NULL or pc_rtx.
2878 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
2879 param "sethi" from rtx to rtx_insn *.
2880 (nds32_group_float_insns): Likewise for param "insn".
2881 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
2883 (vax_output_int_subtract): Likewise.
2884 * config/vax/vax.c (vax_output_int_add): Likewise for param
2886 (vax_output_int_subtract): Likewise.
2887 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
2888 (emit_pattern_after): Likewise for param "after".
2889 (emit_insn_after): Likewise.
2890 (emit_jump_insn_after): Likewise.
2891 (emit_call_insn_after): Likewise.
2892 (emit_debug_insn_after): Likewise.
2893 (emit_pattern_before): Likewise for param "before".
2894 (emit_insn_before): Likewise.
2895 (emit_jump_insn_before): Likewise.
2896 * final.c (get_insn_template): Likewise for param "insn", removing
2898 * output.h (get_insn_template): Likewise for 2nd param.
2899 * rtl.h (emit_insn_before): Likewise.
2900 (emit_jump_insn_before): Likewise.
2901 (emit_debug_insn_before_noloc): Likewise.
2902 (emit_insn_after): Likewise.
2903 (emit_jump_insn_after): Likewise.
2904 (emit_call_insn_after): Likewise.
2905 (emit_debug_insn_after): Likewise.
2906 (set_insn_deleted): Likewise for param.
2908 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
2911 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
2912 floating point modes, so that IFmode is numerically greater than
2913 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
2914 to declare the ordering. This prevents IFmode from being
2915 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
2916 machine. Include rs6000-modes.h to share the fractional values
2917 between genmodes* and the rest of the compiler.
2921 * config/rs6000/rs6000-modes.h: New file.
2922 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
2923 meaning of rs6000_long_double_size so that 126..128 selects an
2924 appropriate 128-bit floating point type.
2925 (rs6000_option_override_internal): Likewise.
2926 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
2927 (TARGET_LONG_DOUBLE_128): Change the meaning of
2928 rs6000_long_double_size so that 126..128 selects an appropriate
2929 128-bit floating point type.
2930 (LONG_DOUBLE_TYPE_SIZE): Update comment.
2931 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
2932 source and destination to match the standard usage.
2933 (truncifkf2): Likewise.
2934 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
2935 ISA 2.07 to use an explicit clobber, instead of passing in a
2937 (copysign<mode>3_soft): Likewise.
2939 2018-06-18 David Malcolm <dmalcolm@redhat.com>
2941 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
2942 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
2943 (vect_slp_analyze_instance_dependence): Likewise.
2944 (vect_enhance_data_refs_alignment): Likewise.
2945 (vect_analyze_data_refs_alignment): Likewise.
2946 (vect_slp_analyze_and_verify_instance_alignment
2947 (vect_analyze_data_ref_accesses): Likewise.
2948 (vect_prune_runtime_alias_test_list): Likewise.
2949 (vect_analyze_data_refs): Likewise.
2950 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
2951 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
2952 (vect_analyze_scalar_cycles_1): Likewise.
2953 (vect_get_loop_niters): Likewise.
2954 (vect_analyze_loop_form_1): Likewise.
2955 (vect_update_vf_for_slp): Likewise.
2956 (vect_analyze_loop_operations): Likewise.
2957 (vect_analyze_loop): Likewise.
2958 (vectorizable_induction): Likewise.
2959 (vect_transform_loop): Likewise.
2960 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
2961 * tree-vect-slp.c (vect_analyze_slp): Likewise.
2962 (vect_make_slp_decision): Likewise.
2963 (vect_detect_hybrid_slp): Likewise.
2964 (vect_slp_analyze_operations): Likewise.
2965 (vect_slp_bb): Likewise.
2966 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
2967 (vectorizable_bswap): Likewise.
2968 (vectorizable_call): Likewise.
2969 (vectorizable_simd_clone_call): Likewise.
2970 (vectorizable_conversion): Likewise.
2971 (vectorizable_assignment): Likewise.
2972 (vectorizable_shift): Likewise.
2973 (vectorizable_operation): Likewise.
2974 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
2976 2018-06-18 Martin Sebor <msebor@redhat.com>
2978 PR tree-optimization/81384
2979 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
2980 * builtins.c (expand_builtin_strnlen): New function.
2981 (expand_builtin): Call it.
2982 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
2983 * builtins.def (BUILT_IN_STRNLEN): New.
2984 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
2985 Warn for bounds in excess of maximum object size.
2986 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
2987 single-value ranges. Handle strnlen.
2988 (handle_builtin_strlen): Handle strnlen.
2989 (strlen_check_and_optimize_stmt): Same.
2990 * doc/extend.texi (Other Builtins): Document strnlen.
2992 2018-06-18 Maya Rashish <coypu@sdf.org>
2994 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
2995 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
2996 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
2998 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
3000 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
3002 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3004 * tree.c (escaped_string::escape): Replace cast to char * by
3005 const_cast<char *> (unescaped).
3007 2018-06-18 Nick Clifton <nickc@redhat.com>
3010 * tree.c (escaped_string): New class. Converts an unescaped
3011 string into its escaped equivalent.
3012 (warn_deprecated_use): Use the new class to convert the
3013 deprecation message, if present.
3014 (test_escaped_strings): New self test.
3015 (test_c_tests): Add test_escaped_strings.
3016 * doc/extend.texi (deprecated): Add a note that the
3017 deprecation message is affected by the -fmessage-length
3018 option, and that control characters will be escaped.
3019 (#pragma GCC error): Document this pragma.
3020 (#pragma GCC warning): Likewise.
3021 * doc/invoke.texi (-fmessage-length): Document this option's
3022 effect on the #warning and #error preprocessor directives and
3023 the deprecated attribute.
3025 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
3027 * tree.c (decl_value_expr_lookup): Revert latest change.
3028 (decl_value_expr_insert): Likewise.
3030 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
3032 * gimplify.c (nonlocal_vlas): Delete.
3033 (nonlocal_vla_vars): Likewise.
3034 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
3036 (gimplify_body): Do not create and destroy nonlocal_vlas.
3037 * tree-nested.c: Include diagnostic.h.
3038 (use_pointer_in_frame): Tweak.
3039 (lookup_field_for_decl): Add assertion and declare the transformation.
3040 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
3041 internal error when the reference is in a wrong context. Do not
3042 create a debug decl by default.
3043 (note_nonlocal_block_vlas): Delete.
3044 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
3045 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
3046 create a debug decl by default.
3047 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
3048 call is in a wrong context.
3049 (fixup_vla_decls): New function.
3050 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
3051 debug variables were created.
3052 * tree.c (decl_value_expr_lookup): Add checking assertion.
3053 (decl_value_expr_insert): Likewise.
3055 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
3058 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
3059 * tree-scalar-evolution.c (interpret_expr): Likewise.
3060 (expression_expensive_p): Likewise.
3061 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
3062 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
3063 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
3064 (ssa_defined_by_minus_one_stmt_p): New.
3066 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
3069 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
3070 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
3071 * dojump.c (do_jump): Likewise.
3072 * expr.c (expand_expr_real_2): Check operand type's sign.
3073 * fold-const.c (const_unop): Handle ABSU_EXPR.
3074 (fold_abs_const): Likewise.
3075 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
3076 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
3077 (strip_sign_op_1): Likesise.
3078 * match.pd: Add new pattern to generate ABSU_EXPR.
3079 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
3080 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3081 * tree-eh.c (operation_could_trap_helper_p): Likewise.
3082 * tree-inline.c (estimate_operator_cost): Likewise.
3083 * tree-pretty-print.c (dump_generic_node): Likewise.
3084 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
3085 * tree.def (ABSU_EXPR): New.
3087 2018-06-16 Jakub Jelinek <jakub@redhat.com>
3090 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
3091 documented as preserved for backward compatibility only.
3092 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
3094 PR rtl-optimization/86108
3095 * bb-reorder.c (create_forwarder_block): Renamed to ...
3096 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
3097 jump from new landing pad to the second part.
3098 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
3101 2018-06-15 Jakub Jelinek <jakub@redhat.com>
3104 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
3105 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
3106 Only call store_expr for halves if the mode is the same.
3109 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
3110 Fix up comment formatting.
3112 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
3114 * typed-splay-tree.h (typed_splay_tree::remove): New function.
3115 (typed_splay_tree::closure,
3116 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
3117 (typed_splay_tree::typed_splay_tree,
3118 typed_splay_tree::operator =): Declared private.
3119 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
3120 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
3121 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
3122 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
3123 typed_splay_tree::splay_tree_splay,
3124 typed_splay_tree::splay_tree_foreach_helper,
3125 typed_splay_tree::splay_tree_insert,
3126 typed_splay_tree::splay_tree_remove,
3127 typed_splay_tree::splay_tree_lookup,
3128 typed_splay_tree::splay_tree_predecessor,
3129 typed_splay_tree::splay_tree_successor,
3130 typed_splay_tree::splay_tree_min,
3131 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
3132 (typed_splay_tree::root, typed_splay_tree::comp,
3133 typed_splay_tree::delete_key,
3134 typed_splay_tree::delete_value): New data members.
3135 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
3136 typed_splay_tree::remove.
3138 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
3140 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
3141 -mginv and -mno-ginv to the assembler.
3142 * config/mips/mips.opt (-mcrc): New option.
3144 * doc/invoke.text (-mcrc): Document.
3147 2018-06-15 Nick Clifton <nickc@redhat.com>
3150 * tree.c (escaped_string): New class. Converts an unescaped
3151 string into its escaped equivalent.
3152 (warn_deprecated_use): Use the new class to convert the
3153 deprecation message, if present.
3154 (test_escaped_strings): New self test.
3155 (test_c_tests): Add test_escaped_strings.
3156 * doc/extend.texi (deprecated): Add a note that the
3157 deprecation message is affected by the -fmessage-length
3158 option, and that control characters will be escaped.
3159 (#pragma GCC error): Document this pragma.
3160 (#pragma GCC warning): Likewise.
3161 * doc/invoke.texi (-fmessage-length): Document this option's
3162 effect on the #warning and #error preprocessor directives and
3163 the deprecated attribute.
3165 2018-06-15 Richard Biener <rguenther@suse.de>
3167 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
3168 here, also noting vector size used.
3169 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
3170 size used in MSG_OPTIMIZED_LOCATIONS dump.
3171 (pass_slp_vectorize::execute): Adjust.
3173 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
3176 * config/arc/arc.c (arc_return_address_register): Fix
3179 2018-06-15 Richard Biener <rguenther@suse.de>
3182 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
3183 leave useless conversion stripping to force_gimple_operand_gsi.
3184 (gimplify_build2): Likewise.
3185 (gimplify_build1): Likewise.
3187 2018-06-15 Richard Biener <rguenther@suse.de>
3190 * tree-cfg.c (move_stmt_op): unshare invariant addresses
3191 before adjusting their block.
3193 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
3195 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
3196 multilibs for *-*-rtems*.
3197 * config/riscv/t-rtems: New file.
3199 2018-06-14 Jakub Jelinek <jakub@redhat.com>
3202 * match.pd ((A +- CST1) +- CST2): Punt if last resort
3203 unsigned_type_for returns NULL.
3206 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
3207 subregs of multi-word pseudos unless the float mode has word size.
3209 2018-06-14 Richard Biener <rguenther@suse.de>
3212 * tree-vect-generic.c (build_word_mode_vector_type): Remove
3213 duplicate and harmful type_hash_canon.
3214 * tree.c (type_hash_canon): Assert we didn't find ourselves.
3216 2018-06-14 Richard Biener <rguenther@suse.de>
3219 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
3222 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
3224 * config/rtems.h (STDINT_LONG32): Define.
3226 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
3227 Prachi Godbole <prachi.godbole@imgtec.com>
3229 * config/mips/mips-cpus.def: Define P6600.
3230 * config/mips/mips-tables.opt: Regenerate.
3231 * config/mips/mips.c (mips_ucbranch_type): New enum.
3232 (mips_rtx_cost_data): Add support for P6600.
3233 (mips_issue_rate): Likewise.
3234 (mips_multipass_dfa_lookahead): Likewise.
3235 (mips_avoid_hazard): Likewise.
3236 (mips_reorg_process_insns): Likewise.
3237 (mips_classify_branch_p6600): New function.
3238 * config/mips/mips.h (TUNE_P6600): New define.
3239 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
3240 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
3241 * config/mips/mips.md: Include p6600.md.
3242 (processor): Add p6600.
3243 * config/mips/p6600.md: New file.
3244 * doc/invoke.texi: Add p6600 to supported architectures.
3246 2018-06-13 Martin Sebor <msebor@redhat.com>
3248 PR tree-optimization/86114
3249 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
3251 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
3253 2018-06-13 Richard Biener <rguenther@suse.de>
3255 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
3256 Properly set vector type of the intermediate stmt.
3257 * tree-vect-stmts.c (vectorizable_operation): The destination
3258 var always has vectype_out type.
3260 2018-06-13 Jeff Law <law@redhat.com>
3262 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
3263 integer 0 for argument to print_rtl_with_bb.
3264 (rl78_reorg): Likewise.
3266 2018-06-13 David Malcolm <dmalcolm@redhat.com>
3268 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
3269 from rtx to rtx_insn *.
3270 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
3272 (add_sched_insns_for_speculation): Likewise for local "target",
3273 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
3274 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
3275 from rtx_insn ** to rtx_code_label **.
3276 (reorg_emit_nops): Likewise.
3277 (c6x_reorg): Likewise for local "call_labels".
3278 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
3280 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
3281 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
3282 the loops over LABEL_REFS.
3283 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
3285 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
3286 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
3287 (split_branches): Strengthen local "olabel" from rtx to
3288 rtx_insn *, adding a safe_as_a cast.
3289 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
3291 (add_insn_after): Likewise for first two params.
3292 (add_insn_before): Likewise.
3293 (remove_insn): Likewise for param.
3294 (emit_pattern_before_noloc): Likewise for second and third params.
3295 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
3296 (emit_call_insn_before_noloc): Likewise.
3297 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
3299 (emit_barrier_before): Likewise.
3300 (emit_label_before): Strengthen "label" param from "rtx" to
3301 "rtx_code_label *". Strengthen "before" param from "rtx" to
3303 (emit_insn_after_1): Strengthen "after" param from "rtx" to
3305 (emit_pattern_after_noloc): Likewise.
3306 (emit_insn_after_noloc): Likewise.
3307 (emit_jump_insn_after_noloc): Likewise.
3308 (emit_call_insn_after_noloc): Likewise.
3309 (emit_debug_insn_after_noloc): Likewise.
3310 (emit_barrier_after): Likewise.
3311 (emit_label_after): Likewise for both params.
3312 (emit_pattern_after_setloc): Likewise for "after" param. Convert
3313 "loc" param from "int" to "location_t".
3314 (emit_insn_after_setloc): Likewise.
3315 (emit_jump_insn_after_setloc): Likewise.
3316 (emit_call_insn_after_setloc): Likewise.
3317 (emit_debug_insn_after_setloc): Likewise.
3318 (emit_pattern_before_setloc): Likewise for "before" param. Convert
3319 "loc" param from "int" to "location_t".
3320 (emit_pattern_before): Convert NULL_RTX to NULL.
3321 (emit_insn_before_setloc): Convert "loc" param from "int" to
3323 (emit_jump_insn_before_setloc): Likewise.
3324 (emit_call_insn_before_setloc): Likewise.
3325 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
3326 rtx_insn *. Convert "loc" param from "int" to "location_t".
3327 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
3328 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
3329 Convert 3rd param from "int" to "location_t".
3330 (emit_barrier_before, emit_barrier_after, next_real_insn):
3331 Strengthen param from rtx to rtx_insn *.
3332 (emit_label_before): Strengthen 1st param from "rtx" to
3333 "rtx_code_label *". Strengthen 2nd param from "rtx" to
3335 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
3336 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
3337 Strengthen 2nd param from "rtx" to "rtx_insn *".
3338 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
3339 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
3340 Likewise. Convert 3rd param from "int" to "location_t".
3341 (emit_label_after): Strengthen 1st param from "rtx" to
3343 (next_real_insn, remove_insn): Strengthen param from "rtx" to
3345 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
3346 from "rtx" to "rtx_insn *".
3348 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
3350 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
3351 bodies streamed in with -Q.
3352 * dumpfile.c (dump_files): Add lto-stream-out dump file.
3353 * dumpfile.h (tree_dump_index): Add lto_stream_out.
3354 * gimple-streamer-out.c: Include gimple-pretty-print.h
3355 (output_bb): Dump stmts streamed.
3356 * lto-section-out.c: Include print-tree.h
3357 (lto_begin_section): Dump sections created.
3358 (lto_output_decl_index): Dump decl encoded.
3359 * lto-streamer-out.c: Include print-tree.h
3360 (create_output_block): Dump output block created.
3361 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
3362 (output_function): Dump function output.
3363 (output_constructor): Dump constructor streamed.
3364 (write_global_stream): Output indexes encoded.
3365 (produce_asm_for_decls): Dump streams encoded.
3366 * lto-streamer.c (streamer_dump_file): New global var.
3367 * lto-streamer.h (streamer_dump_file): Declare.
3368 * passes.c (ipa_write_summaries): Initialize streamer dump.
3369 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
3372 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
3375 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
3376 offsets for register save directives. Emit a second batch of save
3377 directives, if need be, when the function accesses prior frames.
3379 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
3381 * config/arc/fpu.md (fmasf4): Force operand to register.
3382 (fnmasf4): Likewise.
3384 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
3386 * config/arc/arc-protos.h (arc_pad_return): Remove.
3387 * config/arc/arc.c (machine_function): Remove force_short_suffix
3389 (arc_print_operand): Adjust printing of '&'.
3390 (arc_verify_short): Remove conditional printing of short suffix.
3391 (arc_final_prescan_insn): Remove reference to size_reason.
3392 (pad_return): New function.
3393 (arc_reorg): Call pad_return.
3394 (arc_pad_return): Remove.
3395 (arc_init_machine_status): Remove reference to force_short_suffix.
3396 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
3397 (attr length): When attribute iscompact is true force to 2
3398 regardless; in the case of maybe check if we want to force the
3399 instruction to have 4 bytes length.
3400 (nopv): Change it to generate 4 byte long nop as well.
3401 (blockage): New pattern.
3402 (simple_return): Remove call to arc_pad_return.
3403 (p_return_i): Likewise.
3405 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
3407 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
3409 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
3411 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
3414 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
3416 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
3417 for ARC700 and ARCv2.
3419 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
3422 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
3423 operands[2] instead of operands[1].
3426 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
3428 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
3429 case, check whether the outer register overlaps an unallocatable
3430 register, not just whether it fits the required class.
3432 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
3434 * poly-int.h (can_div_trunc_p): Add new overload in which all values
3436 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
3437 (memrefs_conflict_p): Likewise.
3438 (init_alias_analysis): Likewise.
3439 * cfgexpand.c (expand_debug_expr): Likewise.
3440 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
3441 * cse.c (fold_rtx): Likewise.
3442 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
3443 * expr.c (emit_block_move_hints): Likewise.
3444 (clear_storage_hints, push_block, emit_push_insn): Likewise.
3445 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
3446 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
3447 (emit_group_store): Likewise.
3448 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
3449 to read the PRE/POST_MODIFY increment.
3450 * calls.c (store_one_arg): Use strip_offset.
3451 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
3453 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
3455 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
3456 (simplify_binary_operation_1): Extend CONST_INT handling to
3458 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
3459 than a HOST_WIDE_INT.
3460 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
3462 (adjust_mems, add_stores): Update accodingly.
3463 (vt_canonicalize_addr): Track polynomial offsets.
3464 (emit_note_insn_var_location): Likewise.
3465 (vt_add_function_parameter): Likewise.
3466 (vt_initialize): Likewise.
3468 2018-06-12 Jeff Law <law@redhat.com>
3470 * config.gcc (alpha*-*-freebsd*): Remove.
3471 * config/alpha/freebsd.h: Remove.
3473 2018-06-12 David Malcolm <dmalcolm@redhat.com>
3476 * spellcheck-tree.c (levenshtein_distance): Rename to...
3477 (get_edit_distance): ...this, and update for underlying renaming.
3478 * spellcheck-tree.h (levenshtein_distance): Rename to...
3479 (get_edit_distance): ...this.
3480 * spellcheck.c (levenshtein_distance): Rename to...
3481 (get_edit_distance): ...this. Convert from Levenshtein distance
3482 to Damerau-Levenshtein distance by supporting transpositions of
3483 adjacent characters. Rename "v1" to "v_next" and "v0" to
3485 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
3486 (selftest::test_edit_distance_unit_test_oneway): ...this, and
3487 update for underlying renaming.
3488 (selftest::levenshtein_distance_unit_test): Rename to...
3489 (selftest::test_get_edit_distance_unit): ...this, and update for
3490 underlying renaming.
3491 (selftest::test_find_closest_string): Add example from PR 69968
3492 where transposition helps
3493 (selftest::test_metric_conditions): Update for renaming.
3494 (selftest::test_metric_conditions): Likewise.
3495 (selftest::spellcheck_c_tests): Likewise.
3496 * spellcheck.h (levenshtein_distance): Rename both overloads to...
3497 (get_edit_distance): ...this.
3498 (best_match::consider): Update for renaming.
3500 2018-06-12 Martin Sebor <msebor@redhat.com>
3502 PR tree-optimization/85259
3503 * builtins.c (compute_objsize): Handle constant offsets.
3504 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
3505 true iff a warning has been issued.
3506 * gimple.h (gimple_nonartificial_location): New function.
3507 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
3508 gimple_nonartificial_location and handle -Wno-system-headers.
3509 (handle_builtin_stxncpy): Same.
3511 2018-06-12 Martin Sebor <msebor@redhat.com>
3514 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
3516 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
3518 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
3519 BUILTIN_VEC_XST entries for pointer to double and long long.
3521 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
3524 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
3526 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
3529 2018-06-12 Martin Liska <mliska@suse.cz>
3531 * doc/options.texi: Document IntegerRange.
3533 2018-06-12 Martin Liska <mliska@suse.cz>
3535 * config/i386/i386.opt: Make MPX-related options as Deprecated.
3536 * opt-functions.awk: Handle Deprecated flag.
3537 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
3539 (read_cmdline_option): Report warning for a deprecated option.
3540 * opts.h (struct cl_option): Add new field cl_deprecated.
3541 (CL_ERR_DEPRECATED): New.
3543 2018-06-12 Martin Liska <mliska@suse.cz>
3545 * doc/options.texi: Document Deprecated option flag.
3547 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
3549 * config/arc/arc-arch.h (arc_extras): New enum.
3550 (arc_cpu_t):Add field extra.
3551 (arc_cpu_types): Consider the extras.
3552 * config/arc/arc-cpus.def: Add extras info.
3553 * config/arc/arc-opts.h (processor_type): Consider extra field.
3554 * config/arc/arc.c (arc_override_options): Handle extra field.
3556 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
3558 * config/arc/arc-arch.h: Update ARC_OPTX macro.
3559 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
3561 * config/arc/arc.c (arc_init): Update pic warning.
3562 (irq_range): Update irq range parsing warnings.
3563 (arc_override_options): Update various warning messages.
3564 (arc_handle_aux_attribute): Likewise.
3566 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
3568 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
3570 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3572 * doc/sourcebuild.texi: Document usage of line number 0 in verify
3573 compiler messages directives.
3575 2018-06-12 Matthew Fortune <mfortune@gmail.com>
3577 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
3578 * config/mips/mips-tables.opt: Regenerate.
3579 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
3581 * doc/invoke.texi: Document -march=i6500.
3583 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
3585 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
3586 (i6400_gpmul): Add cpu_unit.
3587 (i6400_gpdiv): Likewise.
3588 (i6400_msa_add_d): Update reservations.
3589 (i6400_msa_int_add) Likewise.
3590 (i6400_msa_short_logic3) Likewise.
3591 (i6400_msa_short_logic2) Likewise.
3592 (i6400_msa_short_logic) Likewise.
3593 (i6400_msa_move) Likewise.
3594 (i6400_msa_cmp) Likewise.
3595 (i6400_msa_short_float2) Likewise.
3596 (i6400_msa_div_d) Likewise.
3597 (i6400_msa_long_logic1) Likewise.
3598 (i6400_msa_long_logic2) Likewise.
3599 (i6400_msa_mult) Likewise.
3600 (i6400_msa_long_float2) Likewise.
3601 (i6400_msa_long_float4) Likewise.
3602 (i6400_msa_long_float5) Likewise.
3603 (i6400_msa_long_float8) Likewise.
3604 (i6400_fpu_fadd): Include frint type.
3605 (i6400_fpu_store): New define_insn_reservation.
3606 (i6400_fpu_load): Likewise.
3607 (i6400_fpu_move): Likewise.
3608 (i6400_fpu_fcmp): Likewise.
3609 (i6400_fpu_fmadd): Likewise.
3610 (i6400_int_mult): Include imul3nc type and update reservation.
3611 (i6400_int_div): Include idiv3 type and update reservation.
3612 (i6400_int_load): Update to check type not move_type.
3613 (i6400_int_store): Likewise.
3614 (i6400_int_prefetch): Set zero latency.
3616 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
3618 * gcc.c: Document new %@{...} sequence.
3619 (LINK_COMMAND_SPEC): Use it for the -L switches.
3620 (cpp_unique_options): Use it for the -I switches.
3621 (at_file_argbuf): New global variable.
3622 (in_at_file): Likewise.
3623 (alloc_args): Create at_file_argbuf.
3624 (clear_args): Truncate at_file_argbuf.
3625 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
3626 (open_at_file): New function.
3627 (close_at_file): Likewise.
3628 (create_at_file): Delete.
3629 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
3632 (validate_switches_from_spec): Deal with %@{...} sequence.
3633 (validate_switches): Likewise.
3634 (driver::finalize): Call clear_args.
3636 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
3638 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
3640 2018-06-11 Martin Sebor <msebor@redhat.com>
3642 * doc/invoke.texi (-Wall): List -Wc++17-compat.
3643 (Wno-class-memaccess): Add @opindex.
3644 (Wno-templates, Wno-multiple-inheritance): Same.
3645 (Wno-virtual-inheritance, Wno-namespaces): Same.
3646 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
3647 (Wno-format-overflow, Wno-format-truncation): Same.
3648 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
3649 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
3650 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
3651 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
3652 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
3653 (Wno-misspelled-isr): Same.
3655 2018-06-11 Martin Sebor <msebor@redhat.com>
3657 * PR tree-optimization/86083
3658 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
3660 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
3662 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
3664 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
3667 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
3668 on the correct operand.
3669 (*movdi_internal64): Ditto.
3671 2018-06-11 Martin Liska <mliska@suse.cz>
3673 PR tree-optimization/86089
3674 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
3676 2018-06-11 Julia Koval <julia.koval@intel.com>
3678 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
3679 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
3680 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
3682 2018-06-11 Olivier Hainque <hainque@adacore.com>
3684 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
3685 for Ada with strict dwarf2.
3687 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
3690 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
3693 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
3695 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
3697 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
3699 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
3702 2018-06-08 Martin Liska <mliska@suse.cz>
3704 * tree-cfg.h (debug_function): Fix argument type to match
3707 2018-06-08 Martin Liska <mliska@suse.cz>
3709 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
3710 Remove usage of MPX-related (and removed) fields.
3711 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
3713 2018-06-08 David Malcolm <dmalcolm@redhat.com>
3715 * cfg.c (debug): Use TDF_NONE rather than 0.
3716 * cfghooks.c (debug): Likewise.
3717 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
3718 (struct dump_option_value_info): Convert to...
3719 (struct kv_pair): ...this template type.
3720 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
3722 (optinfo_verbosity_options): Likewise.
3723 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
3725 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
3726 than int for "optgroup_flags" param.
3727 (dump_generic_expr_loc): Use dump_flags_t rather than int for
3729 (dump_dec): Likewise.
3730 (dump_finish): Use TDF_NONE rather than 0.
3731 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
3732 rather than int for "optgroup_flags" param. Use TDF_NONE rather
3733 than 0. Update for change to option_ptr.
3734 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
3735 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
3736 0. Update for changes to optinfo_verbosity_options and
3738 (opt_info_switch_p): Convert optgroup_flags from int to
3740 (dump_basic_block): Use dump_flags_t rather than int
3741 for "dump_kind" param.
3742 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
3743 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
3744 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
3745 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
3746 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
3747 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
3748 TDF_NONE): Convert from macros to...
3749 (enum dump_flag): ...this new enum.
3750 (dump_flags_t): Update to use enum.
3751 (operator|, operator&, operator~, operator|=, operator&=):
3752 Implement for dump_flags_t.
3753 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
3754 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
3755 Convert from macros to...
3756 (enum optgroup_flag): ...this new enum.
3757 (optgroup_flags_t): New typedef.
3758 (operator|, operator|=): Implement for optgroup_flags_t.
3759 (struct dump_file_info): Convert field "alt_flags" to
3760 dump_flags_t. Convert field "optgroup_flags" to
3762 (dump_basic_block): Use dump_flags_t rather than int for param.
3763 (dump_generic_expr_loc): Likewise.
3764 (dump_dec): Likewise.
3765 (dump_register): Convert param "optgroup_flags" to
3767 (opt_info_enable_passes): Likewise.
3768 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
3770 * gimple-pretty-print.c (debug): Likewise.
3771 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
3772 (merged_store_group::apply_stores): Likewise.
3773 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
3774 * gimple.c (verify_gimple_pp): Likewise.
3775 * graphite-poly.c (print_pbb_body): Likewise.
3776 * passes.c (pass_manager::register_one_dump_file): Convert
3777 local "optgroup_flags" to optgroup_flags_t.
3778 * print-tree.c (print_node): Use TDF_NONE rather than 0.
3780 (debug_body): Likewise.
3781 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
3782 to optgroup_flags_t.
3783 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
3785 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
3786 (convert_mult_to_fma): Likewise.
3787 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
3788 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
3789 * tree-vect-data-refs.c (dump_lower_bound): Convert param
3790 "dump_kind" to dump_flags_t.
3792 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
3794 * config/rs6000/rs6000.c (min, max): Delete.
3796 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
3798 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
3799 -mabi=spe and -mabi=no-spe.
3801 2018-06-08 Martin Liska <mliska@suse.cz>
3803 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
3804 where we expect an existing summary.
3806 2018-06-08 Martin Liska <mliska@suse.cz>
3808 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
3809 * ipa-inline.h (estimate_edge_growth): Likewise.
3811 2018-06-08 Martin Liska <mliska@suse.cz>
3813 * cgraph.c (function_version_hasher::hash): Use
3814 cgraph_node::get_uid ().
3815 (function_version_hasher::equal):
3816 * cgraph.h (cgraph_node::get_uid): New method.
3817 * ipa-inline.c (update_caller_keys): Use
3818 cgraph_node::get_uid ().
3819 (update_callee_keys): Likewise.
3820 * ipa-utils.c (searchc): Likewise.
3821 (ipa_reduced_postorder): Likewise.
3822 * lto-cgraph.c (input_node): Likewise.
3823 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
3824 * symbol-summary.h (symtab_insertion): Likewise.
3825 (symtab_removal): Likewise.
3826 (symtab_duplication): Likewise.
3827 * tree-pretty-print.c (dump_function_header): Likewise.
3828 * tree-sra.c (convert_callers_for_node): Likewise.
3830 2018-06-08 Martin Liska <mliska@suse.cz>
3832 * cgraph.c (symbol_table::create_edge): Always assign a new
3834 (symbol_table::free_edge): Do not recycle numbers.
3835 * cgraph.h (cgraph_edge::get): New method.
3836 * symbol-summary.h (symtab_removal): Use it.
3837 (symtab_duplication): Likewise.
3838 (call_summary::hashable_uid): Remove.
3840 2018-06-08 Martin Liska <mliska@suse.cz>
3842 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
3843 (initialize_growth_caches): Remove.
3844 (free_growth_caches): Likewise.
3845 (do_estimate_edge_time): Use edge_growth_cache.
3846 (do_estimate_edge_size): Likewise.
3847 (do_estimate_edge_hints): Likewise.
3848 * ipa-inline.c (reset_edge_caches): Likewise.
3849 (recursive_inlining): Likewise.
3850 (inline_small_functions): Likewise.
3851 * ipa-inline.h (initialize_growth_caches): Remove.
3852 (estimate_edge_size): Likewise.
3853 (estimate_edge_time): Likewise.
3854 (estimate_edge_hints): Likewise.
3855 (reset_edge_growth_cache): Likewise.
3856 * symbol-summary.h (call_summary::remove): New method.
3858 2018-06-08 Martin Liska <mliska@suse.cz>
3860 * ipa-cp.c (class edge_clone_summary): New summary.
3861 (grow_edge_clone_vectors): Remove.
3862 (ipcp_edge_duplication_hook): Remove.
3863 (class edge_clone_summary_t): New call_summary class.
3864 (ipcp_edge_removal_hook): Remove.
3865 (edge_clone_summary_t::duplicate): New function.
3866 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
3867 (create_specialized_node): Likewise.
3868 (ipcp_driver): Initialize edge_clone_summaries and do not
3871 2018-06-08 Martin Liska <mliska@suse.cz>
3873 * symbol-summary.h (get): New function.
3874 (call_summary::m_initialize_when_cloning): New class member.
3876 2018-06-08 Martin Liska <mliska@suse.cz>
3878 * cgraph.c (cgraph_node::remove): Do not recycle uid.
3879 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
3880 (symbol_table::allocate_cgraph_symbol): Do not set uid.
3881 * passes.c (uid_hash_t): Record removed_nodes by their uids.
3882 (remove_cgraph_node_from_order): Use the removed_nodes set.
3883 (do_per_function_toporder): Likwise.
3884 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
3885 instead of summary_uid.
3886 (symtab_removal): Likewise.
3887 (symtab_duplication): Likewise.
3889 2018-06-08 Martin Liska <mliska@suse.cz>
3891 * ipa-cp.c (ipcp_store_bits_results): Use
3892 ipcp_transformation_sum.
3893 (ipcp_store_vr_results): Likewise.
3894 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
3896 (ipcp_transformation_initialize): ... this.
3897 (ipa_set_node_agg_value_chain):
3898 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
3899 (write_ipcp_transformation_info): Likewise.
3900 (read_ipcp_transformation_info): Likewise.
3901 (ipcp_update_bits): Likewise.
3902 (ipcp_update_vr): Likewise.
3903 (ipcp_transform_function): Likewise.
3904 * ipa-prop.h: Rename ipcp_transformation_summary to
3905 ipcp_transformation.
3906 (class ipcp_transformation_t): New function summary.
3907 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
3908 (ipa_get_agg_replacements_for_node): Likewise.
3910 2018-06-08 Martin Liska <mliska@suse.cz>
3912 * ipa-pure-const.c (struct funct_state_d): Do it class instead
3914 (class funct_state_summary_t): New function_summary class.
3915 (has_function_state): Remove.
3916 (get_function_state): Likewise.
3917 (set_function_state): Likewise.
3918 (add_new_function): Likewise.
3919 (funct_state_summary_t::insert): New function.
3920 (duplicate_node_data): Remove.
3921 (remove_node_data): Remove.
3922 (funct_state_summary_t::duplicate): New function.
3923 (register_hooks): Create new funct_state_summaries.
3924 (pure_const_generate_summary): Use it.
3925 (pure_const_write_summary): Likewise.
3926 (pure_const_read_summary): Likewise.
3927 (propagate_pure_const): Likewise.
3928 (propagate_nothrow): Likewise.
3929 (dump_malloc_lattice): Likewise.
3930 (propagate_malloc): Likewise.
3931 (execute): Do not register hooks, just remove summary
3933 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
3936 2018-06-08 Martin Liska <mliska@suse.cz>
3938 * ipa-reference.c (remove_node_data): Remove.
3939 (duplicate_node_data): Likewise.
3940 (class ipa_ref_var_info_summary_t): New class.
3941 (class ipa_ref_opt_summary_t): Likewise.
3942 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
3943 (get_reference_optimization_summary): Use
3944 ipa_ref_opt_sum_summaries.
3945 (set_reference_vars_info): Remove.
3946 (set_reference_optimization_summary): Likewise.
3947 (ipa_init): Create summaries.
3948 (init_function_info): Use function summary.
3949 (ipa_ref_opt_summary_t::duplicate): New function.
3950 (ipa_ref_opt_summary_t::remove): New function.
3951 (get_read_write_all_from_node): Fix GNU coding style.
3952 (propagate): Use function summary.
3953 (write_node_summary_p): Fix GNU coding style.
3954 (stream_out_bitmap): Likewise.
3955 (ipa_reference_read_optimization_summary): Use function summary.
3956 (ipa_reference_c_finalize): Do not release hooks.
3958 2018-06-08 Martin Liska <mliska@suse.cz>
3960 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
3961 (analyze_function_body): Extract multiple calls of get_create.
3962 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
3963 * ipa-inline.c (recursive_inlining): Use ::get method.
3964 * ipa-inline.h (estimate_edge_growth): Likewise.
3966 2018-06-08 Martin Liska <mliska@suse.cz>
3968 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
3970 (hsa_function_summary::hsa_function_summary): Use the new enum
3972 (hsa_gpu_implementation_p): Use hsa_summaries::get.
3973 * hsa-gen.c (hsa_get_host_function): Likewise.
3974 (get_brig_function_name): Likewise.
3975 * ipa-hsa.c (process_hsa_functions): Likewise.
3976 (ipa_hsa_write_summary): Likewise.
3977 * symbol-summary.h (symtab_duplication): Use ::get function/
3978 (get): New function.
3980 2018-06-08 Martin Liska <mliska@suse.cz>
3982 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
3984 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
3985 (hsa_register_kernel): Likewise.
3986 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
3987 * hsa-gen.c (hsa_get_host_function): Likewise.
3988 (get_brig_function_name): Likewise.
3989 (generate_hsa): Likewise.
3990 (pass_gen_hsail::execute): Likewise.
3991 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
3992 (devirtualization_time_bonus): Likewise.
3993 (ipcp_propagate_stage): Likewise.
3994 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
3995 (edge_set_predicate): Likewise.
3996 (evaluate_conditions_for_known_args): Likewise.
3997 (evaluate_properties_for_edge): Likewise.
3998 (ipa_fn_summary::reset): Likewise.
3999 (ipa_fn_summary_t::duplicate): Likewise.
4000 (dump_ipa_call_summary): Likewise.
4001 (ipa_dump_fn_summary): Likewise.
4002 (analyze_function_body): Likewise.
4003 (compute_fn_summary): Likewise.
4004 (estimate_edge_devirt_benefit): Likewise.
4005 (estimate_edge_size_and_time): Likewise.
4006 (estimate_calls_size_and_time): Likewise.
4007 (estimate_node_size_and_time): Likewise.
4008 (inline_update_callee_summaries): Likewise.
4009 (remap_edge_change_prob): Likewise.
4010 (remap_edge_summaries): Likewise.
4011 (ipa_merge_fn_summary_after_inlining): Likewise.
4012 (ipa_update_overall_fn_summary): Likewise.
4013 (read_ipa_call_summary): Likewise.
4014 (inline_read_section): Likewise.
4015 (write_ipa_call_summary): Likewise.
4016 (ipa_fn_summary_write): Likewise.
4017 (ipa_free_fn_summary): Likewise.
4018 * ipa-hsa.c (process_hsa_functions): Likewise.
4019 (ipa_hsa_write_summary): Likewise.
4020 (ipa_hsa_read_section): Likewise.
4021 * ipa-icf.c (sem_function::merge): Likewise.
4022 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
4023 (do_estimate_edge_time): Likewise.
4024 (estimate_size_after_inlining): Likewise.
4025 (estimate_growth): Likewise.
4026 (growth_likely_positive): Likewise.
4027 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
4028 (inline_call): Likewise.
4029 * ipa-inline.c (caller_growth_limits): Likewise.
4030 (can_inline_edge_p): Likewise.
4031 (can_inline_edge_by_limits_p): Likewise.
4032 (compute_uninlined_call_time): Likewise.
4033 (compute_inlined_call_time): Likewise.
4034 (want_inline_small_function_p): Likewise.
4035 (edge_badness): Likewise.
4036 (update_caller_keys): Likewise.
4037 (update_callee_keys): Likewise.
4038 (recursive_inlining): Likewise.
4039 (inline_small_functions): Likewise.
4040 (inline_to_all_callers_1): Likewise.
4041 (dump_overall_stats): Likewise.
4042 (early_inline_small_functions): Likewise.
4043 (early_inliner): Likewise.
4044 * ipa-inline.h (estimate_edge_growth): Likewise.
4045 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
4046 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
4047 * ipa-prop.h (IPA_NODE_REF): Likewise.
4048 (IPA_EDGE_REF): Likewise.
4049 * ipa-pure-const.c (malloc_candidate_p): Likewise.
4050 (propagate_malloc): Likewise.
4051 * ipa-split.c (execute_split_functions): Likewise.
4052 * symbol-summary.h: Rename get to get_create.
4054 (get_create): Likewise.
4055 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
4057 2018-06-08 Martin Liska <mliska@suse.cz>
4059 * symbol-summary.h (release): Move definition out of class
4061 (symtab_removal): Likewise.
4062 (symtab_duplication): Likewise.
4064 2018-06-08 Martin Liska <mliska@suse.cz>
4066 * symbol-summary.h (function_summary): Move constructor
4067 implementation out of class declaration.
4068 (release): Likewise.
4069 (symtab_insertion): Likewise.
4070 (symtab_removal): Likewise.
4071 (symtab_duplication): Likewise.
4074 2018-06-08 Martin Liska <mliska@suse.cz>
4076 * Makefile.in: Remove support for MPX (macros, related functions,
4077 fields in cgraph_node, ...).
4078 * builtin-types.def (BT_BND): Likewise.
4079 (BT_FN_BND_CONST_PTR): Likewise.
4080 (BT_FN_CONST_PTR_BND): Likewise.
4081 (BT_FN_VOID_PTR_BND): Likewise.
4082 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
4083 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
4084 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
4085 (expand_builtin_mempcpy_with_bounds): Likewise.
4086 (expand_builtin_memset_with_bounds): Likewise.
4087 (expand_builtin_memset_args): Likewise.
4088 (std_expand_builtin_va_start): Likewise.
4089 (expand_builtin): Likewise.
4090 (expand_builtin_with_bounds): Likewise.
4091 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
4092 (DEF_LIB_BUILTIN_CHKP): Likewise.
4093 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
4094 (DEF_CHKP_BUILTIN): Likewise.
4095 (BUILT_IN_MEMCPY): Likewise.
4096 (BUILT_IN_MEMMOVE): Likewise.
4097 (BUILT_IN_MEMPCPY): Likewise.
4098 (BUILT_IN_MEMSET): Likewise.
4099 (BUILT_IN_STPCPY): Likewise.
4100 (BUILT_IN_STRCAT): Likewise.
4101 (BUILT_IN_STRCHR): Likewise.
4102 (BUILT_IN_STRCPY): Likewise.
4103 (BUILT_IN_STRLEN): Likewise.
4104 (BUILT_IN_MEMCPY_CHK): Likewise.
4105 (BUILT_IN_MEMMOVE_CHK): Likewise.
4106 (BUILT_IN_MEMPCPY_CHK): Likewise.
4107 (BUILT_IN_MEMSET_CHK): Likewise.
4108 (BUILT_IN_STPCPY_CHK): Likewise.
4109 (BUILT_IN_STRCAT_CHK): Likewise.
4110 (BUILT_IN_STRCPY_CHK): Likewise.
4111 * calls.c (store_bounds): Likewise.
4112 (emit_call_1): Likewise.
4113 (special_function_p): Likewise.
4114 (maybe_warn_nonstring_arg): Likewise.
4115 (initialize_argument_information): Likewise.
4116 (finalize_must_preallocate): Likewise.
4117 (compute_argument_addresses): Likewise.
4118 (expand_call): Likewise.
4119 * cfgexpand.c (expand_call_stmt): Likewise.
4120 (expand_return): Likewise.
4121 (expand_gimple_stmt_1): Likewise.
4122 (pass_expand::execute): Likewise.
4123 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
4124 (cgraph_node::remove): Likewise.
4125 (cgraph_node::dump): Likewise.
4126 (cgraph_node::verify_node): Likewise.
4127 * cgraph.h (chkp_function_instrumented_p): Likewise.
4128 (symtab_node::get_alias_target): Likewise.
4129 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
4130 (cgraph_local_p): Likewise.
4131 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
4132 (cgraph_edge::rebuild_references): Likewise.
4133 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
4134 (walk_polymorphic_call_targets): Likewise.
4135 (cgraph_node::expand_thunk): Likewise.
4136 (symbol_table::output_weakrefs): Likewise.
4137 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
4138 (ix86_handle_option): Likewise.
4139 * config/i386/constraints.md: Likewise.
4140 * config/i386/i386-builtin-types.def (BND): Likewise.
4144 * config/i386/i386-builtin.def (BDESC_END): Likewise.
4145 (BDESC_FIRST): Likewise.
4147 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
4148 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
4149 * config/i386/i386.c (enum reg_class): Likewise.
4150 (ix86_target_string): Likewise.
4151 (ix86_option_override_internal): Likewise.
4152 (ix86_conditional_register_usage): Likewise.
4153 (ix86_valid_target_attribute_inner_p): Likewise.
4154 (ix86_set_indirect_branch_type): Likewise.
4155 (ix86_set_current_function): Likewise.
4156 (ix86_function_arg_regno_p): Likewise.
4157 (init_cumulative_args): Likewise.
4158 (ix86_function_arg_advance): Likewise.
4159 (ix86_function_arg): Likewise.
4160 (ix86_pass_by_reference): Likewise.
4161 (ix86_function_value_regno_p): Likewise.
4162 (ix86_function_value_1): Likewise.
4163 (ix86_function_value_bounds): Likewise.
4164 (ix86_return_in_memory): Likewise.
4165 (ix86_setup_incoming_vararg_bounds): Likewise.
4166 (ix86_va_start): Likewise.
4167 (indirect_thunk_need_prefix): Likewise.
4168 (print_reg): Likewise.
4169 (ix86_print_operand): Likewise.
4170 (ix86_expand_call): Likewise.
4171 (ix86_output_function_return): Likewise.
4172 (reg_encoded_number): Likewise.
4173 (BDESC_VERIFYS): Likewise.
4174 (ix86_init_mpx_builtins): Likewise.
4175 (ix86_init_builtins): Likewise.
4176 (ix86_emit_cmove): Likewise.
4177 (ix86_emit_move_max): Likewise.
4178 (ix86_expand_builtin): Likewise.
4179 (ix86_builtin_mpx_function): Likewise.
4180 (ix86_get_arg_address_for_bt): Likewise.
4181 (ix86_load_bounds): Likewise.
4182 (ix86_store_bounds): Likewise.
4183 (ix86_load_returned_bounds): Likewise.
4184 (ix86_store_returned_bounds): Likewise.
4185 (ix86_class_likely_spilled_p): Likewise.
4186 (ix86_hard_regno_mode_ok): Likewise.
4187 (x86_order_regs_for_local_alloc): Likewise.
4188 (ix86_mitigate_rop): Likewise.
4189 (ix86_bnd_prefixed_insn_p): Likewise.
4190 (ix86_mpx_bound_mode): Likewise.
4191 (ix86_make_bounds_constant): Likewise.
4192 (ix86_initialize_bounds): Likewise.
4193 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
4194 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
4195 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
4196 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
4197 (TARGET_CHKP_BOUND_MODE): Likewise.
4198 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
4199 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
4200 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
4201 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
4202 * config/i386/i386.h (TARGET_MPX): Likewise.
4203 (TARGET_MPX_P): Likewise.
4204 (VALID_BND_REG_MODE): Likewise.
4205 (FIRST_BND_REG): Likewise.
4206 (LAST_BND_REG): Likewise.
4207 (enum reg_class): Likewise.
4208 (BND_REG_P): Likewise.
4209 (BND_REGNO_P): Likewise.
4210 (BNDmode): Likewise.
4211 (ADJUST_INSN_LENGTH): Likewise.
4212 * config/i386/i386.md: Likewise.
4213 * config/i386/i386.opt: Likewise.
4214 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
4215 (defined): Likewise.
4216 (LINK_MPX): Likewise.
4217 (MPX_SPEC): Likewise.
4218 (LIBMPX_SPEC): Likewise.
4219 (LIBMPXWRAPPERS_SPEC): Likewise.
4220 (CHKP_SPEC): Likewise.
4221 * config/i386/predicates.md: Likewise.
4222 * dbxout.c (dbxout_type): Likewise.
4223 * doc/extend.texi: Likewise.
4224 * doc/invoke.texi: Likewise.
4225 * doc/md.texi: Likewise.
4226 * doc/tm.texi: Likewise.
4227 * doc/tm.texi.in: Likewise.
4228 * dwarf2out.c (is_base_type): Likewise.
4229 (gen_formal_types_die): Likewise.
4230 (gen_subprogram_die): Likewise.
4231 (gen_type_die_with_usage): Likewise.
4232 (gen_decl_die): Likewise.
4233 (dwarf2out_late_global_decl): Likewise.
4234 * expr.c (expand_assignment): Likewise.
4235 (emit_storent_insn): Likewise.
4236 (store_expr_with_bounds): Likewise.
4237 (store_expr): Likewise.
4238 (expand_expr_real_1): Likewise.
4239 * expr.h (store_expr_with_bounds): Likewise.
4240 * function.c (use_register_for_decl): Likewise.
4241 (struct bounds_parm_data): Likewise.
4242 (assign_parms_augmented_arg_list): Likewise.
4243 (assign_parm_find_entry_rtl): Likewise.
4244 (assign_parm_is_stack_parm): Likewise.
4245 (assign_parm_load_bounds): Likewise.
4246 (assign_bounds): Likewise.
4247 (assign_parms): Likewise.
4248 (expand_function_start): Likewise.
4249 * gcc.c (CHKP_SPEC): Likewise.
4250 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
4251 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
4252 (wrestrict_dom_walker::check_call): Likewise.
4253 * gimple.c (gimple_build_call_from_tree): Likewise.
4254 * gimple.h (enum gf_mask): Likewise.
4255 (gimple_call_with_bounds_p): Likewise.
4256 (gimple_call_set_with_bounds): Likewise.
4257 * gimplify.c (gimplify_init_constructor): Likewise.
4258 * ipa-cp.c (initialize_node_lattices): Likewise.
4259 (propagate_constants_across_call): Likewise.
4260 (find_more_scalar_values_for_callers_subset): Likewise.
4261 * ipa-hsa.c (process_hsa_functions): Likewise.
4262 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
4263 * ipa-icf.c (sem_function::merge): Likewise.
4264 * ipa-inline.c (early_inliner): Likewise.
4265 * ipa-pure-const.c (warn_function_noreturn): Likewise.
4266 (warn_function_cold): Likewise.
4267 (propagate_pure_const): Likewise.
4268 * ipa-ref.h (enum GTY): Likewise.
4269 * ipa-split.c (find_retbnd): Likewise.
4270 (consider_split): Likewise.
4271 (split_function): Likewise.
4272 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
4273 * ipa.c (walk_polymorphic_call_targets): Likewise.
4274 (symbol_table::remove_unreachable_nodes): Likewise.
4275 (process_references): Likewise.
4276 (cgraph_build_static_cdtor_1): Likewise.
4277 * lto-cgraph.c (lto_output_node): Likewise.
4278 (output_refs): Likewise.
4279 (compute_ltrans_boundary): Likewise.
4280 (input_overwrite_node): Likewise.
4281 (input_node): Likewise.
4282 (input_cgraph_1): Likewise.
4283 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
4284 * passes.c (pass_manager::execute_early_local_passes): Likewise.
4285 (class pass_chkp_instrumentation_passes): Likewise.
4286 (make_pass_chkp_instrumentation_passes): Likewise.
4287 * passes.def: Likewise.
4288 * rtl.h (struct GTY): Likewise.
4289 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
4290 * stor-layout.c (layout_type): Likewise.
4291 * symtab.c: Likewise.
4292 * target.def: Likewise.
4293 * targhooks.c (default_chkp_bound_type): Likewise.
4294 (default_chkp_bound_mode): Likewise.
4295 (default_builtin_chkp_function): Likewise.
4296 (default_chkp_function_value_bounds): Likewise.
4297 (default_chkp_make_bounds_constant): Likewise.
4298 (default_chkp_initialize_bounds): Likewise.
4299 * targhooks.h (default_chkp_bound_type): Likewise.
4300 (default_chkp_bound_mode): Likewise.
4301 (default_builtin_chkp_function): Likewise.
4302 (default_chkp_function_value_bounds): Likewise.
4303 (default_chkp_make_bounds_constant): Likewise.
4304 (default_chkp_initialize_bounds): Likewise.
4305 * toplev.c (compile_file): Likewise.
4306 (process_options): Likewise.
4307 * tree-core.h (DEF_BUILTIN): Likewise.
4308 (DEF_BUILTIN_CHKP): Likewise.
4309 * tree-inline.c (declare_return_variable): Likewise.
4310 (remap_gimple_stmt): Likewise.
4311 (copy_bb): Likewise.
4312 (initialize_inlined_parameters): Likewise.
4313 (expand_call_inline): Likewise.
4314 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
4315 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
4316 (make_pass_ipa_chkp_produce_thunks): Likewise.
4317 (make_pass_chkp): Likewise.
4318 (make_pass_chkp_opt): Likewise.
4319 (make_pass_chkp_instrumentation_passes): Likewise.
4320 * tree-pretty-print.c (dump_generic_node): Likewise.
4321 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
4322 * tree-ssa-dce.c (propagate_necessity): Likewise.
4323 (eliminate_unnecessary_stmts): Likewise.
4324 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
4325 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
4326 * tree-ssa-sccvn.h: Likewise.
4327 * tree-ssa-strlen.c (get_string_length): Likewise.
4328 (valid_builtin_call): Likewise.
4329 (adjust_last_stmt): Likewise.
4330 (handle_builtin_strchr): Likewise.
4331 (handle_builtin_strcpy): Likewise.
4332 (handle_builtin_stxncpy): Likewise.
4333 (handle_builtin_memcpy): Likewise.
4334 (handle_builtin_strcat): Likewise.
4335 (strlen_check_and_optimize_stmt): Likewise.
4336 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
4337 * tree-streamer-in.c: Likewise.
4338 * tree-streamer.c (record_common_node): Likewise.
4339 * tree.c (tree_code_size): Likewise.
4340 (wide_int_to_tree_1): Likewise.
4341 (type_contains_placeholder_1): Likewise.
4342 (build_common_tree_nodes): Likewise.
4343 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
4344 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
4345 (POINTER_BOUNDS_P): Likewise.
4346 (BOUNDED_TYPE_P): Likewise.
4347 (BOUNDED_P): Likewise.
4348 (CALL_WITH_BOUNDS_P): Likewise.
4349 (pointer_bounds_type_node): Likewise.
4350 * value-prof.c (gimple_ic): Likewise.
4351 * var-tracking.c (vt_add_function_parameters): Likewise.
4352 * varasm.c (make_decl_rtl): Likewise.
4353 (assemble_start_function): Likewise.
4354 (output_constant): Likewise.
4355 (maybe_assemble_visibility): Likewise.
4356 * varpool.c (ctor_for_folding): Likewise.
4357 * chkp-builtins.def: Remove.
4358 * ipa-chkp.c: Remove.
4359 * ipa-chkp.h: Remove.
4360 * rtl-chkp.c: Remove.
4361 * rtl-chkp.h: Remove.
4362 * tree-chkp-opt.c: Remove.
4363 * tree-chkp.c: Remove.
4364 * tree-chkp.h: Remove.
4366 2018-06-07 Carl Love <cel@us.ibm.com>
4368 * gcc/config/rs6000/vsx.md (vextract_fp_from_shorth,
4369 vextract_fp_from_shortl): Add BE support.
4371 2018-06-07 Paul Koning <ni1d@arrl.net>
4373 * compare-elim.c (try_merge_compare): Don't merge compare if
4374 address contains a side effect.
4375 (try_eliminate_compare): Likewise.
4377 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
4379 * config.gcc: Support "tremont".
4380 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
4381 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
4383 * config/i386/i386.c (m_TREMONT): Define.
4384 (processor_target_table): Add "tremont".
4385 (PTA_TREMONT): Define.
4386 (ix86_lea_outperforms): Add TARGET_TREMONT.
4387 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
4388 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
4389 and M_INTEL_GOLDMONT_PLUS.
4390 (fold_builtin_cpu): Add "tremont".
4391 (ix86_add_stmt_cost): Add TARGET_TREMONT.
4392 (ix86_option_override_internal): Add "tremont".
4393 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
4394 (processor_type): Add PROCESSOR_TREMONT.
4395 * config/i386/x86-tune.def: Add m_TREMONT.
4396 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
4398 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
4400 * gcc/config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
4401 symbol defined for msp430i* devices to be lower case.
4403 gcc/testsuite/gcc.target/msp430/
4404 * msp430i-device-symbol.c: New test.
4405 * msp430f-device-symbol.c: New test.
4406 * msp430.h: New test header file.
4408 2018-06-07 Richard Biener <rguenther@suse.de>
4410 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
4411 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
4412 Properly wrap signed arithmetic if overflow wraps.
4414 2018-06-07 Jakub Jelinek <jakub@redhat.com>
4416 PR tree-optimization/69615
4417 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
4418 of a cast from a same precision integral SSA_NAME in a bb dominated
4419 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
4420 cast to utype if rhs2 has already a compatible type.
4422 2018-06-07 Richard Biener <rguenther@suse.de>
4424 PR tree-optimization/85935
4425 * graphite-scop-detection.c (find_params_in_bb): Analyze
4426 condition operands with respect to the correct loop. Assert
4427 the analysis doesn't fail.
4429 2018-06-04 Carl Love <cel@us.ibm.com>
4431 * gcc/config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
4432 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
4433 as it is slightly cheaper.
4434 (first_match_or_eos_index_<mode>):
4435 Calculate index using natural element order.
4436 (first_match_index_<mode>):
4437 Calculate index using natural element order.
4438 (first_match_or_eos_index_<mode>):
4439 Calculate index using natural order.
4440 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
4441 for BE and LE modes.
4442 * gcc/config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
4443 P9V_BUILTIN_VCLZLSBB_V16QI.
4444 * gcc/config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
4447 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
4449 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
4450 indentation and line wrap for many prototypes. Add missing
4451 @smallexample directives around block of prototypes for vec_xl and
4454 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
4456 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
4457 track if we pass or return IEEE 128-bit floating point.
4458 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
4459 C++ mangling that is compatible with GCC 8.1.
4460 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
4461 (init_cumulative_args): Note if we pass or return IEEE 128-bit
4462 floating point types.
4463 (rs6000_function_arg_advance_1): Likewise.
4464 (rs6000_mangle_type): Optionally generate mangled names that match
4465 what GCC 8.1 generated for IEEE 128-bit floating point types.
4466 (rs6000_globalize_decl_name): If we have an external function that
4467 passes or returns IEEE 128-bit types, generate a weak reference
4468 from the mangled name used in GCC 8.1 to the current mangled
4470 (rs6000_init_builtins): Make __ibm128 use the long double type if
4471 long double is IBM extended double. Make __float128 use the long
4472 double type if long double is IEEE 128-bit.
4475 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
4476 macro for __ibm128 built-in functions.
4477 (PACK_IF): Add __ibm128 pack/unpack functions.
4478 (UNPACK_IF): Likewise.
4479 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
4480 enable long double built-in functions if long double is IEEE
4481 128-bit floating point.
4482 (rs6000_invalid_builtin): Update long double built-in function
4484 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
4485 functions, adjust the built-in function to use the long double
4486 built-in function if __ibm128 and long double are the same type.
4487 * doc/extend.texi (PowerPC builtins): Update documention for
4488 __builtin_{,un}pack_longdouble. Add documentation for
4489 __builtin_{,un}pack_ibm128.
4491 2018-06-06 Jim Wilson <jimw@sifive.com>
4493 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
4494 (struct machine_function): New field interrupt_mode.
4495 (riscv_handle_type_attribute): New function. Add forward declaration.
4496 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
4497 (riscv_expand_epilogue): Check interrupt_mode field.
4498 (riscv_set_current_function): Check interrupt attribute args and
4499 set interrupt_mode field.
4500 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
4501 (riscv_sret, riscv_uret): New.
4502 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
4503 new arguments to interrupt attribute.
4505 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
4508 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
4509 Don't handle -mcpu=power8 if -mpower9-vector is also used.
4511 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
4513 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
4514 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
4515 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
4516 several redundant entries.
4518 2018-06-06 David Malcolm <dmalcolm@redhat.com>
4520 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
4521 type from "rtx" to "rtx_insn *".
4522 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
4523 for local "call_insn", removing cast.
4524 (ix86_expand_call): Likewise, introducing a "call_insn" local.
4526 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
4528 PR tree-optimization/86066
4529 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
4530 for BIT_INSERT_EXPR stores.
4532 2018-06-06 Richard Biener <rguenther@suse.de>
4534 PR tree-optimization/86062
4535 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
4536 component refs ontop
4537 of to be offsetted base.
4539 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
4541 * gcc/config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
4542 to be static and remove check on interrupt attribute name.
4544 gcc/testsuite/gcc.target/msp430/
4545 * function-attributes-4.c: New test.
4546 * static-interrupts.c: New test.
4548 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
4550 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
4551 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
4553 2018-06-05 Steve Ellcey <sellcey@cavium.com>
4556 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
4558 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
4559 Remove second argument, change how error is called.
4560 (aarch64_layout_arg): Remove second argument from
4561 aarch64_err_no_fpadvsimd call.
4562 (aarch64_init_cumulative_args): Ditto.
4563 (aarch64_gimplify_va_arg_expr): Ditto.
4564 * config/aarch64/aarch64.md (mov<mode>): Ditto.
4566 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
4568 * config/i386/i386.md (simple_return_indirect_internal): New expander.
4569 (*simple_return_indirect_internal<mode>): Rename from
4570 simple_return_indirect_internal. Use W mode iterator.
4571 (rstorssp): New expander.
4572 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
4573 (clrssbsy): New expander.
4574 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
4576 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
4578 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
4580 (cmse_check_pointed_object): Likewise.
4582 2018-06-05 Martin Liska <mliska@suse.cz>
4584 PR gcov-profile/47618
4585 * doc/invoke.texi: Document how -fprofile-dir format
4588 2018-06-05 Richard Biener <rguenther@suse.de>
4590 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
4591 removal pretend DOM info isn't available so we do not update
4592 it and only remove edges, not dominated blocks. Actually free
4593 DOM info in case we removed something. Remove unreachable blocks.
4594 (mfb_keep_latches): Work with either DOM info or marked backedges.
4595 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
4596 first. Mark backedges if DOM info isn't available.
4597 (Re-)compute DOM info after cleanup_control_flow_pre.
4599 2018-06-05 Richard Biener <rguenther@suse.de>
4601 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
4602 (locus_discrim_hasher::hash): Adjust.
4603 (locus_discrim_hasher::equal): Likewise.
4604 (next_discriminator_for_locus): Work on line directly.
4605 (same_line_p): Pass in expanded locus1 as well.
4606 (assign_discriminators): Avoid redundant location expansions.
4608 2018-06-05 Richard Biener <rguenther@suse.de>
4610 PR tree-optimization/86046
4611 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
4612 if required after clearing TREE_ADDRESSABLE.
4614 2018-06-05 Richard Biener <rguenther@suse.de>
4616 PR tree-optimization/86047
4617 * tree-ssa-loop.c (for_each_index): Glob handling of all
4618 decls and constants and really handle all of them.
4620 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4623 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
4624 qualifier_void_pointer and qualifier_const_void_pointer.
4625 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
4626 (arm_init_builtins): Handle the above.
4627 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
4628 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
4629 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
4632 2018-06-05 Martin Liska <mliska@suse.cz>
4634 * auto-profile.c (read_autofdo_file): Do not use
4635 gcov_ctr_summary struct.
4636 (afdo_callsite_hot_enough_for_early_inline): Likewise.
4637 * coverage.c (struct counts_entry): Likewise.
4638 (read_counts_file): Read just single summary entry.
4639 (get_coverage_counts): Use gcov_summary struct.
4640 * coverage.h (get_coverage_counts): Likewise.
4641 * gcov-dump.c (dump_working_sets): Likewise.
4642 (tag_summary): Dump just single summary.
4643 * gcov-io.c (gcov_write_summary): Write just histogram
4645 (gcov_read_summary): Read just single summary.
4646 (compute_working_sets): Use gcov_summary struct.
4647 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
4648 of GCOV_COUNTERS_SUMMABLE.
4649 (GCOV_COUNTERS_SUMMABLE): Remove.
4650 (GCOV_FIRST_VALUE_COUNTER): Replace with
4651 GCOV_COUNTER_V_INTERVAL.
4652 (struct gcov_ctr_summary): Remove.
4653 (struct gcov_summary): Directly use fields of former
4655 (compute_working_sets): Use gcov_summary struct.
4656 * gcov.c (read_count_file): Do not use ctrs fields.
4657 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
4659 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
4661 * profile.c: Likewise.
4662 * profile.h: Likewise.
4664 2018-06-05 Martin Liska <mliska@suse.cz>
4666 PR gcov-profile/84846
4667 * gcov.c (output_lines): Print working directory only
4668 in intermediate format.
4670 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
4672 * config/s390/s390-builtin-types.def: Add void function type.
4673 * config/s390/s390-builtins.def: Use the function type for the
4676 2018-06-04 Jim Wilson <jimw@sifive.com>
4678 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
4680 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
4681 and maybe_eh_return. Change regno to unsigned int. Use new args to
4682 handle EH_RETURN_DATA_REGNO registers properly.
4683 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
4684 (riscv_expand_epilogue): Update comment. Change argument name and
4685 type. Update code to use new name and type. Pass new args to
4686 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
4688 * config/riscv/riscv.md (NORMAL_RETURN): New.
4689 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
4690 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
4691 (eh_return): Call gen_eh_return_internal and emit barrier.
4692 (eh_return_internal): Call riscv_expand_epilogue.
4694 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
4696 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
4697 bit_insertion field and declare can_be_merged_into method.
4698 (merged_store_group::can_be_merged_into): New method.
4699 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
4700 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
4701 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
4703 2018-06-04 Richard Biener <rguenther@suse.de>
4705 PR tree-optimization/85955
4706 * builtins.c (fold_builtin_sincos): Convert pointers to
4707 destination to appropriate type before dereferencing.
4709 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
4711 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
4713 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
4715 * expr.c (expand_expr_real_1): Force the operand into memory if
4716 its TYPE_MODE is BLKmode and if there is no integer mode for
4717 the number of bits being extracted.
4719 2018-06-04 Jakub Jelinek <jakub@redhat.com>
4723 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
4724 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
4726 2018-06-04 Richard Biener <rguenther@suse.de>
4728 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
4729 (cleanup_tree_cfg_noloop): ... single caller. Do
4730 start_recording_case_labels later.
4732 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
4734 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
4735 to _IMMINTRIN_H_INCLUDED.
4736 * config/i386/pconfigintrin.h: Ditto.
4737 * config/i386/waitpkgintrin.h: Ditto.
4738 * config/i386/immintrin.h: Add includes for sgxintrin.h,
4739 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
4740 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
4741 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
4742 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
4743 waitpkgintrin.h and cldemoteintrin.h.
4745 2018-06-04 Richard Biener <rguenther@suse.de>
4747 PR tree-optimization/86038
4748 * tracer.c (find_best_successor): Check probability for
4749 being initialized, bail out if not.
4751 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
4754 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
4755 of bits to ignore when comparing architectures.
4757 2018-06-04 Jakub Jelinek <jakub@redhat.com>
4759 PR tree-optimization/69615
4760 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
4761 maximum or minimum of the type, try to merge it also as if
4762 range1 is + [-, x - 1] or + [x + 1, -].
4765 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
4767 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
4769 PR tree-optimization/86034
4770 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
4771 the unsigned bitfield type in a bit insertion sequence if it does not
4772 have a larger precision than the bitfield size.
4773 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
4775 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
4777 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
4779 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
4781 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
4782 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
4783 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
4784 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
4786 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
4788 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
4789 Disable -fdelete-null-pointer-checks for ELF toolchain.
4791 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
4792 Kito Cheng <kito.cheng@gmail.com>
4794 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
4795 (nds32le-*-*, nds32be-*-*): Integrate checking process.
4796 (nds32*-*-*): Add glibc and uclibc conditions.
4797 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
4798 (TARGET_EXCEPT_UNWIND_INFO): Define.
4799 * config/nds32/elf.h: New file.
4800 * config/nds32/linux.h: New file.
4801 * config/nds32/nds32-elf.opt: New file.
4802 * config/nds32/nds32-linux.opt: New file.
4803 * config/nds32/nds32-fp-as-gp.c
4804 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
4805 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
4807 (nds32_asm_file_end): Ditto.
4808 (nds32_print_operand): Ditto.
4809 (nds32_insert_attributes): Ditto.
4810 (nds32_init_libfuncs): New function.
4811 (TARGET_HAVE_TLS): Define.
4812 (TARGET_INIT_LIBFUNCS): Define.
4813 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
4815 (TARGET_ELF): Apply different mcmodel setting.
4816 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
4817 been migrated into elf.h and linux.h files.
4818 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
4819 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
4820 (mcmodel): The content has been migrated into nds32-elf.opt and
4821 nds32-linux.opt files.
4822 * config/nds32/t-elf: New file.
4823 * config/nds32/t-linux: New file.
4825 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
4826 Shiva Chen <shiva0217@gmail.com>
4828 * config/nds32/constants.md (unspec_volatile_element): Add
4829 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
4830 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
4832 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
4833 (make_pass_nds32_fp_as_gp): Declare.
4834 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
4836 (nds32_asm_function_end_prologue): Remove unused asm output.
4837 (nds32_asm_function_begin_epilogue): Remove unused asm output.
4838 (nds32_asm_file_start): Output necessary fp_as_gp information.
4839 (nds32_option_override): Adjust register usage.
4840 (nds32_expand_prologue): Consider fp_as_gp situation.
4841 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
4842 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
4845 (simple_return): Ditto.
4846 (omit_fp_begin): Output special directive for fp_as_gp.
4847 (omit_fp_end): Output special directive for fp_as_gp.
4848 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
4849 mforbid-fp-as-gp): New options.
4851 2018-06-01 Mark Wielaard <mark@klomp.org>
4853 * dwarf2out.c (dwarf2out_finish): Remove generation of
4854 DW_AT_loclists_base.
4856 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
4858 * gimple-ssa-store-merging.c: Include gimple-fold.h.
4859 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
4860 (struct merged_store_group): Add bit_insertion field.
4861 (dump_char_array): Use standard hexadecimal format.
4862 (merged_store_group::merged_store_group): Set bit_insertion to false.
4863 (merged_store_group::apply_stores): Use optimal buffer size. Deal
4864 with BIT_INSERT_EXPR stores. Move up code updating the mask and
4865 also print the mask in the dump file.
4866 (pass_store_merging::gate): Minor tweak.
4867 (imm_store_chain_info::coalesce_immediate): Fix wrong association
4868 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
4869 stores with INTEGER_CST stores.
4870 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
4871 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
4872 and use it throughout. Generate bit insertion sequences if need be.
4873 (pass_store_merging::process_store): Remove redundant condition.
4874 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
4876 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
4878 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
4879 the 128-bit floating point types. Fix function comment.
4881 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4883 * config/aarch64/aarch64-simd.md
4884 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
4886 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
4889 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
4891 PR tree-optimization/85989
4892 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
4894 (backprop::intersect_uses): Check it when deciding whether this
4895 is a backedge reference.
4896 (backprop::process_block): Add each phi to m_visited_phis
4897 after visiting it, then clear it at the end.
4899 2018-06-01 Richard Biener <rguenther@suse.de>
4901 * tree-vectorizer.h (vect_dr_stmt): New function.
4902 (vect_get_load_cost): Adjust.
4903 (vect_get_store_cost): Likewise.
4904 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
4905 Use vect_dr_stmt instead of DR_SMTT.
4906 (vect_record_base_alignments): Likewise.
4907 (vect_calculate_target_alignment): Likewise.
4908 (vect_compute_data_ref_alignment): Likewise and make static.
4909 (vect_update_misalignment_for_peel): Likewise.
4910 (vect_verify_datarefs_alignment): Likewise.
4911 (vector_alignment_reachable_p): Likewise.
4912 (vect_get_data_access_cost): Likewise. Pass down
4913 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
4914 (vect_get_peeling_costs_all_drs): Likewise.
4915 (vect_peeling_hash_get_lowest_cost): Likewise.
4916 (vect_enhance_data_refs_alignment): Likewise.
4917 (vect_find_same_alignment_drs): Likewise.
4918 (vect_analyze_data_refs_alignment): Likewise.
4919 (vect_analyze_group_access_1): Likewise.
4920 (vect_analyze_group_access): Likewise.
4921 (vect_analyze_data_ref_access): Likewise.
4922 (vect_analyze_data_ref_accesses): Likewise.
4923 (vect_vfa_segment_size): Likewise.
4924 (vect_small_gap_p): Likewise.
4925 (vectorizable_with_step_bound_p): Likewise.
4926 (vect_prune_runtime_alias_test_list): Likewise.
4927 (vect_analyze_data_refs): Likewise.
4928 (vect_supportable_dr_alignment): Likewise.
4929 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
4930 (vect_gen_prolog_loop_niters): Likewise.
4931 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
4932 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
4934 (vect_recog_mask_conversion_pattern): Likewise.
4935 (vect_try_gather_scatter_pattern): Likewise.
4936 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
4937 to vect_get_store_cost.
4938 (vect_get_store_cost): Get stmt_info instead of DR.
4939 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
4940 (vect_get_load_cost): Get stmt_info instead of DR.
4942 2018-06-01 Richard Biener <rguenther@suse.de>
4945 * gimple-fold.c (var_decl_component_p): Also allow offsetted
4946 vars wrapped in MEM_REFs.
4948 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
4950 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
4951 Fix subreg tests so that we only return a choice between
4952 GENERAL_REGS and FP_REGS if the original classes included both.
4954 2018-06-01 Richard Biener <rguenther@suse.de>
4957 * tree-ssa-structalias.c (get_function_part_constraint):
4958 Handle NULL fi->decl.
4959 (find_func_aliases_for_call): Properly handle indirect
4960 fi from direct call.
4961 (find_func_clobbers): Likewise.
4962 (ipa_pta_execute): Likewise.
4963 (create_variable_info_for): For functions that are ifunc_resolver
4964 resolve to a varinfo that contains the result of the resolver call.
4965 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
4968 2018-05-31 Michael Collison <michael.collison@arm.com>
4970 * config/aarch64/aarch64.md:
4971 (*fix_to_zero_extenddfdi2): New pattern.
4972 * gcc.target/aarch64/fix_extend1.c: New testcase.
4974 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
4978 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
4979 and BUILT_IN_STRNCMP_EQ.
4980 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
4981 BUILT_IN_STRNCMP_EQ.
4982 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
4983 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
4984 (gimple_fold_builtin): Likewise.
4985 * tree-ssa-strlen.c (compute_string_length): New function.
4986 (determine_min_obsize): New function.
4987 (handle_builtin_string_cmp): New function to handle calls to
4988 string compare functions.
4989 (strlen_optimize_stmt): Add handling to builtin string compare
4991 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
4992 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
4993 * tree.c (build_common_builtin_nodes): Add new defines of
4994 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
4996 2018-05-31 Jakub Jelinek <jakub@redhat.com>
4999 * bb-reorder.c (pass_partition_blocks::gate): Return false for
5000 functions with naked attribute.
5002 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
5004 * config/i386/sse.md (avx_vec_concat<mode>):
5005 Substitute concat_tg_mode mode attribute with xtg_mode.
5006 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
5007 (concat_tg_mode): Remove mode attribute.
5009 2018-05-31 Martin Sebor <msebor@redhat.com>
5012 * calls.c (alloc_max_size): Correct a logic error/typo.
5013 Treat excessive arguments as infinite. Warn for invalid arguments.
5014 * doc/invoke.texi (-Walloc-size-larger-than): Update.
5016 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
5019 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
5020 and movx for Haswell.
5022 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
5023 Cesar Philippidis <cesar@codesourcery.com>
5026 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
5027 when emitting error on private/firstprivate reductions.
5028 * omp-low.c (lower_omp_target): Avoid reference-type processing
5029 on pointers for firstprivate clause.
5031 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
5033 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
5036 * config/aarch64/aarch64-simd.md
5037 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
5038 (aarch64_ld1_x3_<mode>): Likewise
5039 (aarch64_st1x2<VALLDIF:mode>): Likewise
5040 (aarch64_st1_x2_<mode>): Likewise
5041 (aarch64_st1x3<VALLDIF:mode>): Likewise
5042 (aarch64_st1_x3_<mode>): Likewise
5043 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
5044 (vld1_s8_x3): Likewise.
5045 (vld1_u16_x3): Likewise.
5046 (vld1_s16_x3): Likewise.
5047 (vld1_u32_x3): Likewise.
5048 (vld1_s32_x3): Likewise.
5049 (vld1_u64_x3): Likewise.
5050 (vld1_s64_x3): Likewise.
5051 (vld1_f16_x3): Likewise.
5052 (vld1_f32_x3): Likewise.
5053 (vld1_f64_x3): Likewise.
5054 (vld1_p8_x3): Likewise.
5055 (vld1_p16_x3): Likewise.
5056 (vld1_p64_x3): Likewise.
5057 (vld1q_u8_x3): Likewise.
5058 (vld1q_s8_x3): Likewise.
5059 (vld1q_u16_x3): Likewise.
5060 (vld1q_s16_x3): Likewise.
5061 (vld1q_u32_x3): Likewise.
5062 (vld1q_s32_x3): Likewise.
5063 (vld1q_u64_x3): Likewise.
5064 (vld1q_s64_x3): Likewise.
5065 (vld1q_f16_x3): Likewise.
5066 (vld1q_f32_x3): Likewise.
5067 (vld1q_f64_x3): Likewise.
5068 (vld1q_p8_x3): Likewise.
5069 (vld1q_p16_x3): Likewise.
5070 (vld1q_p64_x3): Likewise.
5071 (vst1_s64_x2): Likewise.
5072 (vst1_u64_x2): Likewise.
5073 (vst1_f64_x2): Likewise.
5074 (vst1_s8_x2): Likewise.
5075 (vst1_p8_x2): Likewise.
5076 (vst1_s16_x2): Likewise.
5077 (vst1_p16_x2): Likewise.
5078 (vst1_s32_x2): Likewise.
5079 (vst1_u8_x2): Likewise.
5080 (vst1_u16_x2): Likewise.
5081 (vst1_u32_x2): Likewise.
5082 (vst1_f16_x2): Likewise.
5083 (vst1_f32_x2): Likewise.
5084 (vst1_p64_x2): Likewise.
5085 (vst1q_s8_x2): Likewise.
5086 (vst1q_p8_x2): Likewise.
5087 (vst1q_s16_x2): Likewise.
5088 (vst1q_p16_x2): Likewise.
5089 (vst1q_s32_x2): Likewise.
5090 (vst1q_s64_x2): Likewise.
5091 (vst1q_u8_x2): Likewise.
5092 (vst1q_u16_x2): Likewise.
5093 (vst1q_u32_x2): Likewise.
5094 (vst1q_u64_x2): Likewise.
5095 (vst1q_f16_x2): Likewise.
5096 (vst1q_f32_x2): Likewise.
5097 (vst1q_f64_x2): Likewise.
5098 (vst1q_p64_x2): Likewise.
5099 (vst1_s64_x3): Likewise.
5100 (vst1_u64_x3): Likewise.
5101 (vst1_f64_x3): Likewise.
5102 (vst1_s8_x3): Likewise.
5103 (vst1_p8_x3): Likewise.
5104 (vst1_s16_x3): Likewise.
5105 (vst1_p16_x3): Likewise.
5106 (vst1_s32_x3): Likewise.
5107 (vst1_u8_x3): Likewise.
5108 (vst1_u16_x3): Likewise.
5109 (vst1_u32_x3): Likewise.
5110 (vst1_f16_x3): Likewise.
5111 (vst1_f32_x3): Likewise.
5112 (vst1_p64_x3): Likewise.
5113 (vst1q_s8_x3): Likewise.
5114 (vst1q_p8_x3): Likewise.
5115 (vst1q_s16_x3): Likewise.
5116 (vst1q_p16_x3): Likewise.
5117 (vst1q_s32_x3): Likewise.
5118 (vst1q_s64_x3): Likewise.
5119 (vst1q_u8_x3): Likewise.
5120 (vst1q_u16_x3): Likewise.
5121 (vst1q_u32_x3): Likewise.
5122 (vst1q_u64_x3): Likewise.
5123 (vst1q_f16_x3): Likewise.
5124 (vst1q_f32_x3): Likewise.
5125 (vst1q_f64_x3): Likewise.
5126 (vst1q_p64_x3): Likewise.
5128 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
5130 * config/msp430/msp430.c (msp430_output_labelref): Prepend
5131 user_label_prefix to name.
5133 * tree-core.h: Update comment about the format of NAME string
5134 passed to handler in attribute_spec.
5136 * config/msp430/msp430.md: Remove erroneous subreg expression from
5137 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
5138 zero_extend{q,h}isi2.
5140 2018-05-30 Borislav Petkov <bp@suse.de>
5142 * doc/extend.texi: Document some architecture specific
5143 constraints and sort entries.
5145 2018-05-30 Martin Sebor <msebor@redhat.com>
5148 * builtins.c (expand_builtin_stpcpy_1): New function.
5149 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
5150 only if the former succeeds.
5152 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
5154 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
5157 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
5159 * doc/invoke.texi (-flinker-output): Document
5161 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
5163 * passes.c (ipa_write_summaries): Only modify statements if body
5165 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
5166 incrementally linking.
5167 (ipa_passes): Likewise.
5168 * lto-cgraph.c (lto_output_node): When incrementally linking do not
5169 pass down resolution info.
5170 * common.opt (flag_incremental_link): Update info.
5171 * gcc.c (plugin specs): Turn flinker-output=* to
5172 -plugin-opt=-linker-output-known
5173 * toplev.c (compile_file): Also cut compilation when doing incremental
5175 * flag-types. (enum lto_partition_model): Add
5176 LTO_LINKER_OUTPUT_NOLTOREL.
5177 (invoke.texi): Add -flinker-output docs.
5178 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
5179 link same way as WPA; do not stream in dead initializers.
5181 * dwarf2out.c (dwarf2out_die_ref_for_decl,
5182 darf2out_register_external_decl): Support incremental link.
5184 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
5186 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
5188 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
5190 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
5191 it down to simple_object_copy_lto_debug_sections.
5192 (run_gcc): Determine incremental LTO link time and configure
5193 lto1 into non-wpa mode, disable renaming of debug sections.
5195 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
5197 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
5198 descriptions of various incorrectly documented functions.
5200 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
5203 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
5204 address check not strict.
5206 2018-05-30 Richard Biener <rguenther@suse.de>
5208 PR tree-optimization/85964
5209 * tracer.c (better_p): Drop initialized count check, we only
5210 call the function with initialized counts now.
5211 (find_best_successor): Do find a best edge if one
5212 has uninitialized count.
5213 (find_best_predecessor): Likewise. Do BB frequency check only
5214 if count is initialized.
5216 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
5218 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
5219 (aarch64_ldrstr_offset_compare): New.
5220 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
5221 load/store orderings.
5222 (aarch64_gen_adjusted_ldpstp): Likewise.
5224 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
5226 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
5227 Check for subset of GENERAL_REGS and FP_REGS.
5228 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
5231 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
5233 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
5234 and wi::to_poly_offset. Add the current offset and then check
5235 whether the sum fits, rather than using an unchecked addition of
5236 a checked term. Check for a shwi rather than a uhwi.
5237 * expr.c (get_bit_range): Use tree_to_poly_uint64.
5238 (store_constructor): Use poly_int_tree_p.
5239 (expand_expr_real_1): Likewise.
5240 * function.c (assign_temp): Likewise.
5241 * fold-const.c (const_binop): Use poly_int_tree_p and
5243 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
5245 * ipa-icf-gimple.c (func_checker::compare_operand): Use
5246 to_poly_offset for MEM offsets.
5247 * ipa-icf.c (sem_variable::equals): Likewise.
5248 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
5249 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
5250 wi::to_poly_offset for BIT_FIELD_REF offsets.
5251 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
5253 * var-tracking.c (emit_note_insn_var_location): Use
5254 tree_to_poly_uint64.
5256 2018-05-29 Jim Wilson <jimw@sifive.com>
5258 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
5260 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
5263 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
5264 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
5266 (sse4_1_round<mode>2): Use nonimmediate_operand
5267 for operand 1 predicate.
5269 2018-05-29 Martin Sebor <msebor@redhat.com>
5270 Richard Biener <rguenther@suse.de>
5273 * calls.c (get_size_range): Call determine_value_range instead
5274 of get_value_range..
5275 * tree-vrp.h (determine_value_range): Declared new function.
5276 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
5278 2018-05-29 Richard Biener <rguenther@suse.de>
5280 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
5281 sure to use non-pattern stmts for get_earlier_stmt arguments.
5282 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
5283 called on pattern stmts.
5284 (get_later_stmt): Likewise.
5286 2018-05-29 Martin Liska <mliska@suse.cz>
5288 PR gcov-profile/85759
5289 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
5292 2018-05-29 Jakub Jelinek <jakub@redhat.com>
5294 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
5296 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
5300 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
5301 VEC_PACK_FLOAT_EXPR): New tree codes.
5302 * tree-pretty-print.c (op_code_prio): Handle
5303 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
5304 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
5305 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
5306 * tree-inline.c (estimate_operator_cost): Likewise.
5307 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
5308 * fold-const.c (const_binop): Likewise.
5309 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
5310 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
5311 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5312 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
5313 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
5314 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
5315 * expr.c (expand_expr_real_2): Likewise.
5316 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
5317 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
5318 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
5320 * optabs.c (expand_widen_pattern_expr): For
5321 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
5322 sign from result type rather than operand's type.
5323 (expand_binop_directly): For vec_packu_float_optab and
5324 vec_packs_float_optab allow result type to be different from operand's
5326 * optabs-tree.c (optab_for_tree_code): Handle
5327 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
5328 VEC_PACK_FLOAT_EXPR. Formatting fixes.
5329 * tree-vect-generic.c (expand_vector_operations_1): Handle
5330 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
5331 VEC_PACK_FLOAT_EXPR.
5332 * tree-vect-stmts.c (supportable_widening_operation): Handle
5334 (supportable_narrowing_operation): Handle FLOAT_EXPR.
5335 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
5336 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
5337 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
5338 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
5340 (vec_pack<floatprefix>_float_<mode>): New expander.
5341 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
5343 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
5344 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
5345 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
5346 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
5347 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
5349 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
5350 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
5351 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
5352 VEC_PACK_FLOAT_EXPR): Document.
5354 2018-05-29 Richard Biener <rguenther@suse.de>
5356 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
5358 (stmt_vec_info_vec): Make pointer.
5359 (init_stmt_vec_info_vec): Remove.
5360 (free_stmt_vec_info_vec): Likewise.
5361 (set_stmt_vec_info_vec): New function.
5362 (free_stmt_vec_infos): Likewise.
5363 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
5364 (set_vinfo_for_stmt): Likewise.
5365 (get_earlier_stmt): Likewise.
5366 (get_later_stmt): Likewise.
5367 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
5368 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
5369 (vec_info::~vec_info): Free stmt_vec_infos.
5370 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
5371 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
5372 (pass_slp_vectorize::execute): Likewise.
5373 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
5374 (free_stmt_vec_info_vec): Likewise.
5375 (set_stmt_vec_info_vec): New function.
5376 (free_stmt_vec_infos): Likewise.
5377 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
5378 the global stmt_vec_info_vec.
5379 * tree-parloops.c (gather_scalar_reductions): Use
5380 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
5383 2018-05-29 Richard Biener <rguenther@suse.de>
5385 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
5387 2018-05-29 Martin Liska <mliska@suse.cz>
5388 David Malcolm <dmalcolm@redhat.com>
5390 * vec.c (test_reverse): New.
5391 (vec_c_tests): Add new test.
5392 * vec.h (vl_ptr>::reverse): New function.
5394 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
5396 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
5398 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
5401 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
5403 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
5405 2018-05-28 Richard Biener <rguenther@suse.de>
5407 PR tree-optimization/85933
5408 * tree-vect-data-refs.c (vect_record_base_alignments): Only
5409 look at stmts marked as vectorizable.
5411 2018-05-28 Richard Biener <rguenther@suse.de>
5413 PR tree-optimization/85934
5414 * tree-vect-generic.c (expand_vector_operations_1): Hoist
5415 vector boolean check before scalar optimization.
5417 2018-05-28 Jakub Jelinek <jakub@redhat.com>
5419 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
5422 2018-05-28 Mark Wielaard <mark@klomp.org>
5424 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
5425 if it is an expression containing a minus sign.
5427 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
5429 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
5431 2018-05-27 Paul Koning <ni1d@arrl.net>
5433 * config/pdp11/pdp11.md (truncsihi2): Remove.
5435 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
5436 Chung-Ju Wu <jasonwucj@gmail.com>
5438 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
5440 (unaligned_store_dw): Ditto.
5441 * config/nds32/nds32-memory-manipulation.c
5442 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
5443 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
5444 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
5445 (emit_setmem_word_loop): Rename to ...
5446 (emit_setmem_doubleword_loop): ... this.
5447 (nds32_gen_dup_4_byte_to_word_value): New function.
5448 (nds32_gen_dup_8_byte_to_double_word_value): New function.
5449 (nds32_expand_setmem_loop): Refine implementation.
5450 (nds32_expand_setmem_loop_v3m): Ditto.
5451 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
5454 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
5456 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
5458 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
5460 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
5461 (nds32_init_machine_status): Initialize machine->attr_naked_p and
5462 machine->attr_no_prologue_p.
5463 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
5464 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
5465 (nds32_expand_epilogue): Consider attr_naked_p.
5466 (nds32_expand_epilogue_v3pop): Likewise.
5467 (nds32_can_use_return_insn): Likewise.
5468 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
5469 attr_no_prologue_p fields.
5470 * config/nds32/nds32.opt (mret-in-naked-func): New option.
5472 2018-05-27 Jakub Jelinek <jakub@redhat.com>
5475 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
5477 * config/i386/sse.md
5478 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
5480 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
5482 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
5484 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
5486 (*<floatsuffix>floatv2div2sf2): Rename to ...
5487 (*float<floatunssuffix>v2div2sf2): ... this.
5488 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
5489 (float<floatunssuffix>v2div2sf2_mask): ... this.
5490 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
5491 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
5492 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
5494 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
5496 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
5498 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
5500 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
5502 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
5504 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
5505 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
5506 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
5507 gen_ufix_truncv8dfv8si2.
5508 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
5509 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
5510 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
5511 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
5512 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
5513 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
5514 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
5515 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
5517 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
5521 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
5523 2018-05-25 Jim Wilson <jimw@sifive.com>
5525 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
5526 * config/riscv/riscv.c (struct machine_function): Add
5527 interrupt_handler_p and attribute_checked_p fields.
5528 (riscv_attribute_table): Add interrupt.
5529 (riscv_interrupt_type_p): New.
5530 (riscv_save_reg_p): Save extra regs for interrupt handler.
5531 (riscv_use_save_libcall): Return false for interrupt handler.
5532 (riscv_first_stack_step): Add forward declaration.
5533 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
5534 for interrupt handler with large frame. Use it for saved reg list.
5535 (riscv_expand_prologue): Move flag_stack_usage_info support to
5536 eliminate duplication.
5537 (riscv_expand_epilogue): Generate mret for interrupt handler.
5538 (riscv_epilogue_uses): New.
5539 (riscv_can_use_return_insn): Return false for interrupt handler.
5540 (riscv_function_ok_for_sibcall): Likewise.
5541 (riscv_set_current_function): Add interrupt handler support.
5542 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
5543 * config/riscv/riscv.md (UNSPECV_MRET): New.
5545 (riscv_frflags, riscv_fsflags): Use tab after opcode.
5547 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
5549 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
5551 PR tree-optimization/85712
5552 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
5553 this candidate has already been replaced in-situ by a copy.
5555 2018-05-25 Jason Merrill <jason@redhat.com>
5557 PR c++/80485 - inline function non-zero address.
5558 * symtab.c (nonzero_address): Check DECL_COMDAT.
5560 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
5563 * config/alpha/alpha.md (ashlsi3): New insn pattern.
5564 (*ashlsi_se): Rename from *ashldi_se. Define as sign
5565 extension of SImode operation. Use const123_operand predicate.
5566 (*saddsi_1): Remove.
5567 (*saddl_se_1): Ditto.
5569 (*ssubl_se_1): Ditto.
5570 * config/alpha/predicates.md (const123_operand): New predicate.
5571 * config/alpha/constraints.md (P): Use IN_RANGE.
5573 2018-05-25 Richard Biener <rguenther@suse.de>
5575 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
5577 (ref_maybe_used_by_stmt_p): Likewise.
5578 (stmt_may_clobber_ref_p): Likewise.
5579 (stmt_may_clobber_ref_p_1): Likewise.
5580 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
5582 (ref_maybe_used_by_stmt_p): Likewise.
5583 (stmt_may_clobber_ref_p): Likewise.
5584 (stmt_may_clobber_ref_p_1): Likewise.
5585 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
5586 the alias oracle to disambiguate DRs with stmts DR analysis
5588 (vect_analyze_data_refs): Do not give up on not analyzable
5589 DRs for BB vectorization. Remove code truncating the dataref
5592 2018-05-25 Jakub Jelinek <jakub@redhat.com>
5595 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
5596 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
5597 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
5599 2018-05-25 Richard Biener <rguenther@suse.de>
5601 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
5602 function, combining stmt data ref gathering and fatal analysis
5604 (vect_analyze_data_refs): Remove now redudnant code and simplify.
5605 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
5606 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
5607 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
5608 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
5610 2018-05-25 Bin Cheng <bin.cheng@arm.com>
5612 PR tree-optimization/85720
5613 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
5614 SCC if all partitions are builtins.
5615 (version_loop_by_alias_check): New parameter. Generate cancelable
5616 runtime alias check if all partitions are builtins.
5617 (distribute_loop): Update call to above function.
5619 2018-05-25 Bin Cheng <bin.cheng@arm.com>
5621 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
5622 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
5623 (parm_default_def_partition_arg): Ditto.
5624 (set_parm_default_def_partition): Ditto.
5625 (get_parm_default_def_partitions): Ditto and make it static.
5626 (get_undefined_value_partitions): Ditto and make it static.
5627 (remove_ssa_form): Refactor call to init_var_map here.
5628 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
5629 computation for loop region.
5630 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
5631 (register_default_def): Delete.
5632 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
5633 (parm_default_def_partition_arg): Ditto.
5634 (set_parm_default_def_partition): Ditto.
5635 (get_parm_default_def_partitions): Ditto and make it static.
5636 (get_undefined_value_partitions): Ditto and make it static.
5637 (coalesce_with_default, coalesce_with_default): Update comment.
5638 (create_coalesce_list_for_region): New func factored out from
5639 create_outofssa_var_map.
5640 (populate_coalesce_list_for_outofssa): New func factored out from
5641 create_outofssa_var_map and coalesce_ssa_name.
5642 (create_outofssa_var_map): Delete.
5643 (coalesce_ssa_name): Refactor to support live range computation.
5644 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
5645 (get_parm_default_def_partitions): Delete.
5646 (get_undefined_value_partitions): Ditto.
5647 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
5648 computation for loop region.
5649 (new_tree_live_info, loe_visit_block): Ditto.
5650 (live_worklist, set_var_live_on_entry): Ditto.
5651 (calculate_live_on_exit, verify_live_on_entry): Ditto.
5652 * tree-ssa-live.h (struct _var_map): New fields.
5653 (init_var_map): Change decl.
5654 (region_contains_p): New.
5656 2018-05-25 Bin Cheng <bin.cheng@arm.com>
5658 * tree-ssa-live.h (live_merge_and_clear): Delete.
5660 2018-05-25 Richard Biener <rguenther@suse.de>
5663 * tree-dump.c (dequeue_and_dump): Remove access to removed
5664 operand 2 of a SWITCH_EXPR.
5666 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
5668 * doc/sourcebuild.texi (vect_double_cond_arith): Include
5669 multiplication and division.
5670 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
5671 (cond_udiv@var{m}, cond_umod@var{m}): Document.
5672 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
5673 (cond_udiv_optab, cond_umod_optab): New optabs.
5674 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
5675 (IFN_COND_RDIV): New internal functions.
5676 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
5677 TRUNC_MOD_EXPR and RDIV_EXPR.
5678 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
5679 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
5681 (SVE_INT_BINARY): Include mult.
5682 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
5683 (optab, sve_int_op): Handle mult.
5684 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
5686 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
5687 for SVE_INT_BINARY_SD.
5689 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
5691 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
5692 (optab, sve_int_op): Handle div and udiv.
5693 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
5694 for SVE_INT_BINARY_SD.
5695 (*<optab><mode>3): New insn for the same.
5697 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
5699 * tree-vect-patterns.c: Include predict.h.
5700 (vect_recog_divmod_pattern): Restrict check for division support
5701 to when optimizing for size.
5703 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
5705 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
5706 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
5707 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
5708 (gimple_match_op::set_op): Likewise.
5709 (gimple_resimplify4): Declare.
5710 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
5711 (expr::gen_transform): Likewise.
5712 (decision_tree::gen): Generate a simplification routine for 4 operands.
5713 * gimple-match-head.c (gimple_simplify): Add an overload for
5714 4 operands. In the top-level function, handle up to 4 call
5715 arguments and call gimple_resimplify4.
5716 (gimple_resimplify4): New function.
5717 (build_call_internal): Pass a fourth operand.
5718 (maybe_push_to_seq): Likewise.
5719 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
5720 Fold VEC_COND_EXPRs of an operation and a default value into
5721 an IFN_COND_* function if possible.
5722 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
5724 (SVE_COND_FP_BINARY): Include them.
5725 (optab, sve_fp_op): Handle them.
5726 (SVE_INT_BINARY_REV): New code iterator.
5727 (SVE_COND_FP_BINARY_REV): New int iterator.
5728 (commutative): New int attribute.
5729 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
5731 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
5733 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
5734 (*cond_<optab><mode>): New patterns for reversed operands.
5736 2018-05-25 Richard Biener <rguenther@suse.de>
5738 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
5739 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
5740 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
5741 (STMT_VINFO_GROUPED_ACCESS): Adjust.
5742 * tree-vect-data-refs.c (everywhere): Adjust users.
5743 * tree-vect-loop.c (everywhere): Likewise.
5744 * tree-vect-slp.c (everywhere): Likewise.
5745 * tree-vect-stmts.c (everywhere): Likewise.
5746 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
5748 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5750 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
5752 (gcc_cv_as_section_exclude): ... this.
5753 Try Solaris as #exclude syntax.
5754 * configure: Regenerate.
5755 * config.in: Regenerate.
5756 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
5758 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
5759 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
5761 * varasm.c (default_elf_asm_named_section): Don't check if
5762 HAVE_GAS_SECTION_EXCLUDE is defined.
5764 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
5766 * doc/md.texi: Update the documentation of the cond_* optabs
5767 to mention the new final operand. Fix GET_MODE_NUNITS call.
5768 Describe the scalar case too.
5769 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
5770 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
5772 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
5773 (get_conditional_internal_fn): Update comment.
5774 * tree-vect-loop.c (vectorizable_reduction): Pass the original
5775 accumulator value as a final argument to conditional functions.
5776 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
5777 a define_expand and add an "else" operand. Assert for now that
5778 the else operand is equal to operand 2. Use SVE_INT_BINARY and
5779 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
5780 (*cond_<optab><mode>): New patterns.
5781 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
5782 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
5783 (UNSPEC_COND_EOR): Delete.
5784 (optab): Remove associated mappings.
5785 (SVE_INT_BINARY): New code iterator.
5786 (sve_int_op): Remove int attribute and add "minus" to the code
5788 (SVE_COND_INT_OP): Delete.
5789 (SVE_COND_FP_OP): Rename to...
5790 (SVE_COND_FP_BINARY): ...this.
5792 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
5794 * optabs.c (can_reuse_operands_p): New function.
5795 (maybe_legitimize_operands): Try to reuse the results for
5798 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
5800 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
5801 Add {q} suffix to insn mnemonic.
5803 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
5805 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
5806 (msp430_warn_func_return): New.
5808 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
5810 * fold-const.c (tree_nonzero_bits): New function.
5811 * fold-const.h (tree_nonzero_bits): Likewise.
5812 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
5813 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
5815 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
5819 * varasm.c (assemble_alias): Check ifunc_resolver only on
5822 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
5825 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
5826 when memory input operand is handled.
5828 2018-05-24 Luis Machado <luis.machado@linaro.org>
5830 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
5832 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
5834 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
5836 * match.pd: Delay FMA folds until after vectorization.
5838 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
5841 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
5842 address check not strict.
5844 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
5846 * gimple-match.h (gimple_match_op): New class.
5847 (mprts_hook): Replace parameters with a gimple_match_op *.
5848 (maybe_build_generic_op): Likewise.
5849 (gimple_simplified_result_is_gimple_val): Replace parameters with
5850 a const gimple_match_op *.
5851 (gimple_simplify): Replace code_helper * and tree * parameters with
5852 a gimple_match_op * parameter.
5853 (gimple_resimplify1): Replace code_helper *, tree and tree *
5854 parameters with a gimple_match_op * parameter.
5855 (gimple_resimplify2): Likewise.
5856 (gimple_resimplify3): Likewise.
5857 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
5858 parameters with a gimple_match_op * parameter.
5859 * gimple-match-head.c (gimple_simplify): Change prototypes of
5860 auto-generated functions to take a gimple_match_op * instead of
5861 separate code_helper * and tree * parameters. Make the same
5862 change in the top-level overload and update calls to the
5863 gimple_resimplify routines. Update calls to the auto-generated
5864 functions and to maybe_push_res_to_seq in the publicly-facing
5865 operation-specific gimple_simplify overloads.
5866 (gimple_match_op::MAX_NUM_OPS): Define.
5867 (gimple_resimplify1): Replace rcode and ops with a single res_op
5868 parameter. Update call to gimple_simplify.
5869 (gimple_resimplify2): Likewise.
5870 (gimple_resimplify3): Likewise.
5871 (mprts_hook): Replace parameters with a gimple_match_op *.
5872 (maybe_build_generic_op): Likewise.
5873 (build_call_internal): Replace type, nargs and ops with
5874 a gimple_match_op *.
5875 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
5876 with a single gimple_match_op *. Update calls to mprts_hook,
5877 build_call_internal and gimple_simplified_result_is_gimple_val.
5878 Factor out code that is common to the tree_code and combined_fn cases.
5879 * genmatch.c (expr::gen_transform): Replace tem_code and
5880 tem_ops with a gimple_match_op called tem_op. Update calls
5881 to the gimple_resimplify functions and maybe_push_res_to_seq.
5882 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
5883 res_ops. Update call to the gimple_resimplify functions.
5884 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
5885 (decision_tree::gen): Make the functions take a gimple_match_op *
5886 called res_op instead of separate res_code and res_ops parameters.
5887 Update call accordingly.
5888 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
5889 and ops with a single res_op parameter. Update calls to
5890 maybe_build_generic_op and maybe_push_res_to_seq.
5891 (fold_stmt_1): Update calls to gimple_simplify and
5892 replace_stmt_with_simplification.
5893 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
5894 and gimple_simplified_result_is_gimple_val.
5895 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
5897 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
5898 with a gimple_match_op *.
5899 (vn_nary_build_or_lookup): Likewise. Update call to
5900 vn_nary_build_or_lookup_1.
5901 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
5902 gimple_match_op *. Update calls to the gimple_resimplify routines
5903 and to gimple_simplified_result_is_gimple_val.
5904 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
5905 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
5906 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
5907 (visit_nary_op): Likewise.
5908 (visit_reference_op_load): Likewise.
5910 2018-05-23 Luis Machado <luis.machado@linaro.org>
5912 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
5913 modifier for printing the step amount.
5915 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5918 * gcc/tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
5921 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
5923 * doc/sourcebuild.texi (Endianness): New subsubsection.
5925 2018-05-23 Luis Machado <luis.machado@linaro.org>
5927 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
5928 <prefetch_dynamic_strides>: New const bool field.
5929 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
5930 prefetch_dynamic_strides.
5931 (exynosm1_prefetch_tune): Likewise.
5932 (thunderxt88_prefetch_tune): Likewise.
5933 (thunderx_prefetch_tune): Likewise.
5934 (thunderx2t99_prefetch_tune): Likewise.
5935 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
5937 (aarch64_override_options_internal): Update to set
5938 PARAM_PREFETCH_DYNAMIC_STRIDES.
5939 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
5940 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
5941 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
5942 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
5943 prefetch-dynamic-strides setting.
5945 2018-05-23 Luis Machado <luis.machado@linaro.org>
5947 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
5948 <minimum_stride>: New const int field.
5949 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
5950 minimum_stride field defaulting to -1.
5951 (exynosm1_prefetch_tune): Likewise.
5952 (thunderxt88_prefetch_tune): Likewise.
5953 (thunderx_prefetch_tune): Likewise.
5954 (thunderx2t99_prefetch_tune): Likewise.
5955 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
5956 <default_opt_level>: Set to 3.
5957 (aarch64_override_options_internal): Update to set
5958 PARAM_PREFETCH_MINIMUM_STRIDE.
5959 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
5960 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
5961 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
5962 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
5963 stride is constant and is below the minimum stride threshold.
5965 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5967 * config/arm/arm-cpus.in (mode26): Delete.
5968 (armv4): Delete mode26 reference.
5969 * config/arm/arm.c (arm_configure_build_target): Delete use of
5972 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
5974 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
5976 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
5977 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
5979 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
5980 (floatunsdidf2): Ditto.
5982 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
5984 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
5985 (fixuns_trunc<mode>si2_avx512f): Ditto.
5986 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
5987 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
5988 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
5990 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
5992 PR rtl-optimization/79985
5993 * df-scan.c (df_insn_refs_collect): Remove special case for
5994 global registers and asm statements.
5996 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
5998 * extend.texi (Global Register Variables): Rewrite the bullet list.
5999 Note that the register is available for allocation. Note that access
6000 via inline asm must use constraints. Add note about async-signal
6001 handlers. Remove paragraph about automagic register selection.
6003 2018-05-23 Richard Biener <rguenther@suse.de>
6005 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
6006 of fixed offset from memset VN.
6008 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
6010 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
6012 (alloc_cand_and_find_basis): Initialize first_interp field.
6013 (slsr_process_mul): Modify first_interp field.
6014 (slsr_process_add): Likewise.
6015 (slsr_process_cast): Modify first_interp field for each new
6017 (slsr_process_copy): Likewise.
6018 (dump_candidate): Dump first_interp field.
6019 (replace_mult_candidate): Process all interpretations, not just
6021 (replace_rhs_if_not_dup): Likewise.
6022 (replace_one_candidate): Likewise.
6024 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
6026 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
6028 (aarch64_needs_frame_chain): New function.
6029 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
6031 2018-05-23 Sudakshina Das <sudi.das@arm.com>
6034 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
6035 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
6036 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
6037 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
6038 as true for strict-align.
6039 (aarch64_can_inline_p): Perform checks even when callee has no
6040 attributes to check for strict alignment.
6041 * doc/extend.texi (AArch64 Function Attributes): Document
6043 * doc/invoke.texi: (AArch64 Options): Likewise.
6045 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
6047 PR tree-optimization/85853
6048 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
6049 the handling of the root of the node to...
6050 (vect_slp_analyze_node_operations_1): ...this new function,
6051 and run the whole thing with the child nodes' def types
6052 set according to their SLP node's def type.
6054 2018-05-23 Richard Biener <rguenther@suse.de>
6057 * tree-data-ref.c (create_runtime_alias_checks): Defer
6058 and ignore overflow warnings.
6060 2018-05-23 Yury Gribov <tetra2005@gmail.com>
6062 PR tree-optimization/85822
6063 * tree-vrp.c (is_masked_range_test): Fix handling of negative
6066 2018-05-23 Richard Biener <rguenther@suse.de>
6068 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
6069 memset constants via native_interpret_expr.
6071 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
6074 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
6076 (cgraph_node::create_alias): Likewise.
6077 (cgraph_node::get_availability): Check ifunc_resolver instead
6078 of looking up ifunc attribute.
6079 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
6080 * varasm.c (do_assemble_alias): Likewise.
6081 (assemble_alias): Likewise.
6082 (default_binds_local_p_3): Likewise.
6083 * cgraph.h (cgraph_node): Add ifunc_resolver.
6084 (cgraph_node::only_called_directly_or_aliased_p): Return false
6086 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
6088 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
6089 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
6090 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
6091 instead of looking up ifunc attribute.
6093 2018-05-22 Luis Machado <luis.machado@linaro.org>
6095 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
6097 2018-05-22 Martin Sebor <msebor@redhat.com>
6100 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
6101 only when expasion succeeds.
6102 (expand_builtin_strcmp): Same.
6103 (expand_builtin_strncmp): Same.
6105 2018-05-22 Martin Sebor <msebor@redhat.com>
6107 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
6109 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
6110 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6112 * config/aarch64/aarch64-ldpstp.md: Replace uses of
6113 aarch64_mem_pair_operand with memory_operand and delete operand swapping
6115 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
6116 Add check for legitimate_address.
6117 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
6118 (aarch64_swap_ldrstr_operands): New.
6119 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
6122 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
6123 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6125 * config/aarch64/aarch64.md: New patterns to generate stp
6127 (store_pair_sw, store_pair_dw): New patterns to generate stp for
6128 single words and double words.
6129 (load_pair_sw, load_pair_dw): Likewise.
6130 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
6132 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
6134 * config/aarch64/aarch64-ldpstp.md: Modify peephole
6135 for different mode ldpstp and add peephole for merged zero stores.
6137 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
6139 (aarch64_gen_store_pair): Rename calls to match new patterns.
6140 (aarch64_gen_load_pair): Rename calls to match new patterns.
6141 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
6142 (load_pair<DREG:mode><DREG2:mode>): ... This.
6143 (store_pair<mode>): Rename to...
6144 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
6145 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
6147 (V_INT_EQUIV): Handle SImode.
6148 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
6151 2018-05-22 Martin Sebor <msebor@redhat.com>
6154 * calls.c (maybe_warn_nonstring_arg): Use string length to set
6155 or ajust the presumed bound on an operation to avoid unnecessary
6158 2018-05-22 Martin Sebor <msebor@redhat.com>
6160 PR tree-optimization/85826
6161 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
6162 assuming that a DECL necesarily has a constant size.
6164 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
6167 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
6169 2018-05-22 Richard Biener <rguenther@suse.de>
6171 PR tree-optimization/85834
6172 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
6173 non-constant and non-zero memset arguments.
6175 2018-05-22 Martin Liska <mliska@suse.cz>
6178 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
6180 2018-05-22 Richard Biener <rguenther@suse.de>
6182 PR tree-optimization/85863
6183 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
6184 comparisons when vectype is specified.
6185 (vectorizable_condition): Do not specify vectype for
6186 vect_is_simple_cond when SLP vectorizing.
6188 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
6191 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
6192 define __ibm128 as long double.
6193 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
6194 as a distinct type when IEEE 128-bit support is enabled.
6195 (init_float128_ieee): Fix up conversions between IFmode and IEEE
6196 128-bit types to use the correct functions.
6197 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
6198 convert between 128-bit floating point types that have different
6199 modes but the same representation, instead of using gen_lowpart to
6201 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
6203 (IFKF_reg): New attributes to give the register constraints for
6205 (extend<mode>tf2_internal): New insns to mark an explicit
6206 conversion between 128-bit floating point types that have a
6207 different mode but share the same representation.
6209 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
6211 PR tree-optimization/85814
6212 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
6213 a null return from get_strinfo when unsharing the next
6214 strinfo in the chain.
6216 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
6219 * varasm.c (weak_finish): Clean up weak_decls.
6221 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6223 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
6224 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
6225 UNSPEC_UADALP values.
6226 * config/aarch64/iterators.md (ABAL): New int iterator.
6229 (sur): Add mappings for the above.
6230 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
6232 (aarch64_<sur>abal<mode>_4): Likewise.
6233 (aarch64_<sur>adalp<mode>_3): Likewise.
6234 (<sur>sadv16qi): New define_expand.
6236 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
6238 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
6239 (*movdf_internal): Ditto.
6240 (*rcpsf2_sse): Ditto.
6241 (*rsqrtsf2_sse): Ditto.
6242 (*sqrt<mode>2_sse): Ditto.
6244 2018-05-21 Tamar Christina <tamar.christina@arm.com>
6246 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
6248 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
6249 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
6250 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
6251 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
6253 * config/aarch64/arm_neon.h: Likewise.
6254 * config/aarch64/iterators.md (VQ_I): New.
6256 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
6258 * config.gcc: Add arc/t-multilib-linux to tmake_file for
6260 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
6263 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
6265 * config/nds32/constraints.md (S): New constraint.
6266 * config/nds32/nds32.md (call_internal): Use constraint S.
6267 (call_value_internal): Likewise.
6268 (sibcall_internal): Likewise.
6269 (sibcall_value_internal): Likewise.
6271 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
6272 Chung-Ju Wu <jasonwucj@gmail.com>
6274 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
6277 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
6278 Chung-Ju Wu <jasonwucj@gmail.com>
6280 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
6281 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
6282 (nds32_rtx_costs_impl): Simplify.
6283 (nds32_address_cost_impl): Simplify.
6284 (nds32_init_rtx_costs): New function.
6285 (nds32_rtx_costs_speed_prefer): Likewise.
6286 (nds32_rtx_costs_size_prefer): Likewise.
6287 (nds32_address_cost_speed_prefer): Likewise.
6288 (nds32_address_cost_speed_fwprop): Likewise.
6289 (nds32_address_cost_size_prefer): Likewise.
6290 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
6291 * config/nds32/nds32.c (nds32_option_override): Use
6292 nds32_init_rtx_costs function.
6294 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
6296 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
6297 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
6298 (TARGET_PIPELINE_N8): Likewise.
6299 (TARGET_PIPELINE_N10): Likewise.
6300 (TARGET_PIPELINE_N13): Likewise.
6301 (TARGET_PIPELINE_GRAYWOLF): Likewise.
6303 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
6305 * config/nds32/nds32-fpu.md: Update copyright year.
6307 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
6309 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
6311 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
6313 * config/nds32/nds32.c
6314 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
6315 * config/nds32/nds32.opt (minline-asm-r15): New option.
6317 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
6319 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
6321 * config/nds32/nds32.md (abssi2): New pattern.
6323 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
6325 * config/i386/i386.md (rex64namesuffix): New mode attribute.
6326 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
6327 Merge insn pattern from sse_cvtsi2ss<round_name> and
6328 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
6329 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
6330 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
6331 using SWI48 mode iterator.
6332 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
6333 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
6334 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
6335 pattern from sse_cvttss2si<round_saeonly_name>
6336 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
6337 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
6338 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
6339 using SWI48 mode iterator.
6340 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
6341 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
6342 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
6343 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
6344 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
6345 using SWI48 mode iterator.
6346 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
6347 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
6348 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
6349 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
6350 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
6351 SWI48 mode iterator.
6352 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
6353 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
6354 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
6355 pattern from sse_cvttsd2si<round_saeonly_name>
6356 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
6358 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
6360 * config/nds32/nds32-md-auxiliary.c
6361 (nds32_valid_smw_lwm_base_p): Refine.
6362 (nds32_output_smw_single_word): Refine.
6363 (nds32_output_smw_double_word): New.
6364 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
6366 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
6368 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
6369 (nds32_output_stack_pop): Refine.
6370 (nds32_expand_unaligned_load): Refine.
6371 (nds32_expand_unaligned_store): Refine.
6373 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
6374 Chung-Ju Wu <jasonwucj@gmail.com>
6376 * config/nds32/constants.md: Add TP_REGNUM constant.
6377 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
6378 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
6380 * config/nds32/nds32-doubleword.md: Consider flag_pic.
6381 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
6382 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
6383 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
6384 and PIC code generation.
6385 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
6387 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
6389 * config/nds32/nds32.md: Support TLS and PIC.
6390 * config/nds32/nds32.c: Support TLS and PIC.
6391 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
6392 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
6395 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
6397 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
6398 mode with E_ prefix.
6400 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
6401 Chung-Ju Wu <jasonwucj@gmail.com>
6403 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
6404 * config/nds32/nds32-md-auxiliary.c
6405 (symbolic_reference_mentioned_p): New.
6406 (nds32_legitimize_ict_address): New.
6407 (nds32_expand_ict_move): New.
6408 (nds32_indirect_call_referenced_p): New.
6409 (nds32_symbol_binds_local_p): Delete.
6410 (nds32_long_call_p): Modify.
6411 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
6412 * config/nds32/nds32-protos.h
6413 (symbolic_reference_mentioned_p): Declare.
6414 (nds32_legitimize_ict_address): Declare.
6415 (nds32_expand_ict_move): Declare.
6416 (nds32_indirect_call_referenced_p): Declare.
6417 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
6418 (nds32_relax_group): Use nds32_ict_const_p as condition.
6419 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
6420 (nds32_asm_file_start): Output ict_model directive in asm code.
6421 (nds32_legitimate_address_p): Consider indirect call.
6422 (nds32_print_operand): Consider indirect call.
6423 (nds32_print_operand_address): Consider indirect call.
6424 (nds32_insert_attributes): Handle "indirect_call" attribute.
6425 (TARGET_LEGITIMATE_ADDRESS_P): Define.
6426 (TARGET_LEGITIMATE_CONSTANT_P): Define.
6427 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
6428 (TARGET_DELEGITIMIZE_ADDRESS): Define.
6429 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
6430 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
6431 (TARGET_ICT_MODEL_SMALL): Define.
6432 (TARGET_ICT_MODEL_LARGE): Define.
6433 * config/nds32/nds32.md (movsi): Consider ict model.
6434 (call, call_value): Consider ict model.
6435 (sibcall, sibcall_value): Consider ict model.
6436 * config/nds32/nds32.opt (mict-model): New option.
6437 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
6440 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
6441 Monk Chiang <sh.chiang04@gmail.com>
6442 Jim Wilson <jimw@sifive.com>
6444 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
6445 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
6446 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
6447 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
6448 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
6449 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
6450 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
6451 compute save_libcall_adjustment properly.
6452 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
6453 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
6454 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
6455 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
6456 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
6457 (ABI_SPEC): Handle mabi=ilp32e.
6458 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
6459 (RVE): Add RVE mask.
6460 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
6461 <-march>: Add rv32e as an example.
6463 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
6466 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
6467 (intra_create_variable_infos): Handle C++ constructors.
6469 2018-05-18 Martin Liska <mliska@suse.cz>
6471 * passes.def: Remove a redundant pass.
6473 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
6476 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
6478 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6480 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
6482 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
6484 (armv2, armv2a, armv3, armv3m): Delete architectures.
6485 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
6486 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
6487 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
6489 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
6490 (*mulsidi3adddi): Likewise.
6491 (mulsidi3): Likewise.
6492 (*mulsidi3_nov6): Likewise.
6493 (umulsidi3): Likewise.
6494 (umulsidi3_nov6): Likewise.
6495 (umaddsidi4): Likewise.
6496 (*umulsidi3adddi): Likewise.
6497 (smulsi3_highpart): Likewise.
6498 (*smulsi3_highpart_nov6): Likewise.
6499 (umulsi3_highpart): Likewise.
6500 (*umulsi3_highpart_nov6): Likewise.
6501 * config/arm/arm.h (arm_arch3m): Delete.
6502 * config/arm/arm.c (arm_arch3m): Delete.
6503 (arm_option_override_internal): Update armv3-related comment.
6504 (arm_configure_build_target): Delete use of isa_bit_mode32.
6505 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
6506 (arm_rtx_costs_internal): Delete check of arm_arch3m.
6507 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
6509 (mulusa3): Likewise.
6510 * config/arm/arm-protos.h (arm_arch3m): Delete.
6511 * config/arm/arm-tables.opt: Regenerate.
6512 * config/arm/arm-tune.md: Likewise.
6513 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
6514 deleted architectures.
6516 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6518 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
6519 (armv5t, armv5te): New features.
6520 (ARMv5, ARMv5e): Delete fgroups.
6521 (ARMv5t, ARMv5te): Adjust for above changes.
6523 (armv5, armv5e): Delete arches.
6524 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
6526 (*call_reg_arm): Likewise.
6527 (*call_value_reg_armv5): Likewise.
6528 (*call_value_reg_arm): Likewise.
6529 (*call_symbol): Likewise.
6530 (*call_value_symbol): Likewise.
6531 (*sibcall_insn): Likewise.
6532 (*sibcall_value_insn): Likewise.
6534 (prefetch): Likewise.
6535 (define_split and define_peephole2 dependent on arm_arch5):
6537 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
6539 (TARGET_ARM_QBIT): Likewise.
6540 (TARGET_DSP_MULTIPLY): Likewise.
6541 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
6542 (arm_arch5, arm_arch5e): Delete.
6543 (arm_arch5t, arm_arch5te): Declare.
6544 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
6545 (arm_arch5t): Declare.
6546 (arm_option_reconfigure_globals): Update for the above.
6547 (arm_options_perform_arch_sanity_checks): Update comment, replace
6548 use of arm_arch5 with arm_arch5t.
6549 (use_return_insn): Likewise.
6550 (arm_emit_call_insn): Likewise.
6551 (output_return_instruction): Likewise.
6552 (arm_final_prescan_insn): Likewise.
6553 (arm_coproc_builtin_available): Likewise.
6554 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
6555 arm_arch5e with arm_arch5t and arm_arch5te.
6556 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
6557 (arm_arch5t, arm_arch5te): Declare.
6558 * config/arm/arm-tables.opt: Regenerate.
6559 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
6560 * config/arm/t-multilib: Likewise.
6561 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
6562 instead of arm_arch5.
6563 (*call_reg_thumb1): Likewise.
6564 (*call_value_reg_thumb1_v5): Likewise.
6565 (*call_value_reg_thumb1): Likewise.
6566 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
6568 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
6570 2018-05-18 Martin Liska <mliska@suse.cz>
6572 PR gcov-profile/84846
6573 * doc/gcov.texi: Document -t option of gcov tool.
6575 2018-05-18 Martin Liska <mliska@suse.cz>
6577 PR gcov-profile/84846
6578 * gcov.c (print_usage): Add new -t option.
6579 (process_args): Handle the option.
6580 (generate_results): Use stdout as output when requested by
6583 2018-05-18 Martin Liska <mliska@suse.cz>
6585 PR gcov-profile/84846
6586 * coverage.c (coverage_init): Write PWD to .gcno file.
6587 * doc/gcov.texi: Document how working directory is printed.
6588 * gcov-dump.c (dump_gcov_file): Print PWD.
6589 * gcov.c (output_intermediate_file): Likewise.
6590 (read_graph_file): Read PWD string.
6591 (output_lines): Print PWD.
6593 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6596 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
6597 for retval and return false if all args to phi are zero.
6599 2018-05-18 Richard Biener <rguenther@suse.de>
6601 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
6603 (evrp_dom_walker::before_dom_children): Call it.
6605 2018-05-18 Richard Biener <rguenther@suse.de>
6607 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
6608 results when processing array refs with variable index.
6610 2018-05-18 Toon Moene <toon@moene.org>
6612 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
6613 directly after that of -floop-interchange. Indicate that both
6614 options are enabled by default when specifying -O3.
6616 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6618 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
6619 iterator. Delete separate integer-mode vec_set<mode> expander.
6620 (aarch64_simd_vec_setv2di): Delete.
6621 (vec_setv2di): Delete.
6622 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
6623 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
6624 the "w, r" alternative.
6626 2018-05-18 Martin Liska <mliska@suse.cz>
6628 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
6629 * tree-pass.h (make_pass_lower_switch_O0): New function.
6630 * tree-switch-conversion.c (node_has_low_bound): Remove.
6631 (node_has_high_bound): Likewise.
6632 (node_is_bounded): Likewise.
6633 (class pass_lower_switch): Make it a template type and create
6635 (pass_lower_switch::execute): Add template argument.
6636 (make_pass_lower_switch): New function.
6637 (make_pass_lower_switch_O0): New function.
6638 (do_jump_if_equal): Remove.
6639 (emit_case_nodes): Simplify to just handle all 3 cases and leave
6640 all the hard work to tree optimization passes.
6642 2018-05-18 Martin Liska <mliska@suse.cz>
6644 * dbgcnt.c (limit_low): Renamed from limit.
6645 (limit_high): New variable.
6646 (dbg_cnt_is_enabled): Check for upper limit.
6647 (dbg_cnt): Adjust dumping.
6648 (dbg_cnt_set_limit_by_index): Add new argument for high
6650 (dbg_cnt_set_limit_by_name): Likewise.
6651 (dbg_cnt_process_single_pair): Parse new format.
6652 (dbg_cnt_process_opt): Use strtok.
6653 (dbg_cnt_list_all_counters): Remove 'value' and add
6655 * doc/invoke.texi: Document changes.
6657 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
6659 * doc/sourcebuild.texi (scalar_all_fma): Document.
6660 * tree.def (FMA_EXPR): Delete.
6661 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
6662 * internal-fn.c (ternary_direct): New macro.
6663 (expand_ternary_optab_fn): Likewise.
6664 (direct_ternary_optab_supported_p): Likewise.
6665 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
6666 * builtins.c (fold_builtin_fma): Delete.
6667 (fold_builtin_3): Don't call it.
6668 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
6669 * expr.c (expand_expr_real_2): Likewise.
6670 * fold-const.c (operand_equal_p): Likewise.
6671 (fold_ternary_loc): Likewise.
6672 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
6673 * gimple.c (DEFTREECODE): Likewise.
6674 * gimplify.c (gimplify_expr): Likewise.
6675 * optabs-tree.c (optab_for_tree_code): Likewise.
6676 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6677 * tree-eh.c (operation_could_trap_p): Likewise.
6678 (stmt_could_throw_1_p): Likewise.
6679 * tree-inline.c (estimate_operator_cost): Likewise.
6680 * tree-pretty-print.c (dump_generic_node): Likewise.
6681 (op_code_prio): Likewise.
6682 * tree-ssa-loop-im.c (stmt_cost): Likewise.
6683 * tree-ssa-operands.c (get_expr_operands): Likewise.
6684 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
6685 * fold-const-call.h (fold_fma): Delete.
6686 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
6687 CFN_FNMA and CFN_FNMS.
6689 * genmatch.c (combined_fn): New enum.
6690 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
6691 (commutative_op): New function.
6692 (commutate): Use it. Handle more than 2 operands.
6693 (dt_operand::gen_gimple_expr): Use commutative_op.
6694 (parser::parse_expr): Allow :c to be used with non-binary
6695 operators if the commutative operand is known.
6696 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
6697 CFN_FMS, CFN_FNMA and CFN_FNMS.
6698 (backprop::process_assign_use): Remove FMA_EXPR handling.
6699 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
6700 (gen_hsa_fma): New function.
6701 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
6702 IFN_FNMA and IFN_FNMS.
6703 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
6704 * gimple-fold.h (follow_all_ssa_edges): Declare.
6705 * gimple-fold.c (follow_all_ssa_edges): New function.
6706 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
6707 gimple_build interface and use follow_all_ssa_edges to fold the result.
6708 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
6709 instead of checking for optabs directly.
6710 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
6711 rather than FMA_EXPRs.
6712 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
6713 call to IFN_FMA instead of an FMA_EXPR.
6715 2018-05-17 Jim Wilson <jimw@sifive.com>
6717 * expr.c (do_tablejump): When converting index to Pmode, if we have a
6718 sign extended promoted subreg, and the range does not have the sign bit
6719 set, then do a sign extend.
6721 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
6722 test, check for sign extended subreg and/or constant operands, and
6723 do a sign extend in that case.
6725 2018-05-17 Steve Ellcey <sellcey@cavium.com>
6727 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
6728 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
6730 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
6731 Change logics_shift_reg to logics_shift_imm.
6732 (thunderx2t99_fp_loadpair_basic): Delete.
6733 (thunderx2t99_fp_storepair_basic): Delete.
6734 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
6735 (thunderx2t99_asimd_polynomial): Delete.
6736 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
6737 and neon_fp_mul_d_scalar_q.
6738 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
6739 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
6740 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
6741 (thunderx2t99_asimd_lut): Add missing tbl types.
6742 (thunderx2t99_asimd_ext): Delete.
6743 (thunderx2t99_asimd_load1_1_mult): Delete.
6744 (thunderx2t99_asimd_load1_2_mult): Delete.
6745 (thunderx2t99_asimd_load1_ldp): New.
6746 (thunderx2t99_asimd_load1): New.
6747 (thunderx2t99_asimd_load2): Add missing *load2* types.
6748 (thunderx2t99_asimd_load3): New.
6749 (thunderx2t99_asimd_load4): New.
6750 (thunderx2t99_asimd_store1_1_mult): Delete.
6751 (thunderx2t99_asimd_store1_2_mult): Delete.
6752 (thunderx2t99_asimd_store2_mult): Delete.
6753 (thunderx2t99_asimd_store2_onelane): Delete.
6754 (thunderx2t99_asimd_store_stp): New.
6755 (thunderx2t99_asimd_store1): New.
6756 (thunderx2t99_asimd_store2): New.
6757 (thunderx2t99_asimd_store3): New.
6758 (thunderx2t99_asimd_store4): New.
6760 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
6762 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
6763 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
6765 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
6766 Segher Boessenkool <segher@kernel.crashing.org>
6769 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
6772 2018-05-17 Richard Biener <rguenther@suse.de>
6774 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
6775 for pruning loop and prune defs feeding only already visited PHIs.
6777 2018-05-17 Richard Biener <rguenther@suse.de>
6779 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
6781 2018-05-17 Bin Cheng <bin.cheng@arm.com>
6782 Richard Biener <rguenther@suse.de>
6784 PR tree-optimization/85793
6785 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
6786 for VMAT_ELEMENTWISE.
6788 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
6790 * internal-fn.h (lookup_internal_fn): Declare
6791 * internal-fn.c (lookup_internal_fn): New function.
6792 * gimple.c (gimple_build_call_from_tree): Handle calls to
6794 * gimple-pretty-print.c (dump_gimple_call): Print "." before
6795 internal function names.
6796 * tree-pretty-print.c (dump_generic_node): Likewise.
6797 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
6799 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
6801 * gimple-fold.h (gimple_build): Make the function forms take
6802 combined_fn rather than built_in_function.
6803 (gimple_simplify): Likewise.
6804 * gimple-match-head.c (gimple_simplify): Likewise.
6805 * gimple-fold.c (gimple_build): Likewise.
6806 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
6807 rather than gimple_build_call_internal.
6808 (get_initial_defs_for_reduction): Likewise.
6809 (vect_create_epilog_for_reduction): Likewise.
6810 (vectorizable_live_operation): Likewise.
6812 2018-05-17 Martin Liska <mliska@suse.cz>
6814 * gimple-ssa-sprintf.c (format_directive): Do not use
6815 space in between 'G_' and '('.
6817 2018-05-17 Jakub Jelinek <jakub@redhat.com>
6820 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
6821 even if the mask is not all ones.
6824 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
6826 (ix86_gimple_fold_builtin): Likewise.
6829 * config/i386/i386.c: Include tree-vector-builder.h.
6830 (ix86_vector_shift_count): New function.
6831 (ix86_fold_builtin): Fold shift builtins by scalar count.
6832 (ix86_gimple_fold_builtin): Likewise.
6834 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
6835 _mm512_setzero): New intrinsics.
6837 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
6838 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6840 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
6841 code generation for cases where splatting a value is not useful.
6842 * simplify-rtx.c (simplify_ternary_operation): Simplify
6843 vec_merge across a vec_duplicate and a paradoxical subreg forming
6844 a vector mode to a vec_concat.
6846 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
6848 * config.gcc: Support "goldmont-plus".
6849 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
6851 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6852 PROCESSOR_GOLDMONT_PLUS.
6853 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
6854 (processor_target_table): Add "goldmont-plus".
6855 (PTA_GOLDMONT_PLUS): Define.
6856 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
6857 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
6858 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
6859 (fold_builtin_cpu): Add "goldmont-plus".
6860 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
6861 (ix86_option_override_internal): Add "goldmont-plus".
6862 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
6863 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
6864 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
6865 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
6867 2018-05-17 Richard Biener <rguenther@suse.de>
6869 PR tree-optimization/85757
6870 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
6871 remove defs that only feed that PHI from further processing.
6873 2018-05-16 Jim Wilson <jimw@sifive.com>
6875 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
6877 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
6879 2018-05-16 Mark Wielaard <mark@klomp.org>
6881 * dwarf2out.c (count_index_strings): New function.
6882 (output_indirect_strings): Call count_index_strings and generate
6883 header for dwarf_version >= 5.
6885 2018-05-16 Mark Wielaard <mark@klomp.org>
6887 * dwarf2out.c (dwarf_FORM): New function.
6888 (set_indirect_string): Use dwarf_FORM.
6889 (reset_indirect_string): Likewise.
6890 (size_of_die): Likewise.
6891 (value_format): Likewise.
6892 (output_die): Likewise.
6893 (add_skeleton_AT_string): Likewise.
6894 (output_macinfo_op): Likewise.
6895 (index_string): Likewise.
6896 (output_index_string_offset): Likewise.
6897 (output_index_string): Likewise.
6898 (count_index_strings): Likewise.
6900 2018-05-16 Carl Love <cel@us.ibm.com>
6902 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
6903 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
6905 2018-05-16 Martin Jambor <mjambor@suse.cz>
6907 * ipa-prop.c (ipa_free_all_edge_args): Remove.
6908 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
6910 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
6912 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
6913 (fnma<mode>4): Likewise.
6914 (fms<mode>4): Likewise.
6915 (fnms<mode>4): Likewise.
6916 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
6917 (aarch64_fnma<mode>4): Likewise.
6918 (aarch64_fms<mode>4): Likewise.
6919 (aarch64_fnms<mode>4): Likewise.
6920 (aarch64_fnmadd<mode>4): Likewise.
6922 2018-05-16 Jason Merrill <jason@redhat.com>
6924 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
6926 2018-05-16 Richard Biener <rguenther@suse.de>
6928 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
6929 (dump_stmt_cost): Declare.
6930 (add_stmt_cost): Dump cost we add.
6931 (add_stmt_costs): New function.
6932 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
6934 (vect_analyze_stmt): Adjust prototype.
6935 (vectorizable_condition): Likewise.
6936 (vectorizable_live_operation): Likewise.
6937 (vectorizable_reduction): Likewise.
6938 (vectorizable_induction): Likewise.
6939 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
6940 cost vector to pass to vectorizable_ and record afterwards.
6941 (vect_model_reduction_cost): Take cost vector argument and adjust.
6942 (vect_model_induction_cost): Likewise.
6943 (vectorizable_reduction): Likewise.
6944 (vectorizable_induction): Likewise.
6945 (vectorizable_live_operation): Likewise.
6946 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
6947 SLP_TREE_NUMBER_OF_VEC_STMTS.
6948 (vect_analyze_slp_cost_1): Remove.
6949 (vect_analyze_slp_cost): Likewise.
6950 (vect_slp_analyze_node_operations): Take visited args and
6951 a target cost vector. Avoid processing already visited stmt sets.
6952 (vect_slp_analyze_operations): Use a local cost vector to gather
6953 costs and register those of non-discarded instances.
6954 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
6955 (vect_schedule_slp_instance): Remove copying of
6956 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
6958 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
6959 adding cost. Record cost entry location.
6960 (vect_prologue_cost_for_slp_op): Function to compute cost of
6961 a constant or invariant generated for SLP vect in the prologue,
6962 split out from vect_analyze_slp_cost_1.
6963 (vect_model_simple_cost): Make static. Adjust for SLP costing.
6964 (vect_model_promotion_demotion_cost): Likewise.
6965 (vect_model_store_cost): Likewise, make static.
6966 (vect_model_load_cost): Likewise.
6967 (vectorizable_bswap): Add cost vector arg and adjust.
6968 (vectorizable_call): Likewise.
6969 (vectorizable_simd_clone_call): Likewise.
6970 (vectorizable_conversion): Likewise.
6971 (vectorizable_assignment): Likewise.
6972 (vectorizable_shift): Likewise.
6973 (vectorizable_operation): Likewise.
6974 (vectorizable_store): Likewise.
6975 (vectorizable_load): Likewise.
6976 (vectorizable_condition): Likewise.
6977 (vectorizable_comparison): Likewise.
6978 (can_vectorize_live_stmts): Likewise.
6979 (vect_analyze_stmt): Likewise.
6980 (vect_transform_stmt): Adjust calls to vectorizable_*.
6981 * tree-vectorizer.c: Include gimple-pretty-print.h.
6982 (dump_stmt_cost): New function.
6984 2018-05-16 Richard Biener <rguenther@suse.de>
6986 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
6987 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
6988 * tree-ssa-dse.c: Include tree-ssa-loop.h.
6989 (check_name): New callback.
6990 (dse_classify_store): Track cycles via a visited bitmap of PHI
6991 defs and simplify handling of in-loop and across loop dead stores
6992 and properly fail for loop-variant refs. Handle byte-tracking with
6993 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
6996 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
6998 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
6999 (vect_get_mask_type_for_stmt): Likewise.
7000 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
7002 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
7003 to determine the statement's vector type and the vector type that
7004 should be used for calculating nunits. Deal with cases in which
7005 the type has to be deferred.
7006 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
7007 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
7008 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
7009 (vect_determine_vf_for_stmt): New functions, split out from...
7010 (vect_determine_vectorization_factor): ...here.
7011 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
7012 (vect_get_mask_type_for_stmt): New functions, split out from
7013 vect_determine_vectorization_factor.
7015 2018-05-16 Richard Biener <rguenther@suse.de>
7017 * tree-cfg.c (verify_gimple_assign_ternary): Properly
7018 verify the [VEC_]COND_EXPR embedded comparison.
7020 2018-05-15 Martin Sebor <msebor@redhat.com>
7022 PR tree-optimization/85753
7023 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
7024 RECORD_TYPE in addition to ARRAY_TYPE.
7026 2018-05-15 Martin Sebor <msebor@redhat.com>
7029 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
7031 2018-05-15 Richard Biener <rguenther@suse.de>
7033 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
7034 add by_clobber_p one. Change algorithm to collect all defs
7035 representing uses we need to walk and try reducing them to
7036 a single one before failing.
7037 (dse_dom_walker::dse_optimize_stmt): Adjust.
7039 2018-05-13 Mark Wielaard <mark@klomp.org>
7041 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
7042 (size_of_loc_descr): Likewise.
7043 (output_loc_operands): Likewise.
7044 (output_loc_operands_raw): Likewise.
7045 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
7046 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
7047 (hash_loc_operands): Likewise.
7048 (compare_loc_operands): Likewise.
7050 2018-05-14 Mark Wielaard <mark@klomp.org>
7052 * dwarf2out.c (count_index_addrs): New function.
7053 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
7055 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7057 PR tree-optimization/83648
7058 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
7059 return value as malloc candidate.
7061 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7064 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
7065 param as true in call to suggest_attribute.
7067 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
7069 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
7070 -mreadonly-in-sdata.
7072 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7074 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
7076 (aarch64_crypto_aesd_fused): Likewise.
7078 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
7080 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
7081 (movsi_aarch64): Likewise.
7082 (load_pairsi): Likewise.
7083 (load_pairdi): Likewise.
7084 (store_pairsi): Likewise.
7085 (store_pairdi): Likewise.
7086 (load_pairsf): Likewise.
7087 (load_pairdf): Likewise.
7088 (store_pairsf): Likewise.
7089 (store_pairdf): Likewise.
7090 (zero_extend): Likewise.
7091 (trunc): Swap alternatives.
7092 (fcvt_target): Add '?' to prefer w over r.
7094 2018-05-14 Jakub Jelinek <jakub@redhat.com>
7097 * config/i386/i386.md: Disallow non-commutative arithmetics in
7098 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
7099 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
7100 in the peephole2 before it.
7102 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
7104 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
7105 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
7106 (ix86_handle_option): Handle -mcldemote.
7107 * config.gcc: New header.
7108 * config/i386/cldemoteintrin.h: New file.
7109 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
7110 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
7112 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7113 OPTION_MASK_ISA_CLDEMOTE.
7114 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
7115 (ix86_valid_target_attribute_inner_p): Ditto.
7116 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
7117 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
7118 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
7119 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
7120 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
7122 * config/i386/i386.opt: Add -mcldemote.
7123 * config/i386/x86intrin.h: New header.
7124 * doc/invoke.texi: Add -mcldemote.
7126 2018-05-14 Richard Biener <rguenther@suse.de>
7128 * doc/match-and-simplify.texi: Adjust :s documentation.
7130 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
7132 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
7133 intended memcpy size.
7134 (REORDER_45): Likewise.
7136 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
7138 * sort.cc: New file.
7139 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
7140 * vec.c (qsort_chk): Use gcc_qsort.
7141 * Makefile.in (OBJS-libcommon): Add sort.o.
7142 (build/sort.o): New target. Use it...
7143 (BUILD_RTL): ... here, and...
7144 (build/gencfn-macros): ... here, and...
7145 (build/genmatch): ... here.
7147 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
7148 Chung-Ju Wu <jasonwucj@gmail.com>
7150 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
7151 * config/nds32/nds32-graywolf.md: New file.
7152 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
7153 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
7155 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
7156 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
7157 * config/nds32/nds32.md (pipeline_model): Add graywolf.
7158 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
7159 * config/nds32/pipelines.md: Include n15 settings.
7161 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
7162 Chung-Ju Wu <jasonwucj@gmail.com>
7164 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
7165 * config/nds32/nds32-n13.md: New file.
7166 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
7167 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
7169 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
7170 * config/nds32/nds32.md (pipeline_model): Add n13.
7171 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
7172 * config/nds32/pipelines.md: Include n13 settings.
7174 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
7175 Chung-Ju Wu <jasonwucj@gmail.com>
7177 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
7178 * config/nds32/nds32-n10.md: New file.
7179 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
7180 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
7182 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
7183 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
7184 * config/nds32/nds32.md (pipeline_model): Add n10.
7185 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
7186 * config/nds32/pipelines.md: Include n10 settings.
7188 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
7189 Kito Cheng <kito.cheng@gmail.com>
7190 Chung-Ju Wu <jasonwucj@gmail.com>
7192 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
7193 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
7194 Add enum values for DSP extension instructions.
7195 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
7197 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
7198 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
7200 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
7201 * config/nds32/nds32-dspext.md: New file for DSP implementation.
7202 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
7203 * config/nds32/nds32-intrinsic.md: Likewise.
7204 * config/nds32/nds32_intrinsic.h: Likewise.
7205 * config/nds32/nds32-md-auxiliary.c: Likewise.
7206 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
7207 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
7208 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
7209 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
7210 * config/nds32/nds32-protos.h: New declarations for DSP extension.
7211 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
7212 TYPE_DMAC in switch statement.
7213 * config/nds32/nds32.c: New checking and implementation for DSP
7214 extension instructions.
7215 * config/nds32/nds32.h: Likewise.
7216 * config/nds32/nds32.md: Likewise.
7217 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
7218 * config/nds32/predicates.md: Implement new predicates for DSP
7221 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
7223 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
7224 Reformat alternatives and attributes so it is easier to identify
7225 which constraints/attributes go with which instruction.
7226 (mov<mode>_hardfloat32, FMOVE64): Likewise.
7227 (mov<mode>_softfloat32, FMOVE64): Likewise.
7228 (mov<mode>_hardfloat64, FMOVE64): Likewise.
7229 (mov<mode>_softfloat64, FMOVE64): Likewise.
7231 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
7233 * doc/extend.texi (PowerPC Built-in Functions): Rename this
7235 (Basic PowerPC Built-in Functions): The new name of the
7236 subsection previously known as "PowerPC Built-in Functions".
7237 (Basic PowerPC Built-in Functions Available on all Configurations):
7239 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
7240 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
7241 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
7242 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
7244 2018-05-11 Martin Jambor <mjambor@suse.cz>
7247 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
7250 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
7253 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
7255 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
7257 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
7258 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
7259 (ix86_handle_option): Handle -mwaitpkg.
7260 * config.gcc: New header.
7261 * config/i386/cpuid.h (bit_WAITPKG): New bit.
7262 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
7263 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
7265 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7266 OPTION_MASK_ISA_WAITPKG.
7267 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
7268 (ix86_option_override_internal): Add PTA_WAITPKG.
7269 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
7270 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
7271 IX86_BUILTIN_TPAUSE.
7272 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
7273 __builtin_ia32_umwait and __builtin_ia32_tpause.
7274 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
7275 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
7276 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
7277 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
7278 UNSPECV_TPAUSE): New.
7279 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
7280 * config/i386/i386.opt: Add -mwaitpkg.
7281 * config/i386/waitpkgintrin.h: New file.
7282 * config/i386/x86intrin.h: New header.
7283 * doc/invoke.texi: Add -mwaitpkg.
7285 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
7288 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
7290 (cortex-m0): Use armv6s-m isa.
7291 (cortex-m0plus): Likewise.
7292 (cortex-m1): Likewise.
7293 (cortex-m0.small-multiply): Likewise.
7294 (cortex-m0plus.small-multiply): Likewise.
7295 (cortex-m1.small-multiply): Likewise.
7297 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
7298 Jakub Jelinek <jakub@redhat.com>
7300 PR tree-optimization/85692
7301 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
7302 source permute as well.
7304 2018-05-11 Martin Liska <mliska@suse.cz>
7307 * doc/extend.texi: Document LLVM style format for no_sanitize
7310 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7312 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
7313 mode_supports_vsx_dform_quad to mode_supports_dq_form.
7314 (mode_supports_vsx_dform_quad): Likewise.
7315 (mode_supports_vmx_dform): Move these functions to be next to the
7316 other mode_supports functions.
7317 (mode_supports_dq_form): Likewise.
7318 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
7319 mode_supports_dq_form.
7320 (reg_offset_addressing_ok_p): Likewise.
7321 (offsettable_ok_by_alignment): Likewise.
7322 (rs6000_legitimate_offset_address_p): Likewise.
7323 (legitimate_lo_sum_address_p): Likewise.
7324 (rs6000_legitimize_address): Likewise.
7325 (rs6000_legitimize_reload_address): Likewise.
7326 (rs6000_secondary_reload_inner): Likewise.
7327 (rs6000_preferred_reload_class): Likewise.
7328 (rs6000_output_move_128bit): Likewise.
7330 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
7332 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
7333 Generate SImode target register for null target.
7334 <case IX86_BUILTIN_XGETBV>: Ditto.
7335 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
7336 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
7338 2018-05-10 Carl Love <cel@us.ibm.com>
7340 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
7341 dcbtt and dcbtstt if operands[2] is 0.
7343 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
7346 * config/i386/sse.md (usadv64qi): New expander.
7348 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
7350 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
7351 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
7352 -maltivec=be support.
7353 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
7354 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
7355 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
7356 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
7357 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
7358 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
7359 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
7360 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
7361 altivec_vsumsws): Adjust.
7362 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
7363 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
7364 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
7366 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
7367 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
7368 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
7369 (altivec_lve<VI_char>x): Delete expand.
7370 (*altivec_lve<VI_char>x_internal): Rename to...
7371 (altivec_lve<VI_char>x): ... this.
7372 (altivec_lvxl_<mode>): Delete expand.
7373 (*altivec_lvxl_<mode>_internal): Rename to ...
7374 (altivec_lvxl_<mode>): ... this.
7375 (altivec_stvxl_<mode>): Delete expand.
7376 (*altivec_stvxl_<mode>_internal): Rename to ...
7377 (altivec_stvxl_<mode>): ... this.
7378 (altivec_stve<VI_char>x): Delete expand.
7379 (*altivec_stve<VI_char>x_internal): Rename to ...
7380 (altivec_stve<VI_char>x): ... this.
7381 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
7382 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
7383 reduc_plus_scal_<mode>): Adjust.
7384 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
7386 (rs6000_cpu_cpp_builtins): Adjust.
7387 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
7388 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
7389 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
7390 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
7391 -maltivec=be support.
7392 (rs6000_split_vec_extract_var): Adjust.
7393 (rs6000_split_v4si_init): Adjust.
7394 (swap_selector_for_mode): Delete.
7395 (altivec_expand_lvx_be, altivec_expand_stvx_be,
7396 altivec_expand_stvex_be): Delete.
7397 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
7398 -maltivec=be support.
7399 (rs6000_gimple_fold_builtin): Ditto.
7400 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
7402 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
7403 (TARGET_DIRECT_MOVE_64BIT): Adjust.
7404 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
7405 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
7406 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
7407 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
7408 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
7409 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
7410 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
7411 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
7412 anonymous split): Adjust.
7413 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
7414 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
7416 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
7418 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
7419 when --with-gxx-include-dir is also specified.
7420 * configure: Regenerate.
7422 2018-05-09 Jim Wilson <jimw@sifive.com>
7425 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
7426 * config/riscv/t-withmultilib: New.
7427 * config/riscv/withmultilib.h: New.
7428 * doc/install.texi: Document RISC-V --with-multilib-list support.
7430 2018-05-09 Richard Biener <rguenther@suse.de>
7432 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
7434 (vect_bb_vectorization_profitable_p): Adjust. Compute
7435 actual scalar cost using the cost vector and the add_stmt_cost
7438 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
7440 PR rtl-optimization/85645
7441 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
7442 in the REG_CFA_REGISTER note for LR, don't leave it empty.
7444 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
7446 PR rtl-optimization/85645
7447 * shrink-wrap.c (spread_components): Return a boolean saying if
7448 anything was changed.
7449 (try_shrink_wrapping_separate): Iterate spread_components until
7450 nothing changes anymore.
7452 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
7454 PR rtl-optimization/85645
7455 * regrename.c (build_def_use): Also kill the chains that include the
7456 destination of a REG_CFA_REGISTER note.
7458 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
7460 PR rtl-optimization/85645
7461 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
7462 insn that has a REG_CFA_REGISTER note.
7464 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
7466 * cfgexpand.c (expand_clobber): New function.
7467 (expand_gimple_stmt_1): Use it.
7468 * tree-vect-stmts.c (vect_clobber_variable): New function,
7470 (vectorizable_simd_clone_call): ...here.
7471 (vectorizable_store): Emit a clobber either side of an
7472 IFN_STORE_LANES sequence.
7473 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
7475 2018-05-09 Tom de Vries <tom@codesourcery.com>
7478 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
7479 (define_insn "trap_if_false"): Add exit after trap.
7481 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
7483 PR rtl-optimization/85638
7484 * bb-reorder.c: Include common/common-target.h.
7485 (create_forwarder_block): New function extracted from...
7486 (fix_up_crossing_landing_pad): ...here. Rename into...
7487 (dw2_fix_up_crossing_landing_pad): ...this.
7488 (sjlj_fix_up_crossing_landing_pad): New function.
7489 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
7490 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
7491 from both partitions and exit the loop after one iteration.
7493 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
7496 * doc/extend.texi (PowerPC Built-in Functions): Rename this
7498 (Basic PowerPC Built-in Functions): The new name of the
7499 subsection previously known as "PowerPC Built-in Functions".
7500 (Basic PowerPC Built-in Functions Available on all Configurations):
7502 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
7504 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
7506 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
7508 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
7511 2018-05-08 Jim Wilson <jimw@sifive.com>
7513 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
7514 (LD_EMUL_SUFFIX): New.
7515 (LINK_SPEC): Use it.
7517 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
7519 * doc/extend.texi (PowerPC Built-in Functions): Rename this
7521 (Basic PowerPC Built-in Functions): The new name of the
7522 subsection previously known as "PowerPC Built-in Functions".
7523 (Basic PowerPC Built-in Functions Available on all Configurations):
7525 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
7527 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
7529 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
7531 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
7534 2018-05-08 Jakub Jelinek <jakub@redhat.com>
7537 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
7538 after cmpelim optimization.
7540 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
7542 * config.gcc: Support "goldmont".
7543 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
7544 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7546 * config/i386/i386.c (m_GOLDMONT): Define.
7547 (processor_target_table): Add "goldmont".
7548 (PTA_GOLDMONT): Define.
7549 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
7550 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
7551 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
7552 (fold_builtin_cpu): Add "goldmont".
7553 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
7554 (ix86_option_override_internal): Add "goldmont".
7555 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
7556 (processor_type): Add PROCESSOR_GOLDMONT.
7557 * config/i386/i386.md: Add CPU "glm".
7558 * config/i386/glm.md: New file.
7559 * config/i386/x86-tune.def: Add m_GOLDMONT.
7560 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
7562 2018-05-08 Jakub Jelinek <jakub@redhat.com>
7565 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
7567 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
7568 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
7569 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
7572 * config/i386/i386.c (ix86_fold_builtin): Handle
7573 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
7576 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
7577 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
7579 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
7582 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
7583 (check_arch): Likewise.
7584 (check_fpu): Return the result rather than printing it.
7585 (end arch): Fix operator precedence.
7586 (end cpu): Likewise.
7587 (END): Print the result from check_fpu.
7589 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
7590 Alan Hayward <alan.hayward@arm.com>
7591 David Sherwood <david.sherwood@arm.com>
7593 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
7594 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
7595 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
7596 (*fcmuo<mode>_and): New patterns.
7598 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
7600 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
7601 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
7602 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
7603 (cmp_op, sve_imm_con): New code attributes.
7604 (SVE_COND_INT_CMP, imm_con): Delete.
7605 (cmp_op): Remove above unspecs from int attribute.
7606 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
7608 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
7609 comparison-specific unspecs.
7610 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
7611 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
7612 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
7613 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
7614 (*vec_fcm<cmp_op><mode>): Rename to...
7615 (*fcm<cmp_op><mode>): ...this and adjust likewise.
7616 (*vec_fcmuo<mode>): Rename to...
7617 (*fcmuo<mode>): ...this and adjust likewise.
7618 (*pred_fcm<cmp_op><mode>): New pattern.
7619 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
7620 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
7622 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
7624 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
7625 (aarch64_emit_sve_predicated_cond): New function.
7626 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
7627 (aarch64_emit_unspec_cond_or): Replace with...
7628 (aarch64_emit_sve_or_conds): ...this new function. Use
7629 aarch64_emit_sve_ptrue_op for the individual comparisons and
7630 aarch64_emit_binop to OR them together.
7631 (aarch64_emit_inverted_unspec_cond): Replace with...
7632 (aarch64_emit_sve_inverted_cond): ...this new function. Use
7633 aarch64_emit_sve_ptrue_op for the comparison and
7634 aarch64_emit_unop to invert the result.
7635 (aarch64_expand_sve_vec_cmp_float): Update after the above
7636 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
7638 2018-05-07 Nathan Sidwell <nathan@acm.org>
7640 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
7641 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
7642 (Backwards Compatibility): Likewise.
7644 2018-05-07 Luis Machado <luis.machado@linaro.org>
7648 2018-05-07 Luis Machado <luis.machado@linaro.org>
7650 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
7651 <prefetch_dynamic_strides>: New const bool field.
7652 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
7653 prefetch_dynamic_strides.
7654 (exynosm1_prefetch_tune): Likewise.
7655 (thunderxt88_prefetch_tune): Likewise.
7656 (thunderx_prefetch_tune): Likewise.
7657 (thunderx2t99_prefetch_tune): Likewise.
7658 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
7660 (aarch64_override_options_internal): Update to set
7661 PARAM_PREFETCH_DYNAMIC_STRIDES.
7662 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
7663 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
7664 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
7665 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
7666 prefetch-dynamic-strides setting.
7668 2018-05-07 Luis Machado <luis.machado@linaro.org>
7670 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
7671 <minimum_stride>: New const int field.
7672 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
7673 minimum_stride field.
7674 (exynosm1_prefetch_tune): Likewise.
7675 (thunderxt88_prefetch_tune): Likewise.
7676 (thunderx_prefetch_tune): Likewise.
7677 (thunderx2t99_prefetch_tune): Likewise.
7678 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
7679 (aarch64_override_options_internal): Update to set
7680 PARAM_PREFETCH_MINIMUM_STRIDE.
7681 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
7682 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
7683 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
7684 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
7685 stride is constant and is below the minimum stride threshold.
7687 2018-05-07 Luis Machado <luis.machado@linaro.org>
7689 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
7692 2018-05-07 Luis Machado <luis.machado@linaro.org>
7694 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
7695 <prefetch_dynamic_strides>: New const bool field.
7696 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
7697 prefetch_dynamic_strides.
7698 (exynosm1_prefetch_tune): Likewise.
7699 (thunderxt88_prefetch_tune): Likewise.
7700 (thunderx_prefetch_tune): Likewise.
7701 (thunderx2t99_prefetch_tune): Likewise.
7702 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
7704 (aarch64_override_options_internal): Update to set
7705 PARAM_PREFETCH_DYNAMIC_STRIDES.
7706 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
7707 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
7708 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
7709 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
7710 prefetch-dynamic-strides setting.
7712 2018-05-07 Luis Machado <luis.machado@linaro.org>
7714 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
7715 <minimum_stride>: New const int field.
7716 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
7717 minimum_stride field.
7718 (exynosm1_prefetch_tune): Likewise.
7719 (thunderxt88_prefetch_tune): Likewise.
7720 (thunderx_prefetch_tune): Likewise.
7721 (thunderx2t99_prefetch_tune): Likewise.
7722 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
7723 (aarch64_override_options_internal): Update to set
7724 PARAM_PREFETCH_MINIMUM_STRIDE.
7725 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
7726 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
7727 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
7728 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
7729 stride is constant and is below the minimum stride threshold.
7731 2018-05-06 Jakub Jelinek <jakub@redhat.com>
7734 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
7735 the type is addressable. Don't force op into register if it has
7738 2018-05-05 Roland McGrath <mcgrathr@google.com>
7741 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
7742 any section for which we don't know a specific type it should have,
7743 regardless of name. Previously this was done only for the exact
7744 names ".init_array", ".fini_array", and ".preinit_array".
7745 (default_elf_asm_named_section): Add comment about
7746 relationship with default_section_type_flags and SECTION_NOTYPE.
7747 (get_section): Don't consider it a type conflict if one side has
7748 SECTION_NOTYPE and the other doesn't, as long as neither has the
7749 SECTION_BSS et al used in the default_section_type_flags logic.
7751 2018-05-05 Tom de Vries <tom@codesourcery.com>
7754 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
7755 (workaround_barsyncs): New function.
7756 (nvptx_reorg): Use workaround_barsyncs.
7757 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
7758 (define_expand "nvptx_membar_cta"): New define_expand.
7759 (define_insn "*nvptx_membar_cta"): New insn.
7761 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
7763 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
7764 To improve optimization opportunities.
7765 * builtin-types.def: The new needed builtin types for the above.
7767 2018-05-04 Richard Biener <rguenther@suse.de>
7769 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
7770 * gimple-ssa-store-merging.c
7771 (imm_store_chain_info::output_merged_store): Remove redundant create,
7772 release split_store vector contents on failure.
7773 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
7774 scalar stmt vector on cache hit.
7776 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
7778 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
7780 * config.gcc (powerpc-xilinx-eabi*): Remove.
7781 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
7783 (fusion_addis_mem_combo_load): Ditto.
7784 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
7786 (rs6000_cpu_cpp_builtins): Ditto.
7787 * config/rs6000/rs6000-linux.c
7788 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
7789 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
7790 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
7792 (rs6000_setup_reg_addr_masks): Ditto.
7793 (rs6000_init_hard_regno_mode_ok): Ditto.
7794 (rs6000_option_override_internal): Ditto.
7795 (legitimate_lo_sum_address_p): Ditto.
7796 (rs6000_legitimize_address): Ditto.
7797 (rs6000_legitimize_reload_address): Ditto.
7798 (rs6000_legitimate_address_p): Ditto.
7799 (abi_v4_pass_in_fpr): Ditto.
7800 (setup_incoming_varargs): Ditto.
7801 (rs6000_gimplify_va_arg): Ditto.
7802 (rs6000_split_multireg_move): Ditto.
7803 (rs6000_savres_strategy): Ditto.
7804 (rs6000_emit_prologue_components): Ditto.
7805 (rs6000_emit_epilogue_components): Ditto.
7806 (rs6000_emit_prologue): Ditto.
7807 (rs6000_emit_epilogue): Ditto.
7808 (rs6000_elf_file_end): Ditto.
7809 (rs6000_function_value): Ditto.
7810 (rs6000_libcall_value): Ditto.
7811 * config/rs6000/rs6000.h: Ditto.
7812 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
7813 (TARGET_MINMAX): ... this. New.
7814 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
7815 * config/rs6000/rs6000.md: Remove Xilinx FP support.
7816 (*movsi_internal1_single): Delete.
7817 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
7818 mfpu=, mxilinx-fpu): Delete.
7819 * config/rs6000/singlefp.h: Delete.
7820 * config/rs6000/sysv4.h: Remove Xilinx FP support.
7821 * config/rs6000/t-rs6000: Ditto.
7822 * config/rs6000/t-xilinx: Delete.
7823 * gcc/config/rs6000/titan.md: Adjust for fp_type removal.
7824 * gcc/config/rs6000/vsx.md: Remove Xilinx FP support.
7825 (VStype_simple): Delete.
7826 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
7827 * config/rs6000/xfpu.h: Delete.
7828 * config/rs6000/xfpu.md: Delete.
7829 * config/rs6000/xilinx.h: Delete.
7830 * config/rs6000/xilinx.opt: Delete.
7831 * gcc/doc/invoke.texi (RS/6000 and PowerPC Options): Remove
7832 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
7834 2018-05-04 Tom de Vries <tom@codesourcery.com>
7837 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
7840 2018-05-04 Richard Biener <rguenther@suse.de>
7843 * tree-complex.c (update_complex_assignment): We are always in SSA form.
7844 (expand_complex_div_wide): Likewise.
7845 (expand_complex_operations_1): Likewise.
7846 (expand_complex_libcall): Preserve EH info of the original stmt.
7847 (tree_lower_complex): Handle removed blocks.
7848 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
7849 on complex multiplication and division libcall builtins.
7851 2018-05-04 Richard Biener <rguenther@suse.de>
7854 * fold-const.c (negate_expr_p): Restrict negation of operand
7855 zero of a division to when we know that can happen without
7857 (fold_negate_expr_1): Likewise.
7859 2018-05-04 Jakub Jelinek <jakub@redhat.com>
7862 * real.h (real_nextafter): Declare.
7863 * real.c (real_nextafter): New function.
7864 * fold-const-call.c (fold_const_nextafter): New function.
7865 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
7866 CASE_CFN_NEXTTOWARD.
7867 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
7868 even when arg1_mode is different from arg0_mode.
7870 2018-05-03 Nathan Sidwell <nathan@acm.org>
7872 * doc/extend.texi (Deprecated Features): Remove
7874 (Backwards Compatibility): Likewise.
7875 * doc/invoke.texi (C++ Language Options): Likewise.
7876 (C++ Dialect Options): Likewise.
7878 2018-05-03 Jakub Jelinek <jakub@redhat.com>
7881 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
7882 _mm512_mask_mullox_epi64): New intrinsics.
7884 2018-05-03 Tom de Vries <tom@codesourcery.com>
7887 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
7888 dump files): Add offload-tree.
7890 2018-05-03 Richard Biener <rguenther@suse.de>
7892 PR tree-optimization/85615
7893 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
7894 to loops not nested in BBs loop father to avoid creating multi-entry
7897 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7899 PR tree-optimization/70291
7900 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
7901 arguments. Change return type to tree. Emit libcall as a new
7902 statement rather than replacing existing one when inplace_p is true.
7903 (expand_complex_multiplication_components): New function.
7904 (expand_complex_multiplication): Expand floating-point complex
7905 multiplication using the above.
7906 (expand_complex_division): Rename inner_type parameter to type.
7907 Update expand_complex_libcall call-site.
7908 (expand_complex_operations_1): Update expand_complex_multiplication
7909 and expand_complex_division call-sites.
7911 2018-05-02 Jakub Jelinek <jakub@redhat.com>
7914 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
7915 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
7916 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
7917 the highest significant bit of the shift count mask is clear. In
7918 check whether and[sq]i3 is needed verify that all significant bits
7919 of the shift count other than the highest are set.
7921 2018-05-02 Tom de Vries <tom@codesourcery.com>
7924 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
7925 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
7926 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
7927 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
7928 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
7929 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
7930 __builtin_goacc_parlevel_size.
7932 2018-05-02 Richard Biener <rguenther@suse.de>
7934 PR tree-optimization/85597
7935 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
7936 do not use split vect_get_vec_defs call but call vect_get_slp_defs
7939 2018-05-02 Tom de Vries <tom@codesourcery.com>
7942 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
7943 dump files): Add ltrans-tree.
7945 2018-05-02 Tom de Vries <tom@codesourcery.com>
7948 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
7949 dump files): Add wpa-ipa.
7951 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
7953 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
7954 powerpc*-*-linux*paired* target.
7955 * config/rs6000/750cl.h: Delete.
7956 * config/rs6000/paired.h: Delete.
7957 * config/rs6000/paired.md: Delete.
7958 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
7960 * config/rs6000/rs6000-builtin.def: Remove paired float support.
7961 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
7962 comment. Remove paired float support.
7963 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
7964 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
7966 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
7967 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
7969 * config/rs6000/rs6000.c: Remove paired float support.
7970 (paired_expand_vector_init, paired_expand_vector_move,
7971 paired_emit_vector_compare, paired_emit_vector_cond_expr,
7972 (paired_expand_lv_builtin, paired_expand_stv_builtin,
7973 paired_expand_builtin, paired_expand_predicate_builtin,
7974 paired_init_builtins): Delete.
7975 * config/rs6000/rs6000.h: Remove paired float support.
7976 * config/rs6000/rs6000.md: Remove paired float support.
7977 (move_from_CR_ov_bit): Delete.
7978 * config/rs6000/rs6000.opt (mpaired): Delete.
7979 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
7980 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
7982 2018-05-02 Richard Biener <rguenther@suse.de>
7985 * gimplify.c (gimplify_save_expr): When in SSA form allow
7986 SAVE_EXPRs to compute to SSA vars.
7988 2018-05-02 Jakub Jelinek <jakub@redhat.com>
7991 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
7992 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
7993 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
7994 clobber operands[2], instead use a new pseudo. Formatting fixes.
7996 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
7998 PR tree-optimization/85586
7999 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
8000 exit early for statements in the same group if the accesses are
8003 2018-05-02 Tom de Vries <tom@codesourcery.com>
8006 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
8009 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
8011 PR tree-optimization/85143
8012 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
8014 2018-05-01 Tom de Vries <tom@codesourcery.com>
8017 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
8018 not found" error message.
8020 2018-05-01 Tom de Vries <tom@codesourcery.com>
8023 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
8024 * vec.c (test_ordered_remove_if): New function.
8025 (vec_c_tests): Call test_ordered_remove_if.
8026 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
8027 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
8028 * tree-vect-patterns.c (vect_pattern_recog_1): Use
8029 VEC_ORDERED_REMOVE_IF.
8031 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8033 PR tree-optimization/82665
8034 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
8035 pointer subtraction where arguments come from a memchr call.
8037 2018-05-01 Jakub Jelinek <jakub@redhat.com>
8039 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
8040 --push-state --as-needed and --pop-state instead of --as-needed and
8041 --no-as-needed if ld supports it.
8042 * configure: Regenerated.
8045 * doc/install.texi2html: Replace _002d with - and _002a with * in
8046 generated html files using sed.
8048 2018-04-30 David Malcolm <dmalcolm@redhat.com>
8051 * gcc-rich-location.c (blank_line_before_p): New function.
8052 (use_new_line): New function.
8053 (gcc_rich_location::add_fixit_insert_formatted): New function.
8054 * gcc-rich-location.h
8055 (gcc_rich_location::add_fixit_insert_formatted): New function.
8057 2018-04-30 David Malcolm <dmalcolm@redhat.com>
8059 * selftest.c (assert_streq): Rename "expected" and "actual" to
8060 "val1" and "val2". Extend NULL-handling to cover both inputs
8061 symmetrically, while still requiring both to be non-NULL for a pass.
8062 * selftest.h (assert_streq): Rename "expected" and "actual" to
8064 (ASSERT_EQ): Likewise.
8065 (ASSERT_EQ_AT): Likewise.
8066 (ASSERT_KNOWN_EQ): Likewise.
8067 (ASSERT_KNOWN_EQ_AT): Likewise.
8068 (ASSERT_NE): Likewise.
8069 (ASSERT_MAYBE_NE): Likewise.
8070 (ASSERT_MAYBE_NE_AT): Likewise.
8071 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
8072 the assertion to pass.
8073 (ASSERT_STREQ_AT): Likewise.
8075 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
8077 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
8078 interaction with -pie.
8080 2018-04-30 David Malcolm <dmalcolm@redhat.com>
8082 * selftest.h: Fix alphabetization of per-source-file selftest
8085 2018-04-30 Jason Merrill <jason@redhat.com>
8087 PR c++/61982 - dead stores to destroyed objects.
8088 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
8091 2018-04-30 Jason Merrill <jason@redhat.com>
8093 * tree.c (build_clobber): New.
8094 * tree.h: Declare it.
8095 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
8097 2018-04-30 David Malcolm <dmalcolm@redhat.com>
8099 * diagnostic-show-locus.c (layout::layout): Update for
8100 location_get_source_line returning a char_span.
8101 (struct char_span): Move to input.h.
8102 (struct correction): Update for fields in char_span becoming
8104 (struct source_line): Update for location_get_source_line
8105 returning a char_span.
8106 (layout::print_line): Likewise.
8107 * edit-context.c (edited_file::print_content): Likewise.
8108 (edited_file::print_diff_hunk): Likewise.
8109 (edited_file::print_run_of_changed_lines): Likewise.
8110 (edited_file::get_num_lines): Likewise.
8111 (edited_line::edited_line): Likewise.
8112 * final.c (asm_show_source): Likewise.
8113 * input.c (location_get_source_line): Convert return type
8114 from const char * to char_span, losing the final "line_len"
8116 (dump_location_info): Update for the above.
8117 (get_substring_ranges_for_loc): Likewise. Use a char_span
8118 when handling the literal within the line.
8119 (test_reading_source_line): Update for location_get_source_line
8120 returning a char_span.
8121 * input.h (class char_span): Move here from
8122 diagnostic-show-locus.c, converting from a struct to a class.
8123 Make data members private.
8124 (char_span::operator bool): New.
8125 (char_span::length): New.
8126 (char_span::get_buffer): New.
8127 (char_span::operator[]): New.
8128 (char_span::subspan): Make const.
8129 (char_span::xstrdup): New.
8130 (location_get_source_line): Convert return type from const char *
8131 to char_span, losing the final "line_size" param.
8133 2018-04-30 Jan Hubicka <jh@suse.cz>
8135 * lto-wrapper.c (ltrans_priorities): New static var.
8136 (cmp_priority): New.
8137 (run_gcc): Read priorities and if doing parallel build order
8138 the Makefile by them.
8140 2018-04-30 David Malcolm <dmalcolm@redhat.com>
8142 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
8144 2018-04-30 Richard Biener <rguenther@suse.de>
8146 * tree-cfg.c (verify_address): Remove base argument, add
8147 flag whether to check TREE_ADDRESSABLE and do that.
8148 (verify_expr): Remove.
8149 (verify_types_in_gimple_reference): Add pieces from verify_expr.
8150 (verify_gimple_assign_single): Likewise.
8151 (verify_gimple_switch): Likewise.
8152 (verify_expr_location_1): Dereference tp once. Add (disabled)
8153 piece from verify_expr.
8154 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
8156 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
8158 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
8160 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
8162 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
8163 (small_data_pattern): Likewise.
8164 (arc_rewrite_small_data): Likewise.
8165 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
8166 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
8167 (get_symbol_alignment): New function.
8168 (legitimate_small_data_address_p): Likewise.
8169 (legitimate_scaled_address): Update, call
8170 legitimate_small_data_address_p.
8171 (output_sdata): New static variable.
8172 (arc_print_operand): Update how we handle small data operands.
8173 (arc_print_operand_address): Likewise.
8174 (arc_legitimate_address_p): Update, use
8175 legitimate_small_data_address_p.
8176 (arc_rewrite_small_data_p): Remove.
8177 (arc_rewrite_small_data_1): Likewise.
8178 (arc_rewrite_small_data): Likewise.
8179 (small_data_pattern): Likewise.
8180 (compact_sda_memory_operand): Update to use
8181 legitimate_small_data_address_p and get_symbol_alignment.
8182 (prepare_move_operands): Don't rewite sdata pattern.
8183 (prepare_extend_operands): Remove.
8184 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
8186 (zero_extendqisi2): Likewise.
8187 (zero_extendhisi2): Likewise.
8188 (extendqihi2): Likewise.
8189 (extendqisi2): Likewise.
8190 (extendhisi2): Likewise.
8194 * config/arc/constraints.md (Usd): Change it to memory constraint.
8196 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
8198 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
8199 as source of std instructions.
8200 * config/arc/arc.md (movsi_insn): Update pattern predicate to
8201 allow 6-bit constants as source for store instructions.
8202 (movdi_insn): Update instruction pattern to allow 6-bit constants
8203 as source for store instructions.
8205 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
8207 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
8209 2018-04-30 Nathan Sidwell <nathan@acm.org>
8210 Sandra Loosemore <sandra@codesourcery.com>
8212 * dumpfile.c (dump_open): Allow '-' for stdout.
8213 * doc/invoke.texi (Developer Options): Document dump filename
8214 determination early. Document stdin/stdout selection.
8216 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
8218 Microblaze Target: PIC data text relative
8220 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
8221 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
8223 * gcc/config/microblaze/microblaze.h (microblaze_constant_address_p):
8224 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
8225 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
8226 New addressing mode for data-text relative position indepenedent code.
8227 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
8228 'ADDRESS_SYMBOLIC_TXT_REL'.
8229 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
8230 (microblaze_legitimate_pic_operand): Exclude function calls from
8231 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
8232 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
8234 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
8235 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
8236 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
8237 for 'address + offset'.
8238 (microblaze_expand_prologue): Add new function prologue call for
8240 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
8241 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
8242 table in case of TARGET_PIC_DATA_TEXT_REL.
8243 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
8244 * gcc/config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
8245 Add new macros 'UNSPEC_TEXT',
8246 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
8247 + exclude function calls from 'UNSPEC_PLT' in case of data text
8249 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
8250 new target hook for generating address diff vector tables in case of
8252 * doc/tm.texi : Regenerate.
8253 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
8254 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
8255 of addr diff vector generation.
8256 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
8257 target hook definition.
8258 * targhooks.h, gcc/targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
8259 Add default function for generate_pic_addr_diff_vec -> flag_pic.
8260 * doc/invoke.texi (Add new pic option): Add new microblaze pic
8261 option for data text relative.
8263 2018-04-30 Richard Biener <rguenther@suse.de>
8265 * tree-chrec.h (evolution_function_is_constant_p): Remove
8267 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
8269 2018-04-30 Richard Biener <rguenther@suse.de>
8272 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
8274 2018-04-30 Richard Biener <rguenther@suse.de>
8276 PR tree-optimization/28364
8277 PR tree-optimization/85275
8278 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
8279 copying first exit test.
8281 2018-04-28 Mark Wielaard <mark@klomp.org>
8283 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
8285 (dwarf_AT): Handle DW_AT_addr_base.
8286 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
8288 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
8291 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
8292 (*ashl<dwi>3_doubleword_mask_1): Ditto.
8293 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
8294 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
8296 2018-04-28 Richard Biener <rguenther@suse.de>
8298 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
8299 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
8300 to reflect use. Only add interesting stmts.
8302 2018-04-27 Martin Jambor <mjambor@suse.cz>
8305 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
8306 the jump function allows for passing through aggregate values.
8308 2018-04-27 David Malcolm <dmalcolm@redhat.com>
8310 * input.h (in_system_header_at): Convert from macro to inline
8312 (from_macro_expansion_at): Likewise.
8313 (from_macro_definition_at): Likewise.
8315 2018-04-27 Jeff Law <law@redhat.com>
8317 * config.gcc: Mark tile* targets as deprecated/obsolete.
8319 2018-04-27 Richard Biener <rguenther@suse.de>
8321 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
8324 2018-04-27 Richard Biener <rguenther@suse.de>
8326 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
8328 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
8330 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
8331 with Yd constraint. Set "preferred_for_speed" attribute from
8332 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
8334 (*movdi_internal): Ditto.
8335 (movti_interunit splitters): Remove
8336 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
8337 (movdi_interunit splitters): Ditto.
8338 * config/i386/constraints.md (Ye): Remove.
8339 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
8341 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8344 * config/aarch64/constraints.md (Usg): Limit to 31.
8347 2018-04-27 Jakub Jelinek <jakub@redhat.com>
8349 PR tree-optimization/85529
8350 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
8351 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
8352 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
8353 zero extension or masking of the MSB bit.
8354 (optimize_range_tests): Add FIRST_BB argument, pass it through
8355 to optimize_range_tests_var_bound.
8356 (maybe_optimize_range_tests, reassociate_bb): Adjust
8357 optimize_range_tests callers.
8359 2018-04-26 Richard Biener <rguenther@suse.de>
8360 Jakub Jelinek <jakub@redhat.com>
8362 * cgraph.h (symbol_table): Just declare debug method here.
8363 * symtab.c (symbol_table::debug): Define.
8365 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
8367 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
8369 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
8371 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
8372 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
8373 (*movdi_internal): Substitute Yi and Yj constraint with x
8374 and Ym and Yn constraint with y constraint. Update "isa"
8375 attribute and set "preferred_for_speed" attribute from
8376 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
8377 (*movsi_internal): Ditto.
8378 (*movdf_internal): Ditto.
8379 (*movsf_internal): Ditto.
8380 (*zero_extendsidi2): Ditto.
8381 * config/i386/sse.md (vec_set<mode>_0): Ditto.
8382 (sse2_loadld): Ditto.
8383 (*vec_extract<ssevecmodelower>_0): Ditto.
8384 (*vec_extractv4si_0_zext_sse4): Ditto.
8385 (vec_concatv2di): Ditto.
8386 (*vec_dup<mode>): Ditto.
8387 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
8388 * config/i386/constraints.md (Yi): Remove.
8393 2018-04-26 Nathan Sidwell <nathan@acm.org>
8395 * dumpfile.c (dump_open): New.
8396 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
8397 (dump_finish): Detect stdio/stderr by value not name.
8399 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
8401 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
8403 2018-04-26 Tom de Vries <tom@codesourcery.com>
8406 * config/nvptx/nvptx.c (verify_neutering_jumps)
8407 (verify_neutering_labels): New function
8408 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
8410 2018-04-26 Tom de Vries <tom@codesourcery.com>
8413 * config/nvptx/nvptx.c (needs_neutering_p): New function.
8414 (nvptx_single): Use needs_neutering_p to skip over insns that do not
8417 2018-04-26 Richard Biener <rguenther@suse.de>
8418 Tom de Vries <tom@codesourcery.com>
8421 * lto-streamer-out.c (output_function): Fixup loops if required to match
8422 discovery done in the reader.
8424 2018-04-26 Richard Biener <rguenther@suse.de>
8426 PR tree-optimization/85116
8427 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
8428 have a loop exit from the single latch predecessor. Remove
8429 case of header with just condition.
8430 (ch_base::copy_headers): Exclude infinite loops from any
8432 (pass_ch::execute): Record exits.
8434 2018-04-26 Richard Biener <rguenther@suse.de>
8436 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
8437 prologue cost vector and pass it to vect_get_load_cost.
8438 (vect_get_peeling_costs_all_drs): Likewise.
8439 (vect_peeling_hash_get_lowest_cost): Likewise.
8440 (vect_enhance_data_refs_alignment): Likewise.
8442 2018-04-26 Richard Biener <rguenther@suse.de>
8445 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
8446 checking of integer<->pointer conversions.
8447 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
8448 sign-/zero-extending pointer types.
8449 (expand_omp_for_static_chunk): Likewise.
8451 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
8452 Jean Lee <xiaoyur347@gmail.com>
8454 * config/mips/mips.c (mips_asan_shadow_offset): New function.
8455 (TARGET_ASAN_SHADOW_OFFSET): Define.
8456 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
8457 true for -fsanitize=address.
8459 2018-04-25 Mark Wielaard <mark@klomp.org>
8461 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
8464 2018-04-25 Jakub Jelinek <jakub@redhat.com>
8466 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
8467 than "alu", remove explicit "memory" and "imm_disp" attributes.
8468 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
8471 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
8472 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
8473 gen_lowpart_no_emit.
8475 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
8478 * config/i386/i386.c (ix86_expand_builtin): Change memory
8479 operand to XI, extend p0 to Pmode.
8480 * config/i386/i386.md: Change unspec volatile and operand
8481 1 mode to XI, change operand 0 mode to P.
8483 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
8485 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
8486 GET_MODE_MASK before any checking.
8487 (nds32_can_use_bset_p): Likewise.
8488 (nds32_can_use_btgl_p): Likewise.
8490 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
8492 * config/nds32/nds32-doubleword.md: New define_split pattern for
8493 illegal register number.
8495 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
8497 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
8499 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
8501 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
8503 2018-04-25 Richard Biener <rguenther@suse.de>
8505 * lto-streamer.h (LTO_major_version): Bump to 8.
8507 2018-04-25 Jakub Jelinek <jakub@redhat.com>
8509 * BASE-VER: Set to 9.0.0.
8511 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
8513 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
8514 in __abskf2 and __powikf2.
8516 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8519 * config/aarch64/constraints.md (Usg, Usj): New constraints.
8520 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
8521 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
8522 Use the above on operand 2. Reindent.
8523 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
8525 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
8528 * common/config/i386/i386-common.c (ix86_handle_option): Don't
8530 * config/i386/i386.opt (mcet): Removed.
8531 * doc/install.texi: Remove -mcet documentation.
8532 * doc/invoke.texi: Likewise.
8534 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
8537 * doc/install.texi: Remove -mcet from bootstrap-cet.
8539 2018-04-24 Jakub Jelinek <jakub@redhat.com>
8542 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
8543 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
8547 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
8548 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
8549 containing a CONST_VECTOR.
8551 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
8553 * doc/install.texi: Update newlib dependency for nvptx.
8555 2018-04-24 Jakub Jelinek <jakub@redhat.com>
8558 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
8559 instead of INTVAL when shifting x left.
8561 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
8563 PR tree-optimization/85478
8564 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
8565 vect_grouped_store_supported for single element vectors.
8567 2018-04-24 Richard Biener <rguenther@suse.de>
8570 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
8571 load cost increase to the case of non-constant step.
8573 2018-04-24 Jakub Jelinek <jakub@redhat.com>
8576 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
8577 destination if any_malformed_asm.
8579 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
8582 * expr.c (store_field): In the bitfield case, if the value comes from
8583 a function call and is returned in registers by means of a PARALLEL,
8584 do not change the mode of the temporary unless BLKmode and VOIDmode.
8586 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
8588 PR rtl-optimization/85423
8589 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
8590 dependencies to debug insns when the previous insn is non-debug.
8592 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
8594 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
8595 enums into a single definition.
8596 (fls): Fix predicates and printing.
8599 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
8601 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
8602 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
8603 and short u6 immediate.
8604 (check_if_valid_sleep_operand): Remove.
8605 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
8607 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
8609 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
8610 flag_always_save_lp condition.
8611 * config/nds32/nds32.opt (malways-save-lp): New option.
8613 2018-04-22 Shiva Chen <shiva0217@gmail.com>
8615 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
8616 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
8617 * config/nds32/nds32.h
8618 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
8619 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
8621 2018-04-22 Shiva Chen <shiva0217@gmail.com>
8623 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
8624 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
8626 2018-04-22 Shiva Chen <shiva0217@gmail.com>
8627 Chung-Ju Wu <jasonwucj@gmail.com>
8629 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
8631 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
8632 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
8634 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
8636 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
8638 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
8640 * config/nds32/nds32-protos.h (nds32_data_alignment,
8641 nds32_local_alignment): Declare.
8642 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
8643 nds32_local_alignment): New functions.
8644 (TARGET_CONSTANT_ALIGNMENT): Define.
8645 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
8647 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
8649 * config/nds32/nds32.c
8650 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
8651 (TARGET_MODES_TIEABLE_P): Likewise.
8653 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
8655 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
8658 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
8659 Chung-Ju Wu <jasonwucj@gmail.com>
8661 * config/nds32/constants.md (unspec_volatile_element): Add enum values
8662 for unaligned access.
8663 * config/nds32/nds32-intrinsic.c: Implementation of expanding
8665 * config/nds32/nds32-intrinsic.md: Likewise.
8666 * config/nds32/nds32_intrinsic.h: Likewise.
8667 * config/nds32/nds32.h (nds32_builtins): Likewise.
8668 * config/nds32/nds32.opt (munaligned-access): New option.
8669 * config/nds32/nds32.c (nds32_asm_file_start): Display
8670 flag_unaligned_access status.
8672 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
8674 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
8675 -mno-relax is present.
8676 * config/riscv/linux.h (LINK_SPEC): Ditto.
8678 2018-04-20 Martin Sebor <msebor@redhat.com>
8681 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
8683 (gimple_fold_builtin_stxcpy_chk): Same.
8684 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
8686 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
8689 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
8690 __powikf2 when long double is IEEE 128-bit.
8692 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
8694 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
8695 step to make sure stack always aligned.
8697 2018-04-20 Carl Love <cel@us.ibm.com>
8700 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
8701 size check for arg0.
8703 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
8704 Tom de Vries <tom@codesourcery.com>
8707 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
8708 Emit insns for calls too.
8709 (nvptx_find_par): Always look for worker-level predecessor insn.
8710 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
8712 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
8713 (nvptx_process_pars): Propagate frames for calls.
8715 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
8718 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
8720 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
8721 (ix86_handle_option): Don't handle OPT_mibt.
8722 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
8724 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
8726 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
8727 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
8728 (ix86_target_macros): Define __CET__ with flag_cf_protection
8729 for -fcf-protection.
8730 * config/i386/i386.c (isa2_opts): Remove -mibt.
8731 * config/i386/i386.h (TARGET_IBT): Removed.
8732 (TARGET_IBT_P): Likewise.
8733 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
8734 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
8735 * config/i386/i386.opt (mcet): Update help message.
8738 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
8739 -mcet as an alias for -mshstk.
8741 2018-04-20 Richard Biener <rguenther@suse.de>
8744 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
8745 complexity by forcing a single use of the multiply operand.
8747 2018-04-20 Martin Jambor <mjambor@suse.cz>
8750 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
8751 recursion dependency to only apply to non-clones.
8753 2018-04-20 Martin Jambor <mjambor@suse.cz>
8756 * ipa-cp.c (create_specialized_node): Check that clones of
8757 self-recursive edges exist during IPA-CP.
8759 2018-04-19 Toon Moene <toon@moene.org>
8761 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
8764 2018-04-19 Jakub Jelinek <jakub@redhat.com>
8766 PR tree-optimization/85467
8767 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
8768 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
8769 VECTOR_CST element to type.
8771 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
8774 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
8775 * config/i386/i386.md (builtin_setjmp_setup): Removed.
8776 (builtin_longjmp): Likewise.
8777 (save_stack_nonlocal): New pattern.
8778 (restore_stack_nonlocal): Likewise.
8780 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
8783 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
8784 Replace ASM_OUTPUT_LABEL with fprintf.
8786 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
8789 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
8790 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
8791 * config/i386/i386-c.c (ix86_target_macros_internal): Also
8792 define __IBT__ and __SHSTK__ for -fcf-protection.
8793 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
8795 (ix86_trampoline_init): Likewise.
8796 (x86_output_mi_thunk): Likewise.
8797 (ix86_notrack_prefixed_insn_p): Likewise.
8798 (ix86_option_override_internal): Don't disallow -fcf-protection.
8799 * config/i386/i386.md (rdssp<mode>): Also enable for
8801 (incssp<mode>): Likewise.
8802 (nop_endbr): Likewise.
8803 * config/i386/i386.opt (mcet): Change help message to built-in
8807 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
8808 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
8809 enable CET built-in functions.
8811 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
8813 * common/config/i386/i386-common.c
8814 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
8815 OPTION_MASK_ISA_MOVDIRI_UNSET,
8816 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
8817 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
8818 * config.gcc (movdirintrin.h): New header.
8819 * config/i386/cpuid.h (bit_MOVDIRI,
8820 bit_MOVDIR64B): New bits.
8821 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
8823 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
8824 (VOID, PVOID, PCVOID)): New function types.
8825 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
8826 __builtin_ia32_directstoreu_u64,
8827 __builtin_ia32_movdir64b): New builtins.
8828 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
8829 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
8831 (ix86_valid_target_attribute_inner_p): Ditto.
8832 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
8833 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
8834 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
8835 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
8836 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
8837 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
8838 (movdiri<mode>, movdir64b_<mode>): New.
8839 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
8840 * config/i386/immintrin.h: Include movdirintrin.h.
8841 * config/i386/movdirintrin.h: New file.
8842 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
8844 2018-04-19 Richard Biener <rguenther@suse.de>
8847 * cfg.c (clear_bb_flags): When loop state says we have
8848 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
8850 2018-04-19 Richard Biener <rguenther@suse.de>
8852 PR tree-optimization/84737
8853 * tree-vect-data-refs.c (vect_copy_ref_info): New function
8854 copying restrict info.
8855 (vect_setup_realignment): Use it.
8856 * tree-vectorizer.h (vect_copy_ref_info): Declare.
8857 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
8858 the first DR to all generated stores.
8859 (vectorizable_load): Likewise for loads.
8861 2018-04-19 Jakub Jelinek <jakub@redhat.com>
8863 PR tree-optimization/85446
8864 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
8865 the integral and pointer types to have the same precision.
8867 * doc/install.texi: Document --disable-cet being the default and
8870 2018-04-18 Martin Liska <mliska@suse.cz>
8872 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
8875 2018-04-18 Martin Liska <mliska@suse.cz>
8878 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
8881 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
8882 arguments if they are comparable.
8884 2018-04-18 Martin Liska <mliska@suse.cz>
8887 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
8890 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
8893 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
8896 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
8897 ENDBR after calling __morestack.
8899 2018-04-18 David Malcolm <dmalcolm@redhat.com>
8902 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
8903 by using gcc_base_ver to generate a gcc_driver_version, and use
8904 it when generating GCC_DRIVER_NAME.
8905 * configure: Regenerate.
8907 2018-04-18 Jakub Jelinek <jakub@redhat.com>
8910 * config.gcc: Obsolete powerpc*-*-*spe*.
8912 2018-04-17 Jakub Jelinek <jakub@redhat.com>
8915 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
8916 (stabstr_D): Change type of unum from unsigned int to
8917 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
8920 2018-04-17 Jim Wilson <jimw@sifive.com>
8923 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
8924 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
8925 Set arg_pointer_offset after using pretend_args_size.
8927 2018-04-17 Jakub Jelinek <jakub@redhat.com>
8929 PR rtl-optimization/85431
8930 * dse.c (record_store): Ignore zero width stores.
8933 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
8934 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
8935 __builtin_stack_restore rather than after it.
8936 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
8937 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
8938 argument instead of virtual_dynamic_stack_rtx.
8940 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
8942 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
8944 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8945 Add note to error message to explain internal mapping of overloaded
8946 built-in function name to non-overloaded built-in function name.
8947 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
8950 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
8953 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
8954 where the inputs overlap with the output.
8956 2018-04-17 Jakub Jelinek <jakub@redhat.com>
8958 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
8959 (=v, v) alternative and explicit "memory" attribute.
8960 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
8961 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
8963 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
8964 "sselog1" type instead of "sselog".
8965 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
8966 "sselog". Remove explicit "memory" attribute.
8967 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
8968 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
8970 (vec_extract_hi_v32hi): Merge all alternatives into one, use
8971 "sselog1" type instead of "sselog". Remove explicit "memory"
8973 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
8974 use "sselog1" type instead of "sselog". Remove explicit "memory"
8976 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
8977 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
8979 (vec_extract_hi_v64qi): Merge all alternatives into one, use
8980 "sselog1" type instead of "sselog". Remove explicit "memory"
8982 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
8983 use "sselog1" type instead of "sselog". Remove explicit "memory"
8987 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
8990 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
8993 2018-04-17 Martin Jambor <mjambor@suse.cz>
8996 * ipa-cp.c (create_specialized_node): Call
8997 expand_all_artificial_thunks if necessary.
8999 2018-04-17 Martin Liska <mliska@suse.cz>
9002 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
9003 in message, remote space in between '_G' and '('.
9005 2018-04-17 Jakub Jelinek <jakub@redhat.com>
9008 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
9009 avx512f_vmcmp<mode>3<round_saeonly_name>,
9010 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
9011 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
9012 avx512f_rndscale<mode><round_saeonly_name>,
9013 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
9014 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
9015 Use %<iptr>2 instead of %2 for -masm=intel.
9016 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
9017 avx512f_vcvttss2usi<round_saeonly_name>,
9018 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
9020 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
9021 avx512f_vcvttsd2usi<round_saeonly_name>,
9022 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
9023 Use %q1 instead of %1 for -masm=intel.
9024 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
9025 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
9026 of %3 for -masm=intel.
9027 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
9029 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
9031 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
9033 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
9034 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
9036 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
9038 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
9039 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
9040 %g1 and one with %0 and %1.
9041 (avx512er_vmrcp28<mode><round_saeonly_name>,
9042 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
9044 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
9045 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
9046 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
9047 of %0 and %{%4%} for -masm=intel.
9048 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
9049 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
9050 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
9051 order of %0 and %{%5%}%{z%} for -masm=intel.
9053 2018-04-17 Jan Hubicka <jh@suse.cz>
9056 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
9058 2018-04-17 Martin Liska <mliska@suse.cz>
9061 * multiple_target.c (create_dispatcher_calls): Set apostrophes
9062 for target_clone error message. Make default implementation
9063 clone to be a local declaration.
9064 (separate_attrs): Add new argument and check for an empty
9066 (expand_target_clones): Handle it.
9067 (ipa_target_clone): Make redirection just for target_clones
9070 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
9071 Tom de Vries <tom@codesourcery.com>
9074 * omp-expand.c (expand_oacc_for): Add dummy false branch for
9075 tiled basic blocks without omp continue statements.
9077 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
9080 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
9081 vec_extract expression as having side effects to make sure it gets
9084 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
9087 * config/i386/i386.c (get_builtin_code_for_version): Check
9090 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
9093 * gcc/config.gcc: Support "skylake".
9094 * gcc/config/i386/i386-c.c (ix86_target_macros_internal): Handle
9096 * gcc/config/i386/i386.c (m_SKYLAKE): Define.
9097 (processor_target_table): Add "skylake".
9098 (ix86_option_override_internal): Add "skylake".
9099 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
9100 PROCESSOR_CANNONLAKE.
9101 (get_builtin_code_for_version): Fix priority for
9102 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
9103 PROCESSOR_SKYLAKE-AVX512.
9104 * gcc/config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
9105 (processor_type): Add PROCESSOR_SKYLAKE.
9107 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
9108 Jason Merrill <jason@redhat.com>
9111 * convert.c (convert_to_integer_1): Use direct recursion for
9112 enumeral types and types with a precision less than the number
9113 of bits in their mode.
9115 2018-04-16 Julia Koval <julia.koval@intel.com>
9118 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
9119 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
9121 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
9124 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
9125 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
9126 and -mno-direct-move.
9128 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
9131 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
9132 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
9133 Ensure negative shifts result in {0}.
9135 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
9137 PR rtl-optimization/79916
9138 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
9139 regs (if any) to define how to gnerate SD moves when LRA is in
9142 2018-04-13 Jakub Jelinek <jakub@redhat.com>
9144 PR rtl-optimization/85393
9145 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
9146 * except.c (expand_dw2_landing_pad_for_region): Make static.
9147 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
9148 a label and unconditional jump to old_bb, rather than
9149 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
9152 PR rtl-optimization/85376
9153 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
9154 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
9155 instead of a specific value.
9157 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
9158 Bin Cheng <bin.cheng@arm.com>
9160 PR tree-optimization/82965
9161 PR tree-optimization/83991
9162 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
9163 by_profile_only parameter.
9164 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
9165 information if the loop was predicted to iterate too many times.
9166 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
9168 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
9171 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
9174 2018-04-13 Martin Liska <mliska@suse.cz>
9175 Jakub Jelinek <jakub@redhat.com>
9178 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
9179 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
9180 * builtins.c (expand_builtin_memory_copy_args): Use
9181 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
9182 handle dest_addr == pc_rtx.
9184 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
9187 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
9188 asked to not generate direct moves.
9189 (fix_trunc<mode>si2_stfiwx): Similar.
9190 (fix_trunc<mode>si2_internal): Similar.
9192 2018-04-12 Jakub Jelinek <jakub@redhat.com>
9195 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
9196 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
9197 lookup if dest in some wider mode is known to be const0_rtx and
9198 if so, record permanent equivalence for it to be ZERO_EXTEND of
9199 the narrower mode destination.
9201 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
9203 * lto-streamer-out.c (output_function): Revert 259346.
9204 * omp-expand.c (expand_oacc_for): Likewise.
9206 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
9208 PR rtl-optimization/85354
9209 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
9210 * sel-sched.c (sel_global_init): ... here.
9212 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
9215 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
9216 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
9217 mode for PE-COFF targets.
9218 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
9219 (i386_pe_asm_lto_end): Likewise.
9220 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
9221 (TARGET_ASM_LTO_END): Likewise.
9222 * config/i386/winnt.c (saved_debug_info_level): New static variable.
9223 (i386_pe_asm_lto_start): New function.
9224 (i386_pe_asm_lto_end): Likewise.
9226 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
9227 Richard Biener <rguenther@suse.de>
9230 * lto-streamer-out.c (output_function): Fix CFG loop state before
9232 * omp-expand.c (expand_oacc_for): Handle calls to internal
9233 functions like regular functions.
9235 2018-04-12 Richard Biener <rguenther@suse.de>
9238 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
9239 for the early LTO debug to properly generate references to it
9240 during DIE emission. Do not re-use that for the skeleton for
9242 (dwarf2out_early_finish): Likewise.
9244 2018-04-12 Jakub Jelinek <jakub@redhat.com>
9247 * config/i386/sse.md
9248 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
9249 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
9250 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
9251 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
9252 and output is a reg, avoid creating invalid lowpart subreg, but
9253 instead split into a 512-bit move. Don't split if not AVX512VL,
9254 input is xmm16+ reg and output is a mem.
9255 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
9256 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
9257 xmm16+ reg and output is a mem.
9259 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9261 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
9262 also for flag_dwarf2_cfi_asm.
9264 2018-04-12 Jakub Jelinek <jakub@redhat.com>
9266 PR rtl-optimization/85342
9267 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
9268 a bool scalar var inside of the loop instead. Don't try to update
9269 recog_data.operand after failed apply_change_group.
9271 2018-04-12 Tom de Vries <tom@codesourcery.com>
9274 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
9275 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
9276 array with flexible array member as array without given dimension.
9277 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
9278 argument for undefined param to true.
9280 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
9283 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
9284 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
9285 from PowerPC section.
9286 * config/rs6000/sysv4.opt (mcall-): Improve help text.
9287 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
9288 help text that is too long.
9289 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
9290 help text that is too long.
9291 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
9292 help text that is too long.
9294 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
9296 * config/alpha/alpha.md (stack_probe_internal): Rename
9297 from "probe_stack". Update all callers.
9299 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
9301 PR rtl-optimization/84566
9302 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
9303 sched_macro_fuse_insns.
9305 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
9308 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
9309 (compute_block_dependences): ... from here.
9311 2018-04-11 Jakub Jelinek <jakub@redhat.com>
9313 PR tree-optimization/85331
9314 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
9315 from int to HOST_WIDE_INT.
9317 2018-04-11 Martin Jambor <mjambor@suse.cz>
9320 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
9321 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
9322 not the same as the source val.
9323 (cgraph_edge_brings_value_p): New parameter.
9324 (gather_edges_for_value): Pass destination value to
9325 cgraph_edge_brings_value_p.
9326 (perhaps_add_new_callers): Likewise.
9327 (get_info_about_necessary_edges): Likewise and exclude values brought
9328 only by self-recursive edges.
9329 (create_specialized_node): Redirect only clones of self-calling edges.
9330 (+self_recursive_pass_through_p): New function.
9331 (find_more_scalar_values_for_callers_subset): Use it.
9332 (find_aggregate_values_for_callers_subset): Likewise.
9333 (known_aggs_to_agg_replacement_list): Removed.
9334 (decide_whether_version_node): Re-calculate known constants for all
9335 remaining context clones.
9337 2018-04-11 Richard Biener <rguenther@suse.de>
9340 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
9341 from early DWARF output.
9342 (dwarf2out_early_finish): Output line info unconditionally into
9343 early DWARF and add reference to it.
9345 2018-04-11 Jakub Jelinek <jakub@redhat.com>
9348 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
9349 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
9350 other than V2DFmode using iptr mode attribute.
9351 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
9353 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
9355 PR rtl-optimization/84659
9356 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
9358 2018-04-11 Jakub Jelinek <jakub@redhat.com>
9361 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
9363 (output_loc_list): Pass address of a dummy size variable even in the
9364 locview handling loop.
9365 (index_location_lists): Add comment on why skip_loc_list_entry can't
9368 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
9371 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
9374 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
9377 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
9378 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
9379 and -mstring-compare-inline-limit.
9381 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
9384 * gcc/config/rs6000/rs6000.md (allocate_stack): Put the residual size
9385 for stack clash protection in a register whenever we need it to be in
9388 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
9390 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
9391 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
9393 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
9396 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
9398 (mlong-double-): Ditto.
9399 * config/rs6000/sysv4.opt (msdata=): Ditto.
9400 (mtls-size=): Ditto.
9402 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
9404 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
9405 erroneous entries for
9406 "vector int vec_ldl (int, long int *)", and
9407 "vector unsigned int vec_ldl (int, unsigned long int *)".
9408 Add comments and entries for
9409 "vector bool char vec_ldl (int, bool char *)",
9410 "vector bool short vec_ldl (int, bool short *)",
9411 "vector bool int vec_ldl (int, bool int *)",
9412 "vector bool long long vec_ldl (int, bool long long *)",
9413 "vector pixel vec_ldl (int, pixel *)",
9414 "vector long long vec_ldl (int, long long *)",
9415 "vector unsigned long long vec_ldl (int, unsigned long long *)".
9416 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
9417 type tree bool_long_long_type_node and correct definition of
9418 bool_V2DI_type_node to make reference to this new type tree.
9419 (rs6000_mangle_type): Replace erroneous reference to
9420 bool_long_type_node with bool_long_long_type_node.
9421 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
9422 comments to emphasize sign distinctions for char and int types and
9423 replace RS6000_BTI_bool_long constant with
9424 RS6000_BTI_bool_long_long constant. Also add comment to restrict
9425 use of RS6000_BTI_pixel.
9426 (bool_long_type_node): Remove this macro definition.
9427 (bool_long_long_type_node): New macro definition
9429 2018-04-10 Jakub Jelinek <jakub@redhat.com>
9431 PR rtl-optimization/85300
9432 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
9433 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
9434 simplify_unary_operation fails.
9436 2018-04-10 Martin Liska <mliska@suse.cz>
9438 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
9439 cgraph_edge and ipa_ref.
9441 2018-04-10 Jakub Jelinek <jakub@redhat.com>
9445 * config/i386/sse.md
9446 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
9447 computation of the VEC_MERGE selector from mask.
9448 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
9449 Fix decoding of the VEC_MERGE selector into mask.
9451 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
9453 PR tree-optimization/85286
9454 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
9456 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
9458 * final.c (final_1): Set insn_last_address as well as
9459 insn_current_address.
9461 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9464 * explow.c (emit_stack_probe): Call validize_mem on memory location
9465 before passing it to gen_probe_stack. Create address operand and
9466 legitimize it for the probe_stack_address case.
9468 2018-04-09 Jan Hubicka <jh@suse.cz>
9471 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
9472 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
9473 * tree.c (free_lang_data_in_type): Fix handling of binfos;
9475 (free_lang_data): Rebuild type inheritance graph.
9477 2018-04-09 Martin Sebor <msebor@redhat.com>
9479 * invoke.texi (-finline-small-functions): Mention other optimization
9481 (-findirect-inlining, -fpartial-inlining): Same.
9482 (-finline-functions-called-once): Same.
9483 (-freorder-blocks-and-partition): Same.
9485 2018-04-09 Jan Hubicka <jh@suse.cz>
9488 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
9489 jumps; choose last target that matches the criteria (i.e.
9490 no partition changes for non-crossing jumps).
9491 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
9492 support for redirecting crossing jumps to non-crossing.
9494 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
9496 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
9497 also for naked functions.
9499 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
9501 * config/arc/arc.md (add_shift): New pattern.
9502 (add_shift2): Likewise.
9503 (sub_shift): Likewise.
9504 (sub_shift_cmp0_noout): Likewise.
9505 (compare_si_ashiftsi): Likewise.
9506 (xbfu_cmp0_noout): New combine pattern.
9507 (xbfu_cmp0"): Likewise.
9508 (movsi_set_cc_insn): Place the predicable variant first.
9509 (commutative_binary_cmp0_noout): Remove clobber.
9510 (commutative_binary_cmp0): New pattern.
9511 (noncommutative_binary_cmp0): Likewise.
9512 (noncommutative_binary_cmp0_noout): Likewise.
9513 (noncommutative_binary_comparison_result_used): Removed.
9514 (rsub_cmp0): New pattern.
9515 (rsub_cmp0_noout): Likewise.
9516 (extzvsi): Changed, keep only meaningful variants.
9517 (SQH, SEZ): New iterators.
9518 (SQH_postfix): New mode attribute.
9519 (SEZ_prefix): New code attribute.
9520 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
9521 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
9522 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
9524 (noncommutative_operator): Check the availability of barrel
9527 2018-04-09 Richard Biener <rguenther@suse.de>
9529 PR tree-optimization/85284
9530 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
9531 Only use the niter constraining form of simple_iv when the exit
9534 2018-04-09 Tom de Vries <tom@codesourcery.com>
9537 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
9538 (define_expand "*memory_barrier"): New define_expand.
9539 (define_insn "memory_barrier"): New insn.
9541 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
9543 PR rtl-optimization/80463
9544 PR rtl-optimization/83972
9545 PR rtl-optimization/83480
9547 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
9548 correct producer for the insn.
9549 (tidy_control_flow): Fixup seqnos in case of debug insns.
9551 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
9553 PR rtl-optimization/83913
9555 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
9556 different sched-times when merging exprs.
9558 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
9560 PR rtl-optimization/83962
9562 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
9563 tidy_fallthru_edge and tidy_control_flow.
9565 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
9567 PR rtl-optimization/83530
9569 * sel-sched.c (force_next_insn): New global variable.
9570 (remove_insn_for_debug): When force_next_insn is true, also leave only
9571 next insn in the ready list.
9572 (sel_sched_region): When the region wasn't scheduled, make another pass
9573 over it with force_next_insn set to 1.
9575 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
9577 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
9579 * config/nds32/constants.md (unspec_volatile_element): Add enum values
9580 for interrupt control.
9581 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
9582 functions for interrupt control.
9583 * config/nds32/nds32-intrinsic.md: Likewise.
9584 * config/nds32/nds32_intrinsic.h: Likewise.
9585 * config/nds32/nds32.h (nds32_builtins): Likewise.
9587 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
9589 * config/nds32/nds32.c (nds32_init_machine_status,
9590 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
9591 strict_aligned_p field.
9592 (nds32_expand_to_rtl_hook): New function.
9593 (TARGET_EXPAND_TO_RTL_HOOK): Define.
9594 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
9596 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
9597 Chung-Ju Wu <jasonwucj@gmail.com>
9599 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
9600 * config/nds32/nds32-n7.md: New file.
9601 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
9602 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
9604 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
9605 * config/nds32/nds32.md (pipeline_model): Add n7.
9606 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
9607 * config/nds32/pipelines.md: Include n7 settings.
9609 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
9610 Chung-Ju Wu <jasonwucj@gmail.com>
9612 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
9613 * config/nds32/nds32-e8.md: New file.
9614 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
9615 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
9617 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
9618 * config/nds32/nds32.md (pipeline_model): Add e8.
9619 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
9620 * config/nds32/pipelines.md: Include e8 settings.
9622 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
9623 Chung-Ju Wu <jasonwucj@gmail.com>
9625 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
9626 * config/nds32/nds32-n8.md: New file.
9627 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
9628 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
9630 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
9631 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
9632 * config/nds32/nds32.md (pipeline_model): Add n8.
9633 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
9634 * config/nds32/pipelines.md: Include n8 settings.
9636 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
9637 Chung-Ju Wu <jasonwucj@gmail.com>
9639 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
9640 * config/nds32/nds32-n9-2r1w.md: New file.
9641 * config/nds32/nds32-n9-3r2w.md: New file.
9642 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
9643 nds32_register_ports): New or modify for cpu n9.
9644 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
9646 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
9647 * config/nds32/nds32-utils.c: New file.
9648 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
9649 TARGET_MUL_SLOW): Define.
9650 * config/nds32/nds32.md (pipeline_model): New attribute.
9651 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
9652 New options that support cpu n9.
9653 * config/nds32/pipelines.md: Include n9 settings.
9654 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
9656 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
9658 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
9659 information if necessary.
9660 (output_cond_branch_compare_zero): Likewise.
9661 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
9662 (nds32_target_alignment): Refine for alignment.
9663 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
9664 (FUNCTION_BOUNDARY): Modify.
9665 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
9667 * config/nds32/nds32.opt (malways-align, malign-functions): New.
9669 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
9671 * config/nds32/constants.md (unspec_volatile_element): Add values for
9672 TLB operation and data prefetch.
9673 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
9674 functions for TLB operation and data prefetch.
9675 * config/nds32/nds32-intrinsic.md: Likewise.
9676 * config/nds32/nds32_intrinsic.h: Likewise.
9677 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
9678 (nds32_print_operand): Likewise.
9679 * config/nds32/nds32.h (nds32_builtins): Likewise.
9681 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
9682 Andrew Pinski <pinsika@gcc.gnu.org>
9685 * match.pd: Use constant_boolean_node of correct type instead of
9686 boolean_true_node or boolean_false_node for simplifying
9687 pointer comparisons to zero.
9689 2018-04-07 Jakub Jelinek <jakub@redhat.com>
9691 PR tree-optimization/80021
9692 * tree.c (verify_type_variant): Make error call in verify_variant_match
9693 translatable and remove final full stop.
9695 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
9697 * config/nds32/constants.md (unspec_volatile_element): Add
9698 UNSPEC_VOLATILE_EH_RETURN.
9699 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
9700 nds32_output_stack_pop): Support dwarf exception handling process.
9701 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
9702 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
9703 exception handling process.
9704 (nds32_compute_stack_frame): Likewise.
9705 (nds32_return_addr_rtx): Likewise.
9706 (nds32_initial_elimination_offset): Likewise.
9707 (nds32_expand_prologue): Likewise.
9708 (nds32_expand_epilogue): Likewise.
9709 (nds32_dynamic_chain_address): New function.
9710 * config/nds32/nds32.h (machine_function): Add fields for dwarf
9712 (DYNAMIC_CHAIN_ADDRESS): Define.
9713 (EH_RETURN_DATA_REGNO): Define.
9714 (EH_RETURN_STACKADJ_RTX): Define.
9715 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
9716 patterns for dwarf exception handling.
9718 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
9720 * config/nds32/nds32.h: Clean up obsolete macros.
9722 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
9724 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
9725 Add enum values for particular instructions.
9726 * config/nds32/nds32-intrinsic.c: Implementation of expanding
9727 particular intrinsic functions.
9728 * config/nds32/nds32-intrinsic.md: Likewise.
9729 * config/nds32/nds32_intrinsic.h: Likewise.
9730 * config/nds32/nds32.h (nds32_builtins): Likewise.
9731 * config/nds32/nds32.md (type): Add pbsad and pbsada.
9732 (btst, ave): New patterns for particular instructions.
9734 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
9736 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
9737 Add enum values for atomic load/store and memory sync.
9738 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
9740 * config/nds32/nds32-intrinsic.md: Likewise.
9741 * config/nds32/nds32_intrinsic.h: Likewise.
9742 * config/nds32/nds32.h (nds32_builtins): Likewise.
9744 2018-04-07 Jakub Jelinek <jakub@redhat.com>
9746 PR tree-optimization/85257
9747 * fold-const.c (native_encode_vector): If not all elts could fit
9748 and off is -1, return 0 rather than offset.
9749 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
9750 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
9751 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
9752 adjust buffer in native_interpret_expr call.
9754 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
9756 * config/nds32/constants.md (unspec_volatile_element): Add cache
9757 control enum values.
9758 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
9759 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
9760 * config/nds32/nds32.c (nds32_cctl_names): New.
9761 (nds32_print_operand): Handle cache control register names.
9762 * config/nds32/nds32.h (nds32_builtins): New enum values.
9763 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
9765 * config/nds32/nds32.md (type): Add mmu.
9766 * config/nds32/pipelines.md (simple_insn): Add mmu.
9768 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
9770 * config/nds32/nds32.md (type): Remove call.
9771 * config/nds32/pipelines.md (simple_insn): Likewise.
9773 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
9775 * config/nds32/constants.md (unspec_volatile_element): Add
9776 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
9777 UNSPEC_VOLATILE_FMFCFG.
9778 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
9779 description for fmfcfg and fmfcsr.
9780 (bdesc_1arg): Add fmtcsr.
9781 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
9782 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
9783 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
9784 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
9785 unspec_fmfcfg): New patterns.
9786 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
9787 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
9788 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
9789 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
9790 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
9791 __nds32__fmfcfg): Define.
9793 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
9795 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
9796 intrinsic register names.
9797 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
9798 intrinsic register enum values and macros.
9800 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
9802 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
9803 for load/store addressing form.
9804 (nds32_print_operand_address): Likewise.
9806 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
9809 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
9810 based on LABEL_REF. Remove useless assertion.
9811 (pic_address_needs_scratch): Fix formatting.
9812 (sparc_legitimize_pic_address): Minor tweaks.
9813 (sparc_delegitimize_address): Adjust assertion accordingly.
9814 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
9815 into symbolic_operand.
9816 (movsi_high_pic_label_ref): Likewise.
9817 (movsi_lo_sum_pic_label_ref): Likewise.
9818 (movdi_pic_label_ref): Likewise.
9819 (movdi_high_pic_label_ref): Likewise.
9820 (movdi_lo_sum_pic_label_ref): Likewise.
9822 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
9824 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
9825 custom LIB_SPEC setup.
9827 2018-04-06 Ruslan Bukin <br@bsdpad.com>
9828 Kito Cheng <kito.cheng@gmail.com>
9830 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
9831 * config/riscv/freebsd.h: New.
9833 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
9835 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
9836 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
9839 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
9840 Kito Cheng <kito.cheng@gmail.com>
9842 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
9843 nds32_output_call, nds32_symbol_binds_local_p): New functions.
9844 * config/nds32/nds32-protos.h (nds32_output_call,
9845 nds32_output_return): Declare.
9846 * config/nds32/nds32.md: Refine all the call and return patterns.
9848 2018-04-06 Jakub Jelinek <jakub@redhat.com>
9851 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
9852 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
9854 PR rtl-optimization/84872
9855 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
9856 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
9859 2018-04-06 Tamar Christina <tamar.christina@arm.com>
9861 * expr.c (copy_blkmode_to_reg): Revert 254862.
9862 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
9864 2018-04-06 Richard Biener <rguenther@suse.de>
9867 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
9868 after seeing a component reference with an adjacent field. Treat
9869 refs to arrays at struct end of external decls similar to
9870 refs to unconstrained commons.
9872 2018-04-06 Jakub Jelinek <jakub@redhat.com>
9875 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
9876 look through SAVE_EXPRs with non-side-effects argument. Adjust
9878 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
9881 2018-04-06 Richard Biener <rguenther@suse.de>
9884 * alias.c (find_base_term): New wrapper around find_base_term
9885 unwinding CSELIB_VAL_PTR changes.
9886 (find_base_term): Do not restore CSELIB_VAL_PTR during the
9889 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9891 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
9893 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
9894 constant definitions.
9895 ("nop"): lr 0,0 -> nopr r0
9896 ("nop_lr0", "nop_lr1"): New insn definitions.
9898 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
9900 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
9901 NDS32_V3PUSH_AVAILABLE_P macro.
9903 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
9904 Chung-Ju Wu <jasonwucj@gmail.com>
9906 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
9907 (nds32*-*-*): Add float and fpu_config into supported_defaults.
9908 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
9909 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
9910 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
9911 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
9912 * config/nds32/constraints.md: New constraints and checking for hard
9913 float configuration.
9914 * config/nds32/iterators.md: New mode iterator and attribute for hard
9915 float configuration.
9916 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
9918 * config/nds32/nds32-fpu.md: New file.
9919 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
9920 deal with hard float code generation.
9921 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
9923 (abi_type, float_reg_number): New enum type.
9924 * config/nds32/nds32-predicates.c: New predicates for hard float.
9925 * config/nds32/nds32-protos.h: Declare functions for hard float.
9926 * config/nds32/nds32.c: Implementation for hard float configuration.
9927 * config/nds32/nds32.h: Definitions for hard float configuration.
9928 * config/nds32/nds32.md: Include hard float machine description and
9929 modify patterns for hard float configuration.
9930 * config/nds32/nds32.opt: New options for hard float configuration.
9931 * config/nds32/predicates.md: New predicates for hard float
9934 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
9936 * common/config/nds32/nds32-common.c
9937 (nds32_option_optimization_table): Enable -mreleax-hint by default.
9939 2018-04-05 Jakub Jelinek <jakub@redhat.com>
9942 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
9943 CONSTRUCTOR_ELT (ctor, ...)->value.
9945 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
9948 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
9950 2018-04-05 Tom de Vries <tom@codesourcery.com>
9953 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
9956 2018-04-05 Shiva Chen <shiva0217@gmail.com>
9957 Kito Cheng <kito.cheng@gmail.com>
9959 * config/nds32/constraints.md (U33): Fine-tune checking condition.
9960 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
9961 * config/nds32/nds32.h (nds32_16bit_address_type): Add
9962 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
9964 2018-04-05 Shiva Chen <shiva0217@gmail.com>
9965 Kito Cheng <kito.cheng@gmail.com>
9967 * config/nds32/constraints.md (Ufe): New memory constraint.
9968 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
9969 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
9970 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
9972 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
9973 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
9975 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
9977 * config/nds32/nds32.md: Use optimize_size in the condition for
9978 alu-shift instructions.
9980 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
9982 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
9984 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
9986 * config/nds32/nds32.md (negsi2): Refine pattern.
9988 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
9989 Chung-Ju Wu <jasonwucj@gmail.com>
9991 * config/nds32/iterators.md (shift_rotate): New code iterator.
9992 (shift): New code attribute.
9993 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
9994 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
9995 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
9996 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
9997 bit-wise operations.
9998 (andsi3, *andsi3): Ditto.
9999 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
10000 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
10001 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
10002 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
10003 nds32_ior_operand, nds32_xor_operand): New predicates.
10005 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
10007 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
10008 (addsi3, subsi3): ... this.
10010 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
10012 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
10014 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
10016 * config/nds32/nds32.md: Adjust indention.
10018 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
10020 * config/nds32/nds32.md (feature): New attribute.
10022 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
10024 * config/nds32/nds32.md (subtype): New attribute.
10026 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10029 * config/arm/arm-builtins.c (arm_expand_builtin): Change
10030 expansion to perform a bitwise AND of the argument followed by a
10031 boolean negation of the result.
10033 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
10035 PR rtl-optimization/84878
10036 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
10037 the basic block. Assert the use reference is not artificial and that
10038 it has an associated insn.
10040 2018-04-04 Michael Matz <matz@suse.de>
10042 * builtins.c (compute_objsize): Pass correct operand
10043 to array_at_struct_end_p.
10045 2018-04-04 Richard Biener <rguenther@suse.de>
10048 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
10049 from contexts for DINFO_LEVEL_TERSE and below.
10051 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
10053 * config/nds32/nds32-doubleword.md (move_<mode>): Require
10054 resiter_operand condition.
10055 * config/nds32/nds32.md (*move<mode>): Ditto.
10057 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
10058 Monk Chiang <sh.chiang04@gmail.com>
10060 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
10062 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
10064 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
10066 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
10067 Kito Cheng <kito.cheng@gmail.com>
10069 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
10070 nds32_cond_code_str, output_cond_branch,
10071 output_cond_branch_compare_zero, nds32_expand_cbranch,
10072 nds32_expand_cstore, nds32_expand_movcc,
10073 nds32_output_cbranchsi4_equality_zero,
10074 nds32_output_cbranchsi4_equality_reg,
10075 nds32_output_cbranchsi4_equality_reg_or_const_int,
10076 nds32_output_cbranchsi4_greater_less_zero: New functions.
10077 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
10078 nds32_expand_cstore, nds32_expand_movcc,
10079 nds32_output_cbranchsi4_equality_zero,
10080 nds32_output_cbranchsi4_equality_reg,
10081 nds32_output_cbranchsi4_equality_reg_or_const_int,
10082 nds32_output_cbranchsi4_greater_less_zero): Declare.
10083 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
10084 nds32_rimm11s_operand): New predicates.
10085 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
10086 * config/nds32/nds32.md: Rewrite all the branch and conditional move
10089 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
10091 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
10092 * config/nds32/nds32.md: Ditto.
10093 * config/nds32/pipelines.md: Ditto.
10095 2018-04-04 Richard Biener <rguenther@suse.de>
10097 PR tree-optimization/85168
10098 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
10099 propagating abnormals.
10101 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
10103 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
10105 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
10106 Kito Cheng <kito.cheng@gmail.com>
10108 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
10109 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
10110 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
10111 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
10112 * config/nds32/nds32.md (sibcall_internal): New.
10113 (sibcall_register): Remove.
10114 (sibcall_immediate): Remove.
10115 (sibcall_value_internal): New.
10116 (sibcall_value_register): Remove.
10117 (sibcall_value_immediate): Remove.
10118 * config/nds32/predicates.md (nds32_general_register_operand): New.
10119 (nds32_call_address_operand): New.
10121 2018-04-03 Jakub Jelinek <jakub@redhat.com>
10123 PR rtl-optimization/85167
10124 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
10125 bb_defs if *split_p, instead preinitialize it to NULL.
10127 PR tree-optimization/85156
10128 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
10129 evaluating the argument multiple times.
10131 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
10133 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
10135 (_mm_cvtpd_ps): Likewise.
10136 (_mm_cvttpd_epi32): Likewise.
10137 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
10138 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
10139 vector, pixel, and bool following altivec.h include.
10141 2018-04-03 Martin Sebor <msebor@redhat.com>
10143 * doc/extend.texi (Common Function Attributes): Clarify.
10144 (const attribute): Likewise.
10145 (pure attribute): Likewise.
10147 2018-04-03 Jakub Jelinek <jakub@redhat.com>
10150 * config/i386/i386.c (ix86_expand_vector_set): Use
10151 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
10153 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
10155 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
10156 instructions when changing rounding bits to preserve precision bits
10157 in the x87 control word.
10159 2018-04-03 Martin Liska <mliska@suse.cz>
10161 PR tree-optimization/82491
10162 * rtl.h (strip_offset_and_add): Replace += suboffset with
10163 poly_uint64 () + suboffset.
10165 2018-03-29 Martin Liska <mliska@suse.cz>
10166 Martin Jambor <mjambor@suse.cz>
10169 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
10170 param_type is not an integral or pointer type.
10172 2018-04-03 Richard Biener <rguenther@suse.de>
10174 * sese.h (recompute_all_dominators): Remove.
10176 2018-04-02 Martin Sebor <msebor@redhat.com>
10178 * doc/invoke.texi (-Wrestrict): Fix typos.
10180 2018-04-02 Jim Wilson <jimw@sifive.com>
10182 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
10183 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
10184 (<optab>di3, <optab>si3_extend): Likewise.
10185 (<optab>si3_mask, <optab>si3_mask_1): New.
10186 (<optab>di3_mask, <optab>di3_mask_1): New.
10187 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
10188 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
10189 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
10191 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
10193 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
10196 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
10198 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
10199 (nds32_canonicalize_comparison): New function.
10201 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
10202 Kito Cheng <kito.cheng@gmail.com>
10203 Kuan-Lin Chen <kuanlinchentw@gmail.com>
10205 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
10206 * config/nds32/constants.md (unspec_volatile_element): Add
10207 UNSPEC_VOLATILE_RELAX_GROUP.
10208 * config/nds32/nds32-relax-opt.c: New file.
10209 * config/nds32/nds32-predicates.c
10210 (nds32_symbol_load_store_p): New function.
10211 * config/nds32/nds32-protos.h
10212 (nds32_symbol_load_store_p): Declare function.
10213 (make_pass_nds32_relax_opt): Declare new rtl pass function.
10214 * config/nds32/nds32.c
10215 (nds32_register_pass): New function to register pass.
10216 (nds32_register_passes): New function to register passes.
10217 * config/nds32/nds32.md (relax_group): New pattern.
10218 * config/nds32/nds32.opt (mrelax-hint): New option.
10219 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
10221 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
10223 * config/nds32/t-nds32: Modify files dependency.
10225 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
10227 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
10228 (PROFILE_HOOK): Define its implementation.
10230 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
10232 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
10233 type and 32-bit size.
10235 2018-04-01 Jakub Jelinek <jakub@redhat.com>
10237 PR middle-end/85090
10238 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
10239 (V_128_256): New mode iterator.
10240 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
10241 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
10242 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
10244 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
10245 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
10247 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
10250 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
10251 NaN inputs correctly.
10253 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
10256 * config/rs6000/vsx.md (??r): New mode attribute.
10257 (*vsx_mov<mode>_64bit): Use it.
10258 (*vsx_mov<mode>_32bit): Likewise.
10260 2018-03-30 Martin Sebor <msebor@redhat.com>
10262 PR tree-optimization/84818
10263 * builtins.c (check_access): Use warning_n.
10265 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10268 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
10270 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
10273 2018-03-30 Julia Koval <julia.koval@intel.com>
10276 * x86-tune.def (movx, partial_reg_dependency): Enable for
10279 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
10281 PR inline-asm/84985
10282 * lra-constraints.c (process_alt_operands): Move setting
10283 this_alternative_matches below.
10285 2018-03-29 Martin Liska <mliska@suse.cz>
10288 * doc/invoke.texi: Document how LTO works with debug info.
10289 Describe auto-load support of binutils. Mention 'x86-64'
10290 as valid option value of -march option.
10292 2018-03-29 Jakub Jelinek <jakub@redhat.com>
10294 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
10297 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
10298 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
10299 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
10302 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
10305 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
10306 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
10307 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
10308 for RS6000_BTM_POWERPC64.
10309 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
10310 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
10311 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
10314 (DIVDEU): Likewise.
10316 2018-03-28 Carl Love <cel@us.ibm.com>
10319 2017-09-27 Carl Love <cel@us.ibm.com>
10321 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
10322 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
10323 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
10326 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10328 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
10329 instead of __vector bool.
10330 (_mm_max_pu8): Likewise.
10331 (_mm_min_pi16): Likewise.
10333 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
10336 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
10337 (DIVWEUO): Likewise.
10338 (DIVDEO): Likewise.
10339 (DIVDEUO): Likewise.
10340 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
10341 DIVWEUO and DIVDEUO.
10342 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
10343 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
10344 (div_extend): Likewise.
10345 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
10347 (__builtin_divweuo): Likewise.
10348 (__builtin_divdeo): Likewise.
10349 (__builtin_divdeuo): Likewise.
10351 2018-03-28 Jakub Jelinek <jakub@redhat.com>
10354 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
10355 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
10357 PR tree-optimization/82004
10358 * gimple-match-head.c (optimize_pow_to_exp): New function.
10359 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
10360 Don't fold to exp if optimize_pow_to_exp is false.
10362 2018-03-28 Martin Liska <mliska@suse.cz>
10365 * calls.c (initialize_argument_information): Fix trailing space.
10366 * common.opt: Fix typo and provide better explanation for
10367 -fsanitize-coverage option.
10368 * config/i386/i386.opt: Fix typo.
10370 2018-03-28 Jakub Jelinek <jakub@redhat.com>
10371 Martin Liska <mliska@suse.cz>
10374 * gimplify.c (asan_poison_variable): Don't do the check for
10375 gimplify_omp_ctxp here.
10376 (gimplify_decl_expr): Do it here.
10377 (gimplify_target_expr): Likewise.
10379 2018-03-28 Martin Liska <mliska@suse.cz>
10382 * config/i386/i386.c (ix86_function_arg_advance): Do not call
10383 chkp_type_bounds_count if MPX is not enabled.
10385 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
10387 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
10389 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10392 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
10393 function to create the function decl for complex long double
10394 multiply and divide for -mabi=ieeelongdouble.
10395 (init_float128_ieee): Call it.
10397 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
10400 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
10401 -fcf-protection=branch -mibt.
10402 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
10404 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10407 * config/arm/arm.c (arm_valid_symbolic_address): Handle
10408 arm_word_relocations.
10410 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
10413 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
10414 extern array declarations.
10416 2018-03-27 Richard Biener <rguenther@suse.de>
10418 PR middle-end/84067
10419 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
10420 explicit single_use checks.
10422 2018-03-27 Richard Biener <rguenther@suse.de>
10424 PR tree-optimization/85082
10425 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
10428 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10430 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
10431 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
10432 Turn on fasynchronous-unwind-tables and funwind-tables.
10434 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
10437 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
10438 (*bmi_blsr_<mode>_ccz): Ditto.
10440 2018-03-26 Tom de Vries <tom@codesourcery.com>
10442 PR tree-optimization/85063
10443 * omp-general.c (offloading_function_p): New function. Factor out
10445 * omp-offload.c (pass_omp_target_link::gate): ... here.
10446 * omp-general.h (offloading_function_p): Declare.
10447 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
10448 with attribute omp declare target for offloading functions.
10450 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
10452 PR tree-optimization/84005
10453 * tree-data-ref.h (get_base_for_alignment): Declare.
10454 * tree-data-ref.c (get_base_for_alignment_1): New function.
10455 (get_base_for_alignment): Likewise.
10456 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
10457 get_base_for_alignment to find a suitable base object, instead
10458 of always using drb->base_address.
10460 2018-03-23 Jakub Jelinek <jakub@redhat.com>
10462 PR inline-asm/85022
10463 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
10464 known size by default.
10466 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
10468 PR inline-asm/85030
10469 * lra-constraints.c (process_alt_operands): Don't match BLKmode
10470 and non BLKmode operands.
10472 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10475 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
10476 Clean up attributes.
10478 2018-03-23 Richard Biener <rguenther@suse.de>
10481 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
10482 we are going to emit early debug for LTO.
10484 2018-03-23 Jakub Jelinek <jakub@redhat.com>
10486 PR inline-asm/85034
10487 * function.c (match_asm_constraints_1): Don't optimize if input
10488 doesn't satisfy general_operand predicate for output's mode.
10490 PR inline-asm/85022
10491 * alias.c (write_dependence_p): Don't require for x_canonicalized
10492 non-VOIDmode if x has VOIDmode.
10495 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
10496 just don't try to optimize it rather than assert it never happens.
10498 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
10500 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
10501 macro expansions for definition of ST_INTERNAL_<mode> and
10502 LD_INTERNAL_<mode> builtins.
10503 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
10505 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
10507 (altivec_expand_st_builtin): Likewise.
10508 (altivec_expand_builtin): Remove calls to deleted functions.
10509 (rs6000_address_for_altivec): Delete this function.
10510 * config/rs6000/vector.md: Remove expands for
10511 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
10513 2018-03-22 Sudakshina Das <sudi.das@arm.com>
10516 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
10517 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
10518 re-computing once computed.
10519 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
10520 (arm_init_machine_status): Initialize
10521 machine->static_chain_stack_bytes.
10523 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
10526 * doc/extend.texi: Add four new prototypes for vec_ld.
10527 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
10528 definitions for more logical presentation.
10529 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
10530 entries for V1TI variants of __builtin_altivec_ld builtin.
10531 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
10532 handling of V1TI variant of LVX icode pattern.
10533 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
10534 (rs6000_gimple_fold_builtin): Likewise.
10535 (altivec_init_builtins): Add code to define
10536 __builtin_altivec_lvx_v1ti function.
10538 2018-03-22 Jakub Jelinek <jakub@redhat.com>
10540 PR inline-asm/84941
10541 * function.c (match_asm_constraints_1): Don't do the optimization
10542 if input isn't a REG, SUBREG, MEM or constant.
10544 2018-03-22 Tom de Vries <tom@codesourcery.com>
10546 PR tree-optimization/84956
10547 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
10548 bb_has_abnormal_pred.
10550 2018-03-22 Jakub Jelinek <jakub@redhat.com>
10553 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
10554 DECL_INITIAL (decl) to decl at the end.
10555 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
10556 adjust the comment.
10558 2018-03-21 Joseph Myers <joseph@codesourcery.com>
10560 * doc/extend.texi (__builtin_tgmath): Document when complex
10561 integer types are treated as _Complex _Float64.
10563 2018-03-21 Tom de Vries <tom@codesourcery.com>
10565 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
10567 2018-03-21 Jakub Jelinek <jakub@redhat.com>
10569 PR tree-optimization/84960
10570 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
10571 if it is ENTRY block, move them into single succ of ENTRY in that case.
10573 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
10575 PR tree-optimization/84811
10576 * poly-int.h (poly_span_traits): Remove the T3 parameter and
10577 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
10578 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
10579 (known_subrange_p): Update accordingly. Cast each value involved
10580 in the size comparison, rather than casting the result of the
10583 2018-03-21 Jakub Jelinek <jakub@redhat.com>
10585 PR tree-optimization/84982
10586 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
10587 by flipping the least significant bit rather than all bits from
10588 bitpos to bitpos + bitsize - 1.
10590 2018-03-21 Nathan Sidwell <nathan@acm.org>
10592 * doc/extend.texi (Deprecated Features): Remove mention of
10593 long-deleted deprecations.
10595 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10598 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
10599 * configure: Regenerate.
10601 2018-03-21 Tom de Vries <tom@codesourcery.com>
10603 PR tree-optimization/83126
10604 * tree-parloops.c (num_phis): New function.
10605 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
10607 2018-03-21 Nathan Sidwell <nathan@acm.org>
10609 * doc/extend.texi (Deprecated Features): Update deprecated flags,
10610 mention anon-struct/union members and trailing attributes.
10612 2018-03-21 Bin Cheng <bin.cheng@arm.com>
10614 PR tree-optimization/84969
10615 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
10616 builtin memset partitions if they set different rhs values.
10618 2018-03-21 Jakub Jelinek <jakub@redhat.com>
10620 PR rtl-optimization/84989
10621 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
10622 VEC_DUPLICATE with scalar result mode.
10624 2018-03-21 Martin Liska <mliska@suse.cz>
10627 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
10628 not intended return statement.
10630 2018-03-21 Martin Liska <mliska@suse.cz>
10633 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
10634 (chkp_find_bound_slots_1): Limit number of iterations.
10636 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
10639 * Minor grammar fixes for x86 options.
10641 2018-03-20 Jakub Jelinek <jakub@redhat.com>
10644 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
10645 holding REG_CFA_RESTORE notes, instead turn them into a USE.
10647 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
10650 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
10651 (altivec_lvx_<mode>_1op): Likewise.
10652 (altivec_stvx_<mode>_2op): Likewise.
10653 (altivec_stvx_<mode>_1op): Likewise.
10654 (altivec_lvx_<VM2:mode>): New define_expand.
10655 (altivec_stvx_<VM2:mode>): Likewise.
10656 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
10657 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
10658 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
10659 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
10660 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
10661 (rs6000_gen_lvx): Likewise.
10662 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
10663 (altivec_expand_stv_builtin): Likewise.
10664 (altivec_expand_builtin): Likewise.
10665 * config/rs6000/vector.md: Likewise.
10667 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10670 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
10673 2018-03-20 Richard Biener <rguenther@suse.de>
10676 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
10677 sign-conversions as zero, fall back to standard scalar_stmt
10680 2018-03-20 Martin Liska <mliska@suse.cz>
10683 * predict.c (rebuild_frequencies): Handle case when we have
10684 PROFILE_ABSENT, but flag_guess_branch_prob is false.
10686 2018-03-20 Jakub Jelinek <jakub@redhat.com>
10689 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
10690 flag_section_anchors.
10691 * varasm.c (use_blocks_for_decl_p): Remove hack for
10692 dw2_force_const_mem.
10695 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
10697 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
10698 be created, use lowpart_subreg of operands[0] rather than operands[0]
10700 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
10701 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
10702 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
10703 and n constraint instead of aarch64_shift_imm_di and Usd.
10704 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
10705 (*aarch64_<optab>_reg_minus<mode>3): ... this.
10707 2018-03-20 Sudakshina Das <sudi.das@arm.com>
10710 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
10711 to favor GPR over NEON registers.
10712 (<shift>di3_neon): Likewise.
10714 2018-03-20 Tom de Vries <tom@codesourcery.com>
10717 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
10718 (nvptx_process_pars): Emit bar.sync asap and alap.
10720 2018-03-20 Tom de Vries <tom@codesourcery.com>
10723 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
10724 seen_label if seen_label is already set.
10726 2018-03-20 Jakub Jelinek <jakub@redhat.com>
10729 * config/i386/i386.c (fold_builtin_cpu): For features above 31
10730 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
10731 Use 1U instead of 1. Formatting fixes.
10734 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
10735 instead of TREE_TYPE (s1) for the return value.
10737 2018-03-19 Jakub Jelinek <jakub@redhat.com>
10739 PR tree-optimization/84946
10740 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
10741 bitsize + bitsize in poly_uint64 rather than poly_int64.
10744 * dwarf2asm.c: Include fold-const.c.
10745 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
10746 of decl rather than decl itself.
10748 PR rtl-optimization/84643
10749 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
10751 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
10754 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
10755 calling assemble_variable.
10757 2018-03-19 Sudakshina Das <sudi.das@arm.com>
10760 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
10761 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
10763 2018-03-19 Jim Wilson <jimw@sifive.com>
10766 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
10767 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
10768 (riscv_first_stack_step): Likewise.
10769 (riscv_option_override): Use STACK_BOUNDARY instead of
10770 MIN_STACK_BOUNDARY.
10771 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
10772 MIN_STACK_BOUNDARY.
10773 (BIGGEST_ALIGNMENT): Set to 128.
10774 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
10775 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
10778 2018-03-19 Richard Biener <rguenther@suse.de>
10780 PR tree-optimization/84933
10781 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
10782 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
10784 2018-03-19 Richard Biener <rguenther@suse.de>
10786 PR tree-optimization/84859
10787 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
10788 (cond_if_else_store_replacement): Perform sinking operation on
10789 single-store BBs regardless of MAX_STORES_TO_SINK setting.
10790 Generalize what a BB with a single eligible store is.
10792 2018-03-19 Richard Biener <rguenther@suse.de>
10794 PR tree-optimization/84929
10795 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
10796 chrec_is_positive against non-chrec arg.
10798 2018-03-19 Tamar Christina <tamar.christina@arm.com>
10801 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
10803 2018-03-18 Martin Liska <mliska@suse.cz>
10805 PR rtl-optimization/84635
10806 * regrename.c (build_def_use): Use matches_mode only when
10809 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
10811 PR tree-optimization/84913
10812 * tree-vect-loop.c (vectorizable_reduction): Don't try to
10813 vectorize chains of COND_EXPRs.
10815 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
10817 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
10819 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
10821 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
10823 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
10825 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
10827 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
10828 Kito Cheng <kito.cheng@gmail.com>
10830 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
10831 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
10832 (nds32_adjust_reg_alloc_order): New function.
10833 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
10835 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
10837 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
10838 nds32_print_operand, nds32_print_operand_address): Use
10839 HOST_WIDE_INT_PRINT_DEC instead.
10841 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
10843 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
10845 2018-03-17 Jakub Jelinek <jakub@redhat.com>
10848 * config/i386/i386.c (initial_ix86_tune_features,
10849 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
10850 unsigned long long.
10851 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
10852 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
10853 rather than 1u << ix86_tune. Formatting fix.
10854 (ix86_option_override_internal): Change ix86_arch_mask from
10855 unsigned int to unsigned HOST_WIDE_INT, initialize to
10856 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
10857 (ix86_function_specific_restore): Likewise.
10859 2018-03-16 Jakub Jelinek <jakub@redhat.com>
10862 * postreload.c (reload_combine_recognize_pattern): Perform
10863 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
10864 truncate_int_for_mode the result for the destination's mode.
10867 * hsa-gen.c (mem_type_for_type): Fix comment typo.
10868 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
10870 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
10873 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
10876 * lra-assigns.c (lra_split_hard_reg_for): Don't use
10877 regno_allocno_class_array and sorted_pseudos.
10878 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
10879 insns where regno is used.
10881 2018-03-16 Martin Liska <mliska@suse.cz>
10884 * multiple_target.c (create_dispatcher_calls): Redirect
10885 reference in the symbol table.
10887 2018-03-16 Martin Liska <mliska@suse.cz>
10890 * multiple_target.c (create_dispatcher_calls): Redirect also
10893 2018-03-16 Jakub Jelinek <jakub@redhat.com>
10897 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
10898 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
10899 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
10901 2018-03-16 Julia Koval <julia.koval@intel.com>
10903 * doc/invoke.texi (Skylake Server): Add CLWB.
10904 Cannonlake): Remove CLWB.
10906 2018-03-16 Jakub Jelinek <jakub@redhat.com>
10908 PR tree-optimization/84841
10909 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
10911 (FLOAT_ONE_CONST_TYPE): Define.
10912 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
10913 (sort_by_operand_rank): Put entries with higher constant_type last
10914 rather than first to match comments.
10916 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
10918 * config/nios2/nios2.md (movsi_internal): Fix thinko in
10921 2018-03-15 Jakub Jelinek <jakub@redhat.com>
10924 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
10925 check and use address of target always.
10927 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
10930 * config/i386/i386.c (indirect_thunk_needed): Update comments.
10931 (indirect_thunk_bnd_needed): Likewise.
10932 (indirect_thunks_used): Likewise.
10933 (indirect_thunks_bnd_used): Likewise.
10934 (indirect_return_needed): New.
10935 (indirect_return_bnd_needed): Likewise.
10936 (output_indirect_thunk_function): Add a bool argument for
10938 (output_indirect_thunk_function): Don't generate alias for
10939 function return thunk.
10940 (ix86_code_end): Call output_indirect_thunk_function to generate
10941 function return thunks.
10942 (ix86_output_function_return): Set indirect_return_bnd_needed
10943 and indirect_return_needed instead of indirect_thunk_bnd_needed
10944 and indirect_thunk_needed.
10946 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
10948 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
10949 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
10950 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
10952 2018-03-15 David Malcolm <dmalcolm@redhat.com>
10953 Paul Hua <paul.hua.gm@gmail.com>
10956 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
10958 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
10960 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
10961 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
10962 resp. SFmode cases.
10964 2018-03-15 Tamar Christina <tamar.christina@arm.com>
10967 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
10968 instead of GET_MODE_SIZE when comparing Units.
10970 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
10973 * varasm.c (hash_section): Return an unchangeble hash value
10974 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
10975 Return !aarch64_can_use_per_function_literal_pools_p ().
10977 2018-03-15 Jakub Jelinek <jakub@redhat.com>
10980 * optabs.c (emit_conditional_move): Pass address of cmode's copy
10981 rather than address of cmode as last argument to prepare_cmp_insn.
10983 2018-03-15 Julia Koval <julia.koval@intel.com>
10985 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
10986 F_AVX512VNNI, F_AVX512BITALG): New.
10988 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
10991 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
10992 insn for floating-point loads and stores.
10994 2018-03-14 Carl Love <cel@us.ibm.com>
10996 * config/rs6000/rs6000-c.c: Add macro definitions for
10997 ALTIVEC_BUILTIN_VEC_PERMXOR.
10998 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
10999 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
11000 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
11001 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
11003 * config/doc/extend.texi: Add prototypes for vec_permxor.
11005 2018-03-14 David Malcolm <dmalcolm@redhat.com>
11008 * diagnostic-show-locus.c (class layout_point): Convert m_line
11009 from int to linenum_type.
11010 (line_span::comparator): Use linenum "compare" function when
11011 comparing line numbers.
11012 (test_line_span): New function.
11013 (layout_range::contains_point): Convert param "row" from int to
11015 (layout_range::intersects_line_p): Likewise.
11016 (layout::will_show_line_p): Likewise.
11017 (layout::print_source_line): Likewise.
11018 (layout::should_print_annotation_line_p): Likewise.
11019 (layout::print_annotation_line): Likewise.
11020 (layout::print_leading_fixits): Likewise.
11021 (layout::annotation_line_showed_range_p): Likewise.
11022 (struct line_corrections): Likewise for field m_row.
11023 (line_corrections::line_corrections): Likewise for param "row".
11024 (layout::print_trailing_fixits): Likewise.
11025 (layout::get_state_at_point): Likewise.
11026 (layout::get_x_bound_for_row): Likewise.
11027 (layout::print_line): Likewise.
11028 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
11029 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
11030 * input.c (selftest::test_linenum_comparisons): New function.
11031 (selftest::input_c_tests): Call it.
11032 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
11033 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
11034 * selftest.h (ASSERT_GT): New macro.
11035 (ASSERT_GT_AT): New macro.
11036 (ASSERT_LT): New macro.
11037 (ASSERT_LT_AT): New macro.
11039 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
11041 PR rtl-optimization/84780
11042 * combine.c (distribute_links): Don't make a link based on pc_rtx.
11044 2018-03-14 Martin Liska <mliska@suse.cz>
11046 * tree.c (record_node_allocation_statistics): Use
11047 get_stats_node_kind.
11048 (get_stats_node_kind): New function extracted from
11049 record_node_allocation_statistics.
11050 (free_node): Use get_stats_node_kind.
11052 2018-03-14 Richard Biener <rguenther@suse.de>
11054 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
11055 that the value-set of ANTIC_IN doesn't grow.
11058 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
11060 (BB_VISITED_WITH_VISITED_SUCCS): New define.
11061 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
11063 2018-03-14 Julia Koval <julia.koval@intel.com>
11065 * config.gcc (icelake-client, icelake-server): New.
11067 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
11068 (initial_ix86_arch_features): Ditto.
11069 (PTA_SKYLAKE): Add SGX.
11070 (PTA_ICELAKE): Remove.
11071 (PTA_ICELAKE_CLIENT): New.
11072 (PTA_ICELAKE_SERVER): New.
11073 (ix86_option_override_internal): Split up icelake on icelake client and
11075 (get_builtin_code_for_version): Ditto.
11076 (fold_builtin_cpu): Ditto.
11077 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
11078 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
11079 * config/i386/i386.h (processor_type): Ditto.
11080 * doc/invoke.texi: Ditto.
11082 2018-03-14 Jakub Jelinek <jakub@redhat.com>
11085 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
11086 INTEGER_CST offset, add it together with bitpos / 8 and
11087 sign extend based on POINTER_SIZE.
11091 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
11094 * config/i386/constraints.md (Yc): New register constraint.
11095 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
11096 Use Yc constraint for alternative 2 of operand 0. Remove
11097 preferred_for_speed attribute.
11099 2018-03-14 Richard Biener <rguenther@suse.de>
11101 PR tree-optimization/84830
11102 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
11103 with the old one to avoid oscillations.
11105 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
11108 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
11110 (assign_by_spills): Return a flag of reload assignment failure.
11111 Do not process the reload assignment failures. Do not spill other
11112 reload pseudos if they has the same reg class. Update n if
11114 (lra_assign): Add a return arg. Set up from the result of
11115 assign_by_spills call.
11116 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
11117 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
11118 usage_insns if it is not NULL.
11119 (spill_hard_reg_in_range): New function.
11120 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
11121 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
11122 function prototypes.
11123 (lra_assign): Change prototype.
11124 * lra.c (lra): Add code to deal with fails by splitting hard reg
11127 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
11129 * config/riscv/riscv.opt (mrelax): New option.
11130 * config/riscv/riscv.c (riscv_file_start): Emit ".option
11131 "norelax" when riscv_mrelax is disabled.
11132 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
11134 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11137 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
11138 reassociation for int modes.
11140 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
11142 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
11143 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
11145 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
11146 * config/aarch64/aarch64-sve.md
11147 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
11148 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
11149 (*extend<mode><Vwide>2): Rename to...
11150 (aarch64_sve_extend<mode><Vwide>2): ...this.
11151 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
11152 renaming the old pattern to...
11153 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
11155 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
11156 define_expand, renaming the old pattern to...
11157 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
11158 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
11159 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
11160 account when deciding which SVE instruction the optab should use.
11161 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
11163 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
11165 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
11166 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
11167 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
11168 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
11169 (tlsdesc_small_<mode>): Turn a define_expand and use
11170 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
11171 (tlsdesc_small_advsimd_<mode>): ...this.
11172 (tlsdesc_small_sve_<mode>): New pattern.
11174 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
11176 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
11177 (UNSPEC_UMUL_HIGHPART): New constants.
11178 (MUL_HIGHPART): New int iteraor.
11179 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
11180 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
11182 (*<su>mul<mode>3_highpart): New define_insn.
11184 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
11187 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
11190 2018-03-13 Martin Liska <mliska@suse.cz>
11193 * (sem_item_optimizer::sem_item_optimizer): Initialize new
11195 (sem_item_optimizer::~sem_item_optimizer): Release it.
11196 (sem_item_optimizer::merge_classes): Register variable aliases.
11197 (sem_item_optimizer::fixup_pt_set): New function.
11198 (sem_item_optimizer::fixup_points_to_sets): Likewise.
11199 * ipa-icf.h: Declare new variables and functions.
11201 2018-03-13 Jakub Jelinek <jakub@redhat.com>
11203 PR middle-end/84834
11204 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
11205 integer_pow2p@2 and test integer_pow2p in condition.
11206 (A < 0 ? C : 0): Similarly for @1.
11208 PR middle-end/84831
11209 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
11210 characters starting at p contain '\0' character, don't look beyond
11214 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
11215 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
11218 * reg-stack.c (change_stack): Change update_end var from int to
11219 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
11220 also call set_block_for_insn on the newly added insns and rescan.
11223 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
11224 on the last operand.
11227 * tree.c (stabilize_reference_1): Return save_expr (e) for
11228 STATEMENT_LIST even if it doesn't have side-effects.
11230 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
11232 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
11234 2018-03-12 Renlin Li <renlin.li@arm.com>
11236 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
11237 aarch64_output_scalar_simd_mov_immediate.
11239 2018-03-12 Martin Sebor <msebor@redhat.com>
11241 PR tree-optimization/83456
11242 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
11243 for perfectly overlapping calls to memcpy.
11244 (gimple_fold_builtin_memory_chk): Same.
11245 (gimple_fold_builtin_strcpy): Handle no-warning.
11246 (gimple_fold_builtin_stxcpy_chk): Same.
11247 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
11249 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
11251 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
11252 parameter. Use it for SFmode.
11253 (rs6000_function_arg_advance_1): Adjust.
11254 (rs6000_function_arg): Adjust.
11255 (rs6000_gimplify_va_arg): Pass false for that new parameter.
11257 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
11259 PR rtl-optimization/84169
11260 PR rtl-optimization/84780
11261 * combine.c (can_combine_p): Check for a 2-insn combination whether
11262 the destination register is used between the two insns, too.
11264 2018-03-12 Richard Biener <rguenther@suse.de>
11266 PR tree-optimization/84803
11267 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
11268 for refs DR analysis didn't process.
11270 2018-03-12 Richard Biener <rguenther@suse.de>
11272 PR tree-optimization/84777
11273 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
11274 force-vectorize loops ignore whether we are optimizing for size.
11276 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
11278 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
11279 (TARGET_MD_ASM_ADJUST): Define.
11281 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
11282 Kito Cheng <kito.cheng@gmail.com>
11283 Chung-Ju Wu <jasonwucj@gmail.com>
11285 * config/nds32/nds32.c (nds32_compute_stack_frame,
11286 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
11287 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
11288 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
11289 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
11290 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
11291 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
11292 * config/nds32/nds32.md (prologue, epilogue): Use macro
11293 NDS32_V3PUSH_AVAILABLE_P to do checking.
11295 2018-03-11 Jakub Jelinek <jakub@redhat.com>
11298 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
11299 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
11300 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
11301 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
11302 addition of most attributes on !orig_type_die or the attribute not
11303 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
11305 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
11306 Chung-Ju Wu <jasonwucj@gmail.com>
11308 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
11309 __NDS32_VH__ macro.
11310 * config/nds32/nds32.opt (mvh): New option.
11312 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
11313 Chung-Ju Wu <jasonwucj@gmail.com>
11315 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
11317 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
11318 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
11321 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
11322 Chung-Ju Wu <jasonwucj@gmail.com>
11324 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
11326 * config/nds32/nds32-multiple.md (strlensi): New pattern.
11327 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
11329 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
11330 Kito Cheng <kito.cheng@gmail.com>
11331 Chung-Ju Wu <jasonwucj@gmail.com>
11333 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
11334 UNSPEC_FFMISM and UNSPEC_FLMISM.
11335 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
11336 for ffb, ffmism and flmism.
11337 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
11338 (unspec_ffmism): Ditto.
11339 (unspec_flmism): Ditto.
11340 (nds32_expand_builtin_impl): Check if string extension is available.
11341 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
11342 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
11344 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
11347 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
11350 * lra-assigns.c (assign_by_spills): Return a flag of reload
11351 assignment failure. Do not process the reload assignment
11352 failures. Do not spill other reload pseudos if they has the same
11354 (lra_assign): Add a return arg. Set up from the result of
11355 assign_by_spills call.
11356 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
11357 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
11358 usage_insns if it is not NULL.
11359 (spill_hard_reg_in_range): New function.
11360 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
11361 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
11362 function prototypes.
11363 (lra_assign): Change prototype.
11364 * lra.c (lra): Add code to deal with fails by splitting hard reg
11367 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
11370 * config/i386/i386.opt: Replace Enforcment with Enforcement.
11372 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
11375 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
11376 (dw_val_node): Add val_symbolic_view.
11377 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
11378 (symview_upper_bound): New.
11379 (new_line_info_table): Initialize symviews_since_reset.
11380 (dwarf2out_source_line): Count symviews_since_reset and set
11381 symview_upper_bound.
11382 (dw_val_equal_p): Handle symview.
11383 (add_AT_symview): New.
11384 (print_dw_val): Handle symview.
11385 (attr_checksum, attr_checksum_ordered): Likewise.
11386 (same_dw_val_p, size_of_die): Likewise.
11387 (value_format, output_die): Likewise.
11388 (add_high_low_attributes): Use add_AT_symview for entry_view.
11389 (dwarf2out_finish): Reset symview_upper_bound, clear
11392 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
11395 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
11396 Add strict argument and use it.
11397 (rs6000_split_multireg_move): Update for new strict argument.
11398 (mem_operand_gpr): Disallow all non-offsettable addresses.
11399 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
11401 2018-03-09 Jakub Jelinek <jakub@redhat.com>
11404 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
11405 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
11406 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
11409 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
11410 decl, use remap_type if we want to use the type.
11412 2018-03-09 Martin Sebor <msebor@redhat.com>
11414 PR tree-optimization/84526
11415 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
11417 (builtin_access::generic_overlap): Be prepared to handle non-array
11420 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
11422 PR rtl-optimization/84682
11423 * lra-constraints.c (process_address_1): Check is_address flag
11424 for address constraints.
11425 (process_alt_operands): Likewise.
11426 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
11427 preprocess_constraints.
11428 * recog.h (preprocess_constraints): Add oploc parameter.
11430 * recog.c (preprocess_constraints): Test address_operand for
11431 CT_ADDRESS constraints.
11433 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
11436 * lra-assigns.c (assign_by_spills): Return a flag of reload
11437 assignment failure. Do not process the reload assignment
11438 failures. Do not spill other reload pseudos if they has the same
11440 (lra_assign): Add a return arg. Set up from the result of
11441 assign_by_spills call.
11442 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
11443 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
11444 usage_insns if it is not NULL.
11445 (spill_hard_reg_in_range): New function.
11446 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
11447 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
11448 function prototypes.
11449 (lra_assign): Change prototype.
11450 * lra.c (lra): Add code to deal with fails by splitting hard reg
11453 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11456 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
11457 Accept complain bool parameter. Only emit errors if it is true.
11458 (arm_parse_cpu_option_name): Likewise.
11459 (arm_target_thumb_only): Adjust callers of the above.
11460 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
11461 prototype to take a default true bool parameter.
11462 (arm_parse_arch_option_name): Likewise.
11464 2018-03-09 David Malcolm <dmalcolm@redhat.com>
11465 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
11469 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
11470 * configure: Regenerate.
11471 * configure.ac ("linker --version-script option"): New.
11472 ("linker soname option"): New.
11474 2018-03-09 Richard Biener <rguenther@suse.de>
11476 PR tree-optimization/84775
11477 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
11478 immediate uses of predicate stmts and mark them modified.
11481 PR tree-optimization/84178
11482 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
11484 (version_loop_for_if_conversion): Delay update_ssa call.
11485 (tree_if_conversion): Delay update_ssa until after predicate
11488 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
11491 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
11492 when the function accesses prior frames.
11494 2018-03-08 Jakub Jelinek <jakub@redhat.com>
11497 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
11498 gen_llsym, otherwise call maybe_gen_llsym.
11500 PR inline-asm/84742
11501 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
11502 has ',' character inside of it.
11504 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11507 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
11508 as clobbering CC_REGNUM.
11510 2018-03-08 Richard Biener <rguenther@suse.de>
11512 PR middle-end/84552
11513 * tree-scalar-evolution.c: Include tree-into-ssa.h.
11514 (follow_copies_to_constant): Do not follow SSA names registered
11517 2018-03-08 Richard Biener <rguenther@suse.de>
11519 PR tree-optimization/84178
11520 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
11522 (version_loop_for_if_conversion): Delay update_ssa call.
11523 (tree_if_conversion): Delay update_ssa until after predicate
11526 2018-03-08 David Malcolm <dmalcolm@redhat.com>
11528 PR tree-optimization/84178
11529 * tree-if-conv.c (release_bb_predicate): Remove the
11530 the assertion that the stmts have NULL use_ops.
11531 Discard the statements, asserting that they haven't
11532 yet been added to a BB.
11534 2018-03-08 Richard Biener <rguenther@suse.de>
11536 PR tree-optimization/84746
11537 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
11538 (phi_translate): Pass in destination ANTIC_OUT set.
11539 (phi_translate_1): Likewise. For a simplified result lookup
11540 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
11541 (phi_translate_set): Adjust.
11542 (do_pre_regular_insertion): Likewise.
11543 (do_pre_partial_partial_insertion): Likewise.
11545 2018-03-08 Martin Liska <mliska@suse.cz>
11547 PR gcov-profile/84735
11548 * doc/gcov.texi: Document usage of profile files.
11549 * gcov-io.h: Document changes in the format.
11551 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
11555 * dwarf2out.c (struct dw_line_info_table): Update comments for
11557 (FORCE_RESET_NEXT_VIEW): New.
11558 (FORCE_RESETTING_VIEW_P): New.
11559 (RESETTING_VIEW_P): Check for -1 too.
11560 (ZERO_VIEW_P): Likewise.
11561 (new_line_info_table): Force-reset next view.
11562 (dwarf2out_begin_function): Likewise.
11563 (dwarf2out_source_line): Simplify zero_view_p initialization.
11564 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
11565 view directly. Omit view when omitting .loc at line 0.
11567 2018-03-08 Jakub Jelinek <jakub@redhat.com>
11569 PR tree-optimization/84740
11570 * tree-switch-conversion.c (process_switch): Call build_constructors
11571 only if info.phi_count is non-zero.
11573 PR tree-optimization/84739
11574 * tree-tailcall.c (find_tail_calls): Check call arguments against
11575 DECL_ARGUMENTS (current_function_decl) rather than
11576 DECL_ARGUMENTS (func) when checking for tail recursion.
11578 2018-03-07 Jakub Jelinek <jakub@redhat.com>
11580 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
11581 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
11582 Volker Reichelt's entry and add entries for people that perform
11583 GCC fuzzy testing and report numerous bugs.
11585 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
11588 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
11589 readonly data in sdata, if that is disabled.
11590 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
11591 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
11592 -mreadonly-in-sdata option.
11594 2018-03-07 Martin Sebor <msebor@redhat.com>
11596 PR tree-optimization/84468
11597 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
11598 basic block when looking for nul assignment.
11600 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
11603 * except.h (output_function_exception_table): Adjust prototype.
11604 * except.c (output_function_exception_table): Remove FNNAME parameter
11605 and add SECTION parameter. Ouput one part of the table at a time.
11606 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
11607 the first part of the exception table and emit unwind directives.
11608 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
11609 (i386_pe_seh_cold_init): Likewise.
11610 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
11611 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
11612 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
11613 (ix86_output_call_insn): Emit a nop in one more case for SEH.
11614 * config/i386/winnt.c: Include except.h.
11615 (struct seh_frame_state): Add reg_offset, after_prologue and
11616 in_cold_section fields.
11617 (i386_pe_seh_end_prologue): Set seh->after_prologue.
11618 (i386_pe_seh_cold_init): New function.
11619 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
11620 to seh->in_cold_section.
11621 (seh_emit_push): Record the offset of the push.
11622 (seh_emit_save): Record the offet of the save.
11623 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
11624 Test seh->after_prologue to disregard the epilogue.
11625 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
11626 (i386_pe_end_cold_function): New function.
11628 2018-03-07 Jakub Jelinek <jakub@redhat.com>
11631 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
11632 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
11635 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
11637 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
11638 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
11640 PR middle-end/84723
11641 * multiple_target.c: Include tree-inline.h and intl.h.
11642 (expand_target_clones): Diagnose and fail if node->definition and
11643 !tree_versionable_function_p (node->decl).
11645 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
11647 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
11649 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
11650 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11651 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
11653 2018-03-06 Jakub Jelinek <jakub@redhat.com>
11656 * combine.c (try_combine): Use reg_or_subregno instead of handling
11657 just paradoxical SUBREGs and REGs.
11659 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
11661 * config/arc/arc.c (arc_finalize_pic): Remove function.
11662 (arc_must_save_register): We use single base PIC register, remove
11663 checks to save/restore the PIC register.
11664 (arc_expand_prologue): Likewise.
11665 * config/arc/arc-protos.h (arc_set_default_type_attributes):
11667 (arc_verify_short): Likewise.
11668 (arc_attr_type): Likewise.
11669 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
11670 (walk_stores): Likewise.
11671 (arc_address_cost): Make it static.
11672 (arc_verify_short): Likewise.
11673 (branch_dest): Likewise.
11674 (arc_attr_type): Likewise.
11675 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
11676 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
11677 (arc_final_prescan_insn): Remove inserting the nops due to
11678 hardware hazards. It is done in reorg step.
11679 (insn_length_variant_t): Remove.
11680 (insn_length_parameters_t): Likewise.
11681 (arc_insn_length_parameters): Likewise.
11682 (arc_get_insn_variants): Likewise.
11683 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
11685 2018-03-06 Jakub Jelinek <jakub@redhat.com>
11687 PR inline-asm/84683
11688 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
11691 PR tree-optimization/84687
11692 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
11694 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
11696 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11698 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
11699 Rename to ppc_speculation_barrier.
11700 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
11701 __builtin_ppc_speculation_barrier.
11703 2018-03-05 Jakub Jelinek <jakub@redhat.com>
11706 * combine.c (combine_simplify_rtx): Don't try to simplify if
11707 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
11710 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
11712 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
11713 to 32 bytes when compiling for POWER9.
11715 2018-03-05 Jakub Jelinek <jakub@redhat.com>
11718 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
11719 regparm >= 3 with no arg reg available also for calls with
11720 flag_force_indirect_call. Pass decl to ix86_function_regparm.
11723 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
11725 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
11727 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
11730 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
11732 2018-03-05 Richard Biener <rguenther@suse.de>
11734 PR tree-optimization/84486
11735 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
11736 When inserting a __builtin_assume_aligned call set the LHS
11737 SSA name alignment info accordingly.
11739 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
11741 PR tree-optimization/84114
11742 * config/aarch64/aarch64.c (aarch64_reassociation_width)
11743 Avoid reassociation of FLOAT_MODE addition.
11745 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
11747 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
11748 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
11749 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
11750 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
11751 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
11752 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
11753 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
11755 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
11756 __builtin_ia32_wbinvd): New builtins.
11757 (SPECIAL_ARGS2): New.
11758 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
11759 (SPECIAL_ARGS2): New.
11760 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
11761 (ix86_valid_target_attribute_inner_p): Ditto.
11762 (ix86_init_mmx_sse_builtins): Add special_args2.
11763 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
11764 TARGET_WBNOINVD_P): New.
11765 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
11766 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
11767 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
11768 * config/i386/immintrin.h (_wbinvd): New intrinsic.
11769 * config/i386/pconfigintrin.h: New file.
11770 * config/i386/wbnoinvdintrin.h: Ditto.
11771 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
11773 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
11775 2018-03-05 Richard Biener <rguenther@suse.de>
11777 PR tree-optimization/84670
11778 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
11780 (BB_VISITED_WITH_VISITED_SUCCS): New define.
11781 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
11782 (compute_antic_aux): Only assert the number of values in ANTIC_IN
11783 doesn't grow if all successors (recursively) were visited at least
11786 2018-03-05 Richard Biener <rguenther@suse.de>
11788 PR tree-optimization/84650
11789 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
11790 if executed in the loop pipeline.
11792 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
11794 * doc/configfiles.texi (Configuration Files): Move info about
11795 conditionalizing $target-protos.h to...
11796 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
11797 differs from $target-protos.h.
11799 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
11800 Chung-Ju Wu <jasonwucj@gmail.com>
11802 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
11803 * config/nds32/nds32-multiple.md (setmemsi): Define.
11804 * config/nds32/nds32-memory-manipulation.c
11805 (nds32_gen_dup_4_byte_to_word_value): New.
11806 (emit_setmem_word_loop): New.
11807 (emit_setmem_byte_loop): New.
11808 (nds32_expand_setmem_loop): New.
11809 (nds32_expand_setmem_loop_v3m): New.
11810 (nds32_expand_setmem_unroll): New.
11811 (nds32_expand_setmem): New.
11813 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
11814 Chung-Ju Wu <jasonwucj@gmail.com>
11816 * config/nds32/nds32-memory-manipulation.c
11817 (nds32_emit_load_store): New.
11818 (nds32_emit_post_inc_load_store): New.
11819 (nds32_emit_mem_move): New.
11820 (nds32_emit_mem_move_block): New.
11821 (nds32_expand_movmemsi_loop_unknown_size): New.
11822 (nds32_expand_movmemsi_loop_known_size): New.
11823 (nds32_expand_movmemsi_loop): New.
11824 (nds32_expand_movmemsi_unroll): New.
11825 (nds32_expand_movmemqi): Rename ...
11826 (nds32_expand_movmemsi): ... to this.
11827 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
11828 (movmemsi): ... to this.
11829 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
11830 (nds32_expand_movmemsi): ... to this.
11832 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
11833 Monk Chiang <sh.chiang04@gmail.com>
11834 Chung-Ju Wu <jasonwucj@gmail.com>
11836 * config/nds32/nds32-protos.h
11837 (nds32_expand_load_multiple): New arguments.
11838 (nds32_expand_store_multiple): Ditto.
11839 (nds32_valid_multiple_load_store): Rename ...
11840 (nds32_valid_multiple_load_store_p): ... to this.
11841 * config/nds32/nds32-memory-manipulation.c
11842 (nds32_expand_load_multiple): Refine implementation.
11843 (nds32_expand_store_multiple): Ditto.
11844 * config/nds32/nds32-multiple.md
11845 (load_multiple): Update nds32_expand_load_multiple interface.
11846 (store_multiple): Update nds32_expand_store_multiple interface.
11847 * config/nds32/nds32-predicates.c
11848 (nds32_valid_multiple_load_store): Rename ...
11849 (nds32_valid_multiple_load_store_p): ... to this and refine
11851 * config/nds32/predicates.md
11852 (nds32_load_multiple_and_update_address_operation): New predicate.
11853 (nds32_store_multiple_and_update_address_operation): New predicate.
11855 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
11856 Chung-Ju Wu <jasonwucj@gmail.com>
11858 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
11859 (combo): New attribute.
11860 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
11862 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
11864 * config/nds32/nds32.opt: Change -mcmodel= default value.
11866 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
11867 Monk Chiang <sh.chiang04@gmail.com>
11868 Chung-Ju Wu <jasonwucj@gmail.com>
11870 * config/nds32/constants.md (unspec_element): New enum.
11871 * config/nds32/constraints.md (Umw): New constraint.
11872 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
11873 * config/nds32/nds32-intrinsic.md: Likewise.
11874 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
11875 (nds32_valid_smw_lwm_base_p): New.
11876 (nds32_output_smw_single_word): New.
11877 (nds32_output_lmw_single_word): New.
11878 (nds32_expand_unaligned_load): New.
11879 (nds32_expand_unaligned_store): New.
11880 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
11881 (nds32_output_smw_single_word): Declare.
11882 (nds32_output_lmw_single_word): Declare.
11883 (nds32_expand_unaligned_load): Declare.
11884 (nds32_expand_unaligned_store): Declare.
11885 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
11886 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
11887 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
11888 NDS32_BUILTIN_UASTORE_DW.
11889 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
11892 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
11893 Kito Cheng <kito.cheng@gmail.com>
11894 Chung-Ju Wu <jasonwucj@gmail.com>
11896 * config/nds32/nds32-intrinsic.c
11897 (nds32_expand_builtin_null_ftype_reg): Delete.
11898 (nds32_expand_builtin_reg_ftype_imm): Ditto.
11899 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
11900 (nds32_read_argument): New.
11901 (nds32_legitimize_target): Ditto.
11902 (nds32_legitimize_argument): Ditto.
11903 (nds32_check_constant_argument): Ditto.
11904 (nds32_expand_unop_builtin): Ditto.
11905 (nds32_expand_unopimm_builtin): Ditto.
11906 (nds32_expand_binop_builtin): Ditto.
11907 (nds32_builtin_decl_impl): Ditto.
11908 (builtin_description): Ditto.
11909 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
11910 (nds32_init_builtins_impl): Ditto.
11911 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
11912 (nds32_builtin_decl): New.
11913 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
11914 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
11916 2018-03-02 Jeff Law <law@redhat.com>
11918 * reorg.c (stop_search_p): Handle DEBUG_INSN.
11919 (redundant_insn, fill_simple_delay_slots): Likewise.
11920 (fill_slots_from_thread): Likewise.
11921 * resource.c (mark_referenced_resources): Likewise.
11922 (mark_set_resources, find_dead_or_set_registers): Likewise.
11924 2018-03-02 Jakub Jelinek <jakub@redhat.com>
11926 * substring-locations.h (format_warning_va): Formatting fix for
11927 ATTRIBUTE_GCC_DIAG.
11928 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
11930 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
11931 * substring-locations.c: Include intl.h.
11932 (format_warning_va): Turned into small wrapper around
11933 format_warning_n_va, renamed to ...
11934 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
11935 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
11937 (format_warning_at_substring_n): New function.
11938 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
11939 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
11940 format_warning_at_substring with just a shorter name instead of
11941 const function pointer.
11942 (fmtwarn_n): New function.
11943 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
11944 appropriate, get rid of all the fmtstr temporaries, move conditionals
11945 with G_() wrapped string literals directly into fmtwarn arguments,
11946 cast dir.len to (int), formatting fixes.
11948 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
11950 * doc/invoke.texi: Remove "Cilk Plus" references.
11952 2018-03-02 Jakub Jelinek <jakub@redhat.com>
11953 Richard Biener <rguenther@suse.de>
11956 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
11957 for error or warning attributes if CALL_FROM_THUNK_P is set.
11960 2018-03-02 Jakub Jelinek <jakub@redhat.com>
11963 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
11964 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
11967 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
11968 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
11970 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
11971 instead of -1U in last predictors element's probability member.
11973 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
11976 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
11977 arguments if they are comparable.
11979 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
11981 PR tree-optimization/84634
11982 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
11983 masks and masked_loop_p with a single loop_masks, making sure it's
11984 null for bb vectorization.
11986 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
11988 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
11989 (vect_analyze_data_ref_access): Use loop->safe_len rather than
11990 loop->force_vectorize to check whether there is no alias.
11992 2018-03-02 Jakub Jelinek <jakub@redhat.com>
11995 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
11997 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
11999 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
12000 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
12001 instead of a loop around prev_real_insn.
12002 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
12005 PR inline-asm/84625
12006 * config/i386/i386.c (ix86_print_operand): Use conditional
12007 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
12010 2018-03-02 Richard Biener <rguenther@suse.de>
12012 PR tree-optimization/84427
12013 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
12014 (bitmap_set_subtract_values): Rewrite to handle multiple
12017 (prune_clobbered_mems): Likewise.
12018 (phi_translate): Take edge instead of pred/phiblock.
12019 (phi_translate_1): Likewise.
12020 (phi_translate_set): Likewise. Insert all translated
12021 exprs for a value into the set, keeping possibly multiple
12022 expressions per value.
12023 (compute_antic_aux): Adjust for phi_translate changes.
12024 When intersecting union the expressions and prune those
12025 not in the final value set, keeping possibly multiple
12026 expressions per value. Do not use value-insertion
12027 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
12028 all expressions. Add verification that the value-sets
12029 only shrink during iteration.
12030 (compute_partial_antic_aux): Adjust for the phi_translate changes.
12031 (do_pre_regular_insertion): Likewise.
12032 (do_pre_partial_partial_insertion): Likewise.
12034 2018-03-02 Richard Biener <rguenther@suse.de>
12037 * config/darwin.c (saved_debug_info_level): New static global.
12038 (darwin_asm_lto_start): Disable debug info generation for LTO out.
12039 (darwin_asm_lto_end): Restore debug info generation settings.
12041 2018-03-01 Martin Liska <mliska@suse.cz>
12044 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12045 volatile arguments.
12047 2018-03-01 Richard Biener <rguenther@suse.de>
12050 * dwarf2out.c (gen_variable_die): Properly handle late VLA
12051 type annotation with LTO when debug was disabled at compile-time.
12053 2018-03-01 Matthew Fortune <mfortune@gmail.com>
12055 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
12057 (mips_final_postscan_insn): Likewise.
12059 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
12061 PR rtl-optimization/84528
12062 * alias.c (init_alias_target): Add commentary.
12063 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
12064 a unique base value if the frame pointer is not eliminated
12065 to the stack pointer.
12067 2018-03-01 Tom de Vries <tom@codesourcery.com>
12069 PR rtl-optimization/83327
12070 * lra-int.h (hard_regs_spilled_into): Declare.
12071 * lra.c (hard_regs_spilled_into): Define.
12072 (init_reg_info): Init hard_regs_spilled_into.
12073 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
12074 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
12075 (process_bb_lives): Handle hard_regs_spilled_into.
12076 (lra_create_live_ranges_1): Before doing liveness propagation, clear
12077 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
12079 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
12081 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
12082 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
12083 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
12084 * config/rs6000/aix72.h: New file.
12086 2018-02-28 Jakub Jelinek <jakub@redhat.com>
12088 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
12089 instead of warning_at with conditional singular and plural messages
12093 * stor-layout.c (update_alignment_for_field): For
12094 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
12095 && !DECL_PACKED (field), do the alignment update, just use
12096 only desired_align instead of MAX (type_align, desired_align)
12098 (place_field): Don't do known_align < desired_align handling
12099 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
12100 is non-NULL, instead do it after rli->prev_field handling and
12101 only if not within a bitfield word. For DECL_PACKED (field)
12102 use type_align of BITS_PER_UNIT.
12104 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
12106 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
12107 superfluous parentheses and trailing spaces.
12109 2018-02-28 Richard Biener <rguenther@suse.de>
12111 PR tree-optimization/84584
12112 * graphite-scop-detection.c (scop_detection::add_scop): Discard
12113 SCoPs with fake exit edge.
12115 2018-02-28 Martin Liska <mliska@suse.cz>
12118 * timevar.c (timer::print): Fix format to properly print 100%
12121 2018-02-28 Richard Biener <rguenther@suse.de>
12123 PR middle-end/84607
12124 * genmatch.c (capture_info::walk_match): Do not mark
12125 captured expressions without operands as expr_p given
12126 they act more like predicates and should be subject to
12127 "lost tail" side-effect preserving.
12129 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
12131 PR rtl-optimization/81611
12132 * auto-inc-dec.c (attempt_change): Move dead note from
12133 mem_insn if it's the next use of regno
12134 (find_address): Take address use of reg holding
12135 non-incremented value. Add parm to limit search to the named
12137 (merge_in_block): Attempt to use a mem insn that is the next
12138 use of the original regno.
12140 2018-02-27 Martin Sebor <msebor@redhat.com>
12143 * gcc/doc/invoke.texi (-Wmissing-attributes): New option.
12144 * gcc/print-tree.c (print_node): Handle DECL_UNINLINABLE.
12146 2018-02-27 Martin Sebor <msebor@redhat.com>
12148 PR translation/84207
12149 * diagnostic-core.h (warning_n, error_n, inform_n): Change
12150 n argument to unsigned HOST_WIDE_INT.
12151 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
12152 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
12153 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
12154 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
12156 2018-02-27 Richard Biener <rguenther@suse.de>
12158 PR tree-optimization/84512
12159 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
12160 Do not use the estimate returned from record_stmt_cost for
12161 the scalar iteration cost but sum properly using add_stmt_cost.
12163 2018-02-27 Richard Biener <rguenther@suse.de>
12165 PR tree-optimization/84466
12166 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
12167 Adjust last change to less strictly validate use operands.
12169 2018-02-27 Martin Liska <mliska@suse.cz>
12171 PR gcov-profile/84548
12172 * gcov.c (process_file): Allow partial overlap and consider it
12173 also as group functions.
12174 (output_lines): Properly calculate range of lines for a group.
12176 2018-02-27 Martin Liska <mliska@suse.cz>
12178 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
12179 'ggc' suffixes. Change first column width.
12180 (timer::print): Fix formatting of the column.
12182 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
12184 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
12185 preserve inline entry blocks for the sake of debug inline
12186 entry point markers alone.
12187 (remove_unused_locals): Suggest in comments a better place to
12188 force the preservation of inline entry blocks that are
12189 otherwise unused, but do not preserve them.
12191 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
12193 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
12195 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
12198 * config/i386/constraints.md (Bs): Replace
12199 ix86_indirect_branch_register with
12200 TARGET_INDIRECT_BRANCH_REGISTER.
12202 * config/i386/i386.md (indirect_jump): Likewise.
12203 (tablejump): Likewise.
12204 (*sibcall_memory): Likewise.
12205 (*sibcall_value_memory): Likewise.
12206 Peepholes of indirect call and jump via memory: Likewise.
12207 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
12208 (*sibcall_value_GOT_32): Likewise.
12209 * config/i386/predicates.md (indirect_branch_operand): Likewise.
12210 (GOT_memory_operand): Likewise.
12211 (call_insn_operand): Likewise.
12212 (sibcall_insn_operand): Likewise.
12213 (GOT32_symbol_operand): Likewise.
12214 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
12216 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
12218 PR rtl-optimization/83496
12219 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
12220 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
12221 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
12222 redundant insn, if any.
12223 (relax_delay_slots): Likewise.
12224 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
12226 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
12228 PR tree-optimization/83965
12229 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
12230 that grouped statements are part of a reduction chain. Return
12231 true if the statement is not marked as a reduction itself but
12232 is part of a group.
12233 (vect_recog_dot_prod_pattern): Don't check whether the statement
12234 is part of a group here.
12235 (vect_recog_sad_pattern): Likewise.
12236 (vect_recog_widen_sum_pattern): Likewise.
12238 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
12241 * final.c (rest_of_clean_state): Also look for calls inside sequences.
12243 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
12246 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
12248 (ix86_output_indirect_function_return): New prototype.
12249 (ix86_split_simple_return_pop_internal): Likewise.
12250 * config/i386/i386.c (indirect_return_via_cx): New.
12251 (indirect_return_via_cx_bnd): Likewise.
12252 (indirect_thunk_name): Handle return va CX_REG.
12253 (output_indirect_thunk_function): Create alias for
12254 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
12255 (ix86_output_indirect_jmp): Remove the bool argument.
12256 (ix86_output_indirect_function_return): New function.
12257 (ix86_split_simple_return_pop_internal): Likewise.
12258 * config/i386/i386.md (*indirect_jump): Don't pass false
12259 to ix86_output_indirect_jmp.
12260 (*tablejump_1): Likewise.
12261 (simple_return_pop_internal): Change it to define_insn_and_split.
12262 Call ix86_split_simple_return_pop_internal to split it for
12263 -mfunction-return=.
12264 (simple_return_indirect_internal): Call
12265 ix86_output_indirect_function_return instead of
12266 ix86_output_indirect_jmp.
12268 2018-02-26 Jakub Jelinek <jakub@redhat.com>
12271 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
12272 memset and value initialization afterwards.
12274 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
12276 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
12278 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12281 * common/config/aarch64/aarch64-common.c
12282 (aarch_option_optimization_table[]): Switch
12283 off fomit-frame-pointer
12285 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
12286 Chung-Ju Wu <jasonwucj@gmail.com>
12288 * config/nds32/nds32-multiple.md (load_multiple): Disallow
12290 (store_multiple): Ditto.
12292 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
12294 * config.gcc: Add --with-cpu support for nds32 target.
12295 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
12296 * config/nds32/nds32.opt: Add -mcpu= option.
12298 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
12300 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
12301 isel=yes): Warn for these deprecated options.
12303 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
12305 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
12306 ISA_2_5_MASKS_EMBEDDED.
12308 2018-02-23 Jakub Jelinek <jakub@redhat.com>
12310 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
12311 p->max as pointers rather than using iterative_hash_expr.
12313 2018-02-23 Carl Love <cel@us.ibm.com>
12315 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
12316 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
12318 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
12319 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
12320 P8V_BUILTIN_VEC_VUNSIGNED2.
12322 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
12325 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
12326 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
12328 * lra-constraints.c (curr_insn_transform): Set up
12329 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
12331 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
12332 * lra-eliminations.c (spill_pseudos): Ditto.
12333 (process_insn_for_elimination): Ditto.
12334 * lra-lives.c (reg_early_clobber_p): Use the new macros.
12335 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
12336 LRA_NON_CLOBBERED_ALT.
12338 2018-02-22 Martin Sebor <msebor@redhat.com>
12340 PR tree-optimization/84480
12341 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
12342 to maybe_diag_stxncpy_trunc. Call it.
12343 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
12344 from gimple_fold_builtin_strcpy. Print inlining stack.
12345 (handle_builtin_stxncpy): Print inlining stack.
12346 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
12348 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
12351 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
12352 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
12353 and -fcheck-pointer-bounds are used together.
12354 (indirect_thunk_prefix): New enum.
12355 (indirect_thunk_need_prefix): New function.
12356 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
12357 "_nt" instead of "_bnd" for NOTRACK prefix.
12358 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
12359 (output_indirect_thunk_function): Likewise.
12361 (ix86_code_end): Update output_indirect_thunk_function calls.
12362 (ix86_output_indirect_branch_via_reg): Replace
12363 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
12364 (ix86_output_indirect_branch_via_push): Likewise.
12365 (ix86_output_function_return): Likewise.
12366 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
12367 incompatible with -fcf-protection=branch and
12368 -fcheck-pointer-bounds.
12370 2018-02-22 Steve Ellcey <sellcey@cavium.com>
12373 * config/aarch64/aarch64.c (aarch64_print_address_internal):
12374 Change gcc_assert call to output_operand_lossage.
12376 2018-02-22 Steve Ellcey <sellcey@cavium.com>
12378 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
12380 2018-02-22 DJ Delorie <dj@redhat.com>
12381 Sebastian Perta <sebastian.perta@renesas.com>
12382 Oleg Endo <olegendo@gcc.gnu.org>
12384 * config/rx/rx.c (rx_rtx_costs): New function.
12385 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
12387 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
12389 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
12391 2018-02-22 Martin Liska <mliska@suse.cz>
12394 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
12395 Add "native" as a possible value.
12397 2018-02-22 Martin Liska <mliska@suse.cz>
12400 * config/i386/i386.c (ix86_option_override_internal):
12401 Add "native" as a possible value for -march and -mtune.
12403 2018-02-22 Jakub Jelinek <jakub@redhat.com>
12406 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
12407 to all type variants.
12409 PR tree-optimization/84503
12410 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
12411 width as info->bitpos + info->bitsize - start.
12412 (merged_store_group::merge_overlapping): Simplify width computation.
12413 (check_no_overlap): New function.
12414 (imm_store_chain_info::try_coalesce_bswap): Compute expected
12415 start + width and last_order of the group, fail if check_no_overlap
12417 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
12418 to group if check_no_overlap fails.
12420 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
12422 * config/rs6000/altivec.md: Delete contraint arguments to
12423 define_expand, define_split, and define_peephole2, and in
12424 define_insn_and_split if always unused.
12425 * config/rs6000/darwin.md: Ditto.
12426 * config/rs6000/dfp.md: Ditto.
12427 * config/rs6000/rs6000.md: Ditto.
12428 * config/rs6000/sync.md: Ditto.
12429 * config/rs6000/vector.md: Ditto.
12430 * config/rs6000/vsx.md: Ditto.
12432 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
12434 * config/rs6000/altivec.md: Write output control strings as braced
12435 blocks instead of double-quoted strings.
12436 * config/rs6000/darwin.md: Ditto.
12437 * config/rs6000/rs6000.md: Ditto.
12438 * config/rs6000/vector.md: Ditto.
12439 * config/rs6000/vsx.md: Ditto.
12441 2018-02-21 Jason Merrill <jason@redhat.com>
12443 PR c++/84314 - ICE with templates and fastcall attribute.
12444 * attribs.c (build_type_attribute_qual_variant): Remove assert.
12446 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
12448 * ipa-cp.c (determine_versionability): Fix comment typos.
12450 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
12453 * ipa-cp.c (determine_versionability): Do not version functions caling
12456 2018-02-21 Martin Liska <mliska@suse.cz>
12459 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
12460 Add "native" as a possible value.
12461 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
12462 the macro when native cpu detection is available.
12464 2018-02-21 Martin Liska <mliska@suse.cz>
12467 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
12468 Add "native" as a possible value.
12469 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
12470 when native cpu detection is available.
12472 2018-02-21 Jakub Jelinek <jakub@redhat.com>
12473 Martin Sebor <msebor@redhat.com>
12475 PR tree-optimization/84478
12476 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
12478 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
12479 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
12480 support which is conservatively correct, for 2 only stay conservative
12481 for maxlen. Formatting and comment capitalization fixes. Add STRICT
12482 argument to the 2 argument get_range_strlen, adjust 6 arg
12483 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
12485 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
12486 (gimple_fold_builtin_strlen): Pass true as last argument to
12489 2018-02-20 Martin Sebor <msebor@redhat.com>
12491 PR middle-end/84095
12492 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
12493 (builtin_memref::set_base_and_offset): Same. Handle inner references.
12494 (builtin_memref::builtin_memref): Factor out parts into
12495 set_base_and_offset and call it.
12497 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
12499 PR middle-end/84406
12500 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
12501 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
12502 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
12503 search at the associated MODE_INT.
12505 2018-02-20 Jeff Law <law@redhat.com>
12507 PR middle-end/82123
12508 PR tree-optimization/81592
12509 PR middle-end/79257
12510 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
12511 for range data rather than using global data.
12512 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
12513 range data rather than using global data.
12514 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
12515 pass it to children as needed.
12516 (struct directive::fmtresult): Similarly.
12517 (struct directive::set_width): Similarly.
12518 (struct directive::set_precision): Similarly.
12519 (format_integer, format_directive, parse_directive): Similarly.
12520 (format_none): Accept unnamed vr_values parameter.
12521 (format_percent, format_floating, format_character): Similarly.
12522 (format_string, format_plain): Similarly.
12523 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
12524 the EVRP range analyzer for range data rather than using global data.
12525 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
12526 gimple-ssa-evrp-analyze.h
12527 (class sprintf_dom_walker): Add after_dom_children member function.
12528 Add evrp_range_analyzer member.
12529 (sprintf_dom_walker::before_dom_children): Call into the EVRP
12530 range analyzer as needed.
12531 (sprintf_dom_walker::after_dom_children): New member function.
12532 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
12534 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
12535 (evrp_range_analyzer::pop_to_marker): Likewise.
12537 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
12539 PR tree-optimization/84419
12540 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
12541 with the required type if its current type is compatible but
12544 2018-02-20 Jakub Jelinek <jakub@redhat.com>
12546 PR middle-end/82004
12547 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
12548 after vectorization.
12550 2018-02-20 Martin Liska <mliska@suse.cz>
12553 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
12554 possible values if we don't have a hint.
12556 2018-02-20 Martin Liska <mliska@suse.cz>
12560 * final.c (shorten_branches): Build align_tab array with one
12562 * opts.c (finish_options): Add alignment option limit check.
12563 (MAX_CODE_ALIGN): Likewise.
12564 (MAX_CODE_ALIGN_VALUE): Likewise.
12565 * doc/invoke.texi: Document maximum allowed option value for
12566 all -falign-* options.
12568 2018-02-19 Jakub Jelinek <jakub@redhat.com>
12571 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
12572 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
12573 * var-tracking.c (emit_note_insn_var_location): Remove all references
12574 to NOTE_INSN_CALL_ARG_LOCATION.
12575 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
12576 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
12577 Use copy_rtx_if_shared.
12578 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
12579 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
12580 (dwarf2out_var_location): Remove handling of
12581 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
12583 * final.c (final_scan_insn): Remove all references to
12584 NOTE_INSN_CALL_ARG_LOCATION.
12585 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
12586 before dumping final insns.
12587 * except.c (emit_note_eh_region_end): Remove all references to
12588 NOTE_INSN_CALL_ARG_LOCATION.
12589 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
12590 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
12591 * config/arc/arc.c (hwloop_optimize): Likewise.
12592 * config/arm/arm.c (create_fix_barrier): Likewise.
12593 * config/s390/s390.c (s390_chunkify_start): Likewise.
12594 * config/sh/sh.c (find_barrier): Likewise.
12595 * config/i386/i386.c (rest_of_insert_endbranch,
12596 ix86_seh_fixup_eh_fallthru): Likewise.
12597 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
12598 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
12599 * config/frv/frv.c (frv_function_prologue): Likewise.
12600 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
12602 (note_outside_basic_block_p): Remove all references to
12603 NOTE_INSN_CALL_ARG_LOCATION.
12604 * gengtype.c (adjust_field_rtx_def): Likewise.
12605 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
12607 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
12608 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
12611 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
12614 PR tree-optimization/84452
12615 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
12616 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
12619 2018-02-19 Martin Liska <mliska@suse.cz>
12622 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
12624 2018-02-19 Martin Liska <mliska@suse.cz>
12625 Richard Sandiford <richard.sandiford@linaro.org>
12627 PR tree-optimization/82491
12628 * gimple-fold.c (get_base_constructor): Make earlier bail out
12631 2018-02-19 Carl Love <cel@us.ibm.com>
12633 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
12634 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
12636 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
12637 P8V_BUILTIN_VEC_NEG.
12639 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
12641 * config/rl78/rl78.md (movdf): New define expand.
12643 2018-02-19 Martin Liska <mliska@suse.cz>
12646 * doc/invoke.texi: Fix typo.
12647 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
12649 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
12651 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
12652 handle rs6000_single_float and rs6000_double_float specially for
12655 2018-02-16 Jeff Law <law@redhat.com>
12657 * config/rx/rx.c (add_pop_cfi_notes): New function.;
12658 (pop_regs): Use it.
12660 2018-02-16 Jakub Jelinek <jakub@redhat.com>
12663 * ipa-inline.c (inline_small_functions): Fix a typo.
12665 2018-02-16 Nathan Sidwell <nathan@acm.org>
12667 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
12669 2018-02-16 Carl Love <cel@us.ibm.com>
12671 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
12672 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
12673 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
12674 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
12675 expansion to P8V_BUILTIN_VEC_FLOAT2.
12677 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
12679 PR rtl-optimization/70023
12680 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
12681 src_regno into account.
12683 2018-02-16 Carl Love <cel@us.ibm.com>
12685 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
12686 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
12687 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
12688 * config/rs6000/rs6000.c: Remove case statements for
12689 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
12690 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
12691 and P9V_BUILTIN_VEC_VINSERT4B.
12692 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
12693 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
12694 * config/rs6000/vsx.md:
12695 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
12698 2018-02-16 Carl Love <cel@us.ibm.com>
12700 * config/rs6000/altivec.h: Add builtin names vec_extract4b
12702 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
12704 * config/rs6000/rs6000-c.c: Add the definitions for
12705 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
12706 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
12707 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
12708 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
12709 definition for insert4b and define insn *insert3b_internal.
12710 * doc/extend.texi: Add documentation for vec_extract4b.
12712 2018-02-16 Nathan Sidwell <nathan@acm.org>
12714 * doc/extend.texi (Backwards Compatibility): Mention friend
12715 injection. Note for-scope is deprecated.
12716 * doc/invoke.texi (-ffriend-injection): Deprecate.
12718 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
12720 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
12721 that moved to I2, also allow destinations that are a paradoxical
12722 subreg (instead of a normal reg).
12724 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
12727 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
12730 2018-02-16 Richard Biener <rguenther@suse.de>
12732 PR tree-optimization/84037
12733 PR tree-optimization/84016
12735 * config/i386/i386.c (ix86_builtin_vectorization_cost):
12736 Adjust vec_construct for the fact we need additional higher latency
12737 128bit inserts for AVX256 and AVX512 vector builds.
12738 (ix86_add_stmt_cost): Scale vector construction cost for
12741 2018-02-16 Richard Biener <rguenther@suse.de>
12743 PR tree-optimization/84417
12744 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
12745 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
12746 (non_rewritable_lvalue_p): Likewise, use poly-ints.
12748 2018-02-16 Martin Liska <mliska@suse.cz>
12751 * internal-fn.def (ASAN_CHECK): Set proper flags.
12752 (ASAN_MARK): Likewise.
12754 2018-02-16 Julia Koval <julia.koval@intel.com>
12756 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
12757 from PTA_CANNONLAKE.
12759 2018-02-16 Jakub Jelinek <jakub@redhat.com>
12762 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
12763 Use ++iter rather than iter++ for std::list iterators.
12764 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
12765 defer deleting them until all nodes in the forest are processed. Do
12766 free even leaf nodes. Change to_process into auto_vec.
12769 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
12770 * vec.h (vec_default_construct): Use memset instead of placement new
12771 if BROKEN_VALUE_INITIALIZATION is defined.
12772 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
12773 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
12776 PR rtl-optimization/83723
12777 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
12778 * lra.c (lra_substitute_pseudo): Likewise. If true, use
12779 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
12781 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
12783 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
12785 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
12787 PR rtl-optimization/81443
12788 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
12789 from inner REGs to paradoxical SUBREGs.
12791 2018-02-16 Richard Biener <rguenther@suse.de>
12793 PR tree-optimization/84399
12794 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
12795 For operands we can analyze at their definition make sure we can
12796 analyze them at each use as well.
12798 2018-02-16 Richard Biener <rguenther@suse.de>
12800 PR tree-optimization/84190
12801 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
12802 volatile accesses if the decl isn't volatile.
12804 2018-02-15 Jason Merrill <jason@redhat.com>
12806 PR c++/84314 - ICE with templates and fastcall attribute.
12807 * attribs.c (build_type_attribute_qual_variant): Don't clobber
12808 TYPE_CANONICAL on an existing type.
12810 2018-02-15 Jakub Jelinek <jakub@redhat.com>
12812 PR tree-optimization/84383
12813 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
12814 dstoff nor call operand_equal_p if dstbase is NULL.
12816 PR tree-optimization/84334
12817 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
12818 also a CONSTANT_CLASS_P, punt.
12820 2018-02-14 Jim Wilson <jimw@sifive.com>
12822 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
12823 first SMALL_OPERAND check. New local min_second_step. Move assert
12824 to where locals are set. Add TARGET_RVC support.
12825 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
12827 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
12829 * doc/invoke.texi: Correct -Wformat-overflow code sample.
12831 2018-02-14 Martin Sebor <msebor@redhat.com>
12833 PR tree-optimization/83698
12834 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
12835 arrays constrain the offset range to their bounds.
12836 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
12837 (builtin_access::overlap): Avoid setting the size of overlap if it's
12839 (maybe_diag_overlap): Also consider arrays when deciding what values
12840 of offsets to include in diagnostics.
12842 2018-02-14 Martin Sebor <msebor@redhat.com>
12845 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
12846 that correspond to the kind of a declaration.
12848 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
12851 * config/pa/pa.md: Load address of PIC label using the linkage table
12852 if the label is nonlocal.
12854 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
12856 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
12857 warning message if user requests -maltivec=be.
12858 * doc/invoke.texi: Document deprecation of -maltivec=be.
12860 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
12863 * config/rs6000/rs6000-c.c: Update definitions for
12864 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
12865 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
12867 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
12870 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
12871 add _get_ssp intrinsics. Remove argument from
12872 __builtin_ia32_rdssp[d|q].
12873 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
12874 * config/i386/i386-builtin.def: Remove argument from
12875 __builtin_ia32_rdssp[d|q].
12876 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
12877 ix86_expand_special_args_builtin for _rdssp[d|q].
12878 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
12879 Clear register before usage.
12880 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
12881 Add documentation for new _get_ssp and _inc_ssp intrinsics.
12883 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
12885 PR tree-optimization/84357
12886 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
12887 operand 1 of an ARRAY_REF too.
12889 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
12892 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
12893 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
12895 (set_of_reg): New struct.
12896 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
12897 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
12898 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
12900 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
12901 Split into bitclr, bitset, bitinvert patterns if appropriate.
12902 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
12903 use rx_fuse_in_memory_bitop.
12904 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
12905 to named insn, correct maximum insn length.
12907 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
12910 * machmode.def: Define a complex mode for PARTIAL_INT.
12911 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
12913 * doc/rtl.texi: Document CSPImode.
12914 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
12916 (msp430_hard_regno_nregs_with_padding): Likewise.
12918 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
12921 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
12923 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
12925 PR rtl-optimization/84169
12926 * combine.c (try_combine): New variable split_i2i3. Set it to true if
12927 we generated a parallel as new i3 and we split that to new i2 and i3
12928 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
12929 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
12930 those to i2, not i1. Partially rewrite this scan code.
12932 2018-02-13 Jakub Jelinek <jakub@redhat.com>
12935 * stor-layout.c (place_field): For variable length fields, adjust
12936 offset_align afterwards not just based on the field's alignment,
12937 but also on the size.
12939 PR middle-end/84309
12940 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
12941 of exps and logs in the use_exp2 case.
12943 2018-02-13 Jeff Law <law@redhat.com>
12945 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
12946 entry for "vector".
12948 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
12951 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
12955 * common.opt (gas-loc-support, gas-locview-support): New.
12956 (ginline-points, ginternal-reset-location-views): New.
12957 * doc/invoke.texi: Document them. Use @itemx where intended.
12958 (gvariable-location-views): Adjust.
12959 * target.def (reset_location_view): New.
12960 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
12961 (TARGET_RESET_LOCATION_VIEW): New.
12962 * doc/tm.texi: Rebuilt.
12963 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
12964 (dwarf2out_default_as_locview_support): New.
12965 (output_asm_line_debug_info): Use option variables.
12966 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
12967 (output_loc_list): Likewise.
12968 (add_high_low_attributes): Check option variables.
12969 Don't output entry view attribute in strict mode.
12970 (gen_inlined_subroutine_die): Check option variables.
12971 (dwarf2out_inline_entry): Likewise.
12972 (init_sections_and_labels): Likewise.
12973 (dwarf2out_early_finish): Likewise.
12974 (maybe_reset_location_view): New, from...
12975 (dwarf2out_var_location): ... here. Call it.
12976 * debug.h (dwarf2out_default_as_loc_support): Declare.
12977 (dwarf2out_default_as_locview_support): Declare.
12978 * hooks.c (hook_int_rtx_insn_0): New.
12979 * hooks.h (hook_int_rtx_insn_0): Declare.
12980 * toplev.c (process_options): Take -gas-loc-support and
12981 -gas-locview-support from dwarf2out. Enable
12982 -gvariable-location-views by default only with locview
12983 assembler support. Enable -ginternal-reset-location-views by
12984 default only if the target defines the corresponding hook.
12985 Enable -ginline-points by default if location views are
12986 enabled; force it disabled if statement frontiers are
12988 * tree-inline.c (expand_call_inline): Check option variables.
12989 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
12991 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
12993 PR tree-optimization/84321
12994 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
12995 handling. Also check whether the anti-range contains any values
12996 that satisfy the mask; switch to a VR_RANGE if not.
12998 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
13001 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
13003 2018-02-13 Martin Jambor <mjambor@suse.cz>
13006 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
13007 of call statements, also set location of a load to a temporary.
13009 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
13011 * config/rl78/rl78.c (add_vector_labels): New function.
13012 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
13013 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
13014 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
13015 which checks that no arguments are passed.
13016 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
13017 * doc/extend.texi: Documentation for the new attribute.
13019 2018-02-13 Andreas Schwab <schwab@suse.de>
13021 * config/riscv/linux.h (CPP_SPEC): Define.
13023 2018-02-13 Jakub Jelinek <jakub@redhat.com>
13026 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
13027 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
13028 OPTION_MASK_ISA_AES as first argument to def_builtin_const
13029 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
13030 instead of OPTION_MASK_ISA_PCLMUL as first argument to
13031 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
13032 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
13033 temporarily for AES and PCLMUL builtins.
13035 PR tree-optimization/84339
13036 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
13037 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
13040 PR middle-end/84309
13041 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
13042 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
13043 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
13045 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
13047 * omp-simd-clone.h: New file.
13048 * omp-simd-clone.c: Include omp-simd-clone.h.
13049 (expand_simd_clones): No longer static.
13050 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
13051 cgraph.h and omp-simd-clone.h.
13052 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
13053 (vect_recog_widen_shift_pattern): Formatting fix.
13054 (vect_pattern_recog_1): Don't check optab for calls.
13057 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
13058 operands[2] into a REG before using gen_lowpart on it.
13060 2018-02-12 Jeff Law <law@redhat.com>
13063 * config/sh/sh.c (find_barrier): Consider a sibling call
13066 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
13067 successfully back substituting a reg.
13069 2018-02-12 Richard Biener <rguenther@suse.de>
13071 PR tree-optimization/84037
13072 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
13073 parameter, move visited init to caller.
13074 (vect_slp_analyze_operations): Separate cost from validity
13075 check, initialize visited once for all instances.
13076 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
13078 * tree-vect-stmts.c (vect_model_simple_cost): Make early
13080 (vect_model_promotion_demotion_cost): Likewise.
13081 (vectorizable_bswap): Guard cost modeling with !slp_node
13082 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
13084 (vectorizable_call): Likewise.
13085 (vectorizable_conversion): Likewise.
13086 (vectorizable_assignment): Likewise.
13087 (vectorizable_shift): Likewise.
13088 (vectorizable_operation): Likewise.
13089 (vectorizable_store): Likewise.
13090 (vectorizable_load): Likewise.
13091 (vectorizable_condition): Likewise.
13092 (vectorizable_comparison): Likewise.
13094 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
13097 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
13098 (ASAN_MARK): Fix fnspec to account for return value, change pointer
13099 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
13101 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
13103 PR middle-end/83665
13104 * params.def (inline-min-speedup): Increase from 8 to 15.
13105 (max-inline-insns-auto): Decrease from 40 to 30.
13106 * ipa-split.c (consider_split): Add some buffer for function to
13107 be considered inlining candidate.
13108 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
13111 2018-02-12 Richard Biener <rguenther@suse.de>
13113 PR tree-optimization/84037
13114 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
13115 matched stmts if we cannot swap the non-matched ones.
13117 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
13119 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
13120 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
13121 _mm_maskz_scalef_round_ss): New intrinsics.
13122 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
13123 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
13124 __builtin_ia32_scalefss_round): Remove.
13125 (__builtin_ia32_scalefsd_mask_round,
13126 __builtin_ia32_scalefss_mask_round): New intrinsics.
13127 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
13128 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
13129 ((match_operand:VF_128 2 "<round_nimm_predicate>"
13130 "<round_constraint>")): Changed to ...
13131 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
13132 "<round_scalar_constraint>")): ... this.
13133 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
13134 %0, %1, %2<round_op3>}"): Changed to ...
13135 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
13136 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
13137 %2<round_scalar_mask_op3>}"): ... this.
13138 * config/i386/subst.md (round_scalar_nimm_predicate): New.
13140 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
13142 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
13143 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
13144 (_mm_maskz_sqrt_round_ss): New intrinsics.
13145 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
13146 (__builtin_ia32_sqrtsd_mask_round)
13147 (__builtin_ia32_sqrtss_mask_round): New builtins.
13148 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
13149 (__builtin_ia32_sqrtss_round): Remove.
13150 (__builtin_ia32_sqrtsd_mask_round)
13151 (__builtin_ia32_sqrtss_mask_round): New builtins.
13152 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
13153 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
13154 ((match_operand:VF_128 1 "vector_operand"
13155 "xBm,<round_constraint>")): Changed to ...
13156 ((match_operand:VF_128 1 "vector_operand"
13157 "xBm,<round_scalar_constraint>")): ... this.
13158 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
13159 %0, %2, %<iptr>1<round_op3>}): Changed to ...
13160 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
13161 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
13162 %<iptr>1<round_scalar_mask_op3>}): ... this.
13163 ((set_attr "prefix" "<round_prefix>")): Changed to ...
13164 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
13166 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
13169 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
13170 Cast vec_cmpeq result to correct type.
13171 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
13172 Cast vec_cmpgt result to correct type.
13174 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
13176 * final.c (final_scan_insn_1): Renamed from...
13177 (final_scan_insn): ... this. New wrapper, to recover
13178 seen from the outermost call in recursive ones.
13179 * config/sparc/sparc.c (output_return): Drop seen from call.
13180 (output_sibcall): Likewise.
13181 * config/visium/visium.c (output_branch): Likewise.
13183 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
13185 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
13188 2018-02-10 Alan Modra <amodra@gmail.com>
13191 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
13192 Specify LR as an input.
13194 2018-02-10 Jakub Jelinek <jakub@redhat.com>
13197 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
13198 remove_member_access_dummy_vars): New functions.
13199 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
13200 lower_omp_1, execute_lower_omp): Use them.
13202 PR rtl-optimization/84308
13203 * shrink-wrap.c (spread_components): Release todo vector.
13205 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
13207 PR rtl-optimization/57193
13208 * ira-color.c (struct allocno_color_data): Add member
13209 conflict_allocno_hard_prefs.
13210 (update_conflict_allocno_hard_prefs): New.
13211 (bucket_allocno_compare_func): Add a preference based on
13212 conflict_allocno_hard_prefs.
13213 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
13214 (color_allocnos): Remove a dead code. Initiate
13215 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
13217 2018-02-09 Jakub Jelinek <jakub@redhat.com>
13220 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
13221 constraint from =wa to wa. Avoid a subreg on the output operand,
13222 instead use a pseudo and subreg it in a move.
13223 (p9_xxbrd_<mode>): Changed to ...
13224 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
13225 (p9_xxbrd_v2df): New expander.
13226 (p9_xxbrw_<mode>): Changed to ...
13227 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
13228 (p9_xxbrw_v4sf): New expander.
13230 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
13232 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
13234 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
13237 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
13238 multiply in 32-bit mode.
13239 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
13240 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
13243 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
13245 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
13246 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
13247 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
13248 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
13250 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
13253 * dwarf2out.c (is_trivial_indirect_ref): New function.
13254 (dwarf2out_late_global_decl): Do not generate a location
13255 attribute for variables that have a non-trivial DECL_VALUE_EXPR
13256 and that are not defined in the current unit.
13258 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
13260 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
13261 instead of a libcall for UNORDERED.
13263 2018-02-09 Tamar Christina <tamar.christina@arm.com>
13266 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
13267 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
13269 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13272 * config/s390/s390.c (s390_set_current_function): Invoke
13273 s390_indirect_branch_settings also if fndecl didn't change.
13275 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
13277 * config/rs6000/rs6000.md (blockage): Set length to zero.
13279 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
13281 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
13283 2018-02-09 Jakub Jelinek <jakub@redhat.com>
13286 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
13287 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
13291 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
13292 PARALLEL incoming that failed vt_get_decl_and_offset check.
13294 PR middle-end/84237
13295 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
13296 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
13298 (get_variable_section): For decls in named .bss* sections pass true as
13299 second argument to bss_initializer_p.
13301 2018-02-09 Marek Polacek <polacek@redhat.com>
13302 Jakub Jelinek <jakub@redhat.com>
13305 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
13306 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
13307 Sync some changes from cxx_fold_indirect_ref.
13309 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
13311 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
13313 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
13314 (BLOCK_INLINE_ENTRY_LABEL): New.
13315 (dwarf2out_var_location): Disregard inline entry markers.
13316 (inline_entry_data): New struct.
13317 (inline_entry_data_hasher): New hashtable type.
13318 (inline_entry_data_hasher::hash): New.
13319 (inline_entry_data_hasher::equal): New.
13320 (inline_entry_data_table): New variable.
13321 (add_high_low_attributes): Add DW_AT_entry_pc and
13322 DW_AT_GNU_entry_view attributes if a pending entry is found
13323 in inline_entry_data_table. Add old entry_pc attribute only
13324 if debug nonbinding markers are disabled.
13325 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
13326 markers are enabled.
13327 (block_within_block_p, dwarf2out_inline_entry): New.
13328 (dwarf2out_finish): Check that no entries remained in
13329 inline_entry_data_table.
13330 * final.c (reemit_insn_block_notes): Handle inline entry notes.
13331 (final_scan_insn, notice_source_line): Likewise.
13332 (rest_of_clean_state): Skip inline entry markers.
13333 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
13335 * gimple.c (gimple_build_debug_inline_entry): New.
13336 * gimple.h (enum gimple_debug_subcode): Add
13337 GIMPLE_DEBUG_INLINE_ENTRY.
13338 (gimple_build_debug_inline_entry): Declare.
13339 (gimple_debug_inline_entry_p): New.
13340 (gimple_debug_nonbind_marker_p): Adjust.
13341 * insn-notes.def (INLINE_ENTRY): New.
13342 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
13343 inline entry marker notes.
13344 (print_insn): Likewise.
13345 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
13346 (INSN_DEBUG_MARKER_KIND): Likewise.
13347 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
13348 * tree-inline.c (expand_call_inline): Build and insert
13349 debug_inline_entry stmt.
13350 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
13351 inline entry blocks early, if nonbind markers are enabled.
13352 (dump_scope_block): Dump fragment info.
13353 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
13354 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
13355 (gimple_build_debug_inline_entry): New.
13356 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
13357 Enable/disable inline entry points too.
13358 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
13359 (DEBUG_INSN): Describe inline entry markers.
13361 * common.opt (gvariable-location-views): New.
13362 (gvariable-location-views=incompat5): New.
13363 * config.in: Rebuilt.
13364 * configure: Rebuilt.
13365 * configure.ac: Test assembler for view support.
13366 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
13367 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
13368 * dwarf2out.c (var_loc_view): New typedef.
13369 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
13370 (dwarf2out_locviews_in_attribute): New.
13371 (dwarf2out_locviews_in_loclist): New.
13372 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
13373 (enum dw_line_info_opcode): Add LI_adv_address.
13374 (struct dw_line_info_table): Add view.
13375 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
13376 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
13377 (zero_view_p): New variable.
13378 (ZERO_VIEW_P): New macro.
13379 (output_asm_line_debug_info): New.
13380 (struct var_loc_node): Add view.
13381 (add_AT_view_list, AT_loc_list): New.
13382 (add_var_loc_to_decl): Add view param. Test it against last.
13383 (new_loc_list): Add view params. Record them.
13384 (AT_loc_list_ptr): Handle loc and view lists.
13385 (view_list_to_loc_list_val_node): New.
13386 (print_dw_val): Handle dw_val_class_view_list.
13387 (size_of_die): Likewise.
13388 (value_format): Likewise.
13389 (loc_list_has_views): New.
13390 (gen_llsym): Set vl_symbol too.
13391 (maybe_gen_llsym, skip_loc_list_entry): New.
13392 (dwarf2out_maybe_output_loclist_view_pair): New.
13393 (output_loc_list): Output view list or entries too.
13394 (output_view_list_offset): New.
13395 (output_die): Handle dw_val_class_view_list.
13396 (output_dwarf_version): New.
13397 (output_compilation_unit_header): Use it.
13398 (output_skeleton_debug_sections): Likewise.
13399 (output_rnglists, output_line_info): Likewise.
13400 (output_pubnames, output_aranges): Update version comments.
13401 (output_one_line_info_table): Output view numbers in asm comments.
13402 (dw_loc_list): Determine current endview, pass it to new_loc_list.
13403 Call maybe_gen_llsym.
13404 (loc_list_from_tree_1): Adjust.
13405 (add_AT_location_description): Create view list attribute if
13406 needed, check it's absent otherwise.
13407 (convert_cfa_to_fb_loc_list): Adjust.
13408 (maybe_emit_file): Call output_asm_line_debug_info for test.
13409 (dwarf2out_var_location): Reset views as needed. Precompute
13410 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
13411 attribute. Set view.
13412 (new_line_info_table): Reset next view.
13413 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
13414 (dwarf2out_source_line): Likewise. Output view resets and labels to
13415 the assembler, or select appropriate line info opcodes.
13416 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
13417 (optimize_string_length): Catch it. Adjust.
13418 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
13419 dw_val_class_view_list, and remove it if no longer needed.
13420 (hash_loc_list): Hash view numbers.
13421 (loc_list_hasher::equal): Compare them.
13422 (optimize_location_lists): Check whether a view list symbol is
13423 needed, and whether the locview attribute is present, and
13424 whether they match. Remove the locview attribute if no longer
13426 (index_location_lists): Call skip_loc_list_entry for test.
13427 (dwarf2out_finish): Call output_asm_line_debug_info for test.
13428 Use output_dwarf_version.
13429 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
13430 (struct dw_val_node): Add val_view_list.
13431 * final.c (SEEN_NEXT_VIEW): New.
13432 (set_next_view_needed): New.
13433 (clear_next_view_needed): New.
13434 (maybe_output_next_view): New.
13435 (final_start_function): Rename to...
13436 (final_start_function_1): ... this. Take pointer to FIRST,
13437 add SEEN parameter. Emit param bindings in the initial view.
13438 (final_start_function): Reintroduce SEEN-less interface.
13439 (final): Rename to...
13440 (final_1): ... this. Take SEEN parameter. Output final pending
13441 next view at the end.
13442 (final): Reintroduce seen-less interface.
13443 (final_scan_insn): Output pending next view before switching
13444 sections or ending a block. Mark the next view as needed when
13445 outputting variable locations. Notify debug backend of section
13446 changes, and of location view changes.
13447 (rest_of_handle_final): Adjust.
13448 * toplev.c (process_options): Autodetect value for debug variable
13449 location views option. Warn on incompat5 without -gdwarf-5.
13450 * doc/invoke.texi (gvariable-location-views): New.
13451 (gvariable-location-views=incompat5): New.
13452 (gno-variable-location-views): New.
13454 2018-02-08 David Malcolm <dmalcolm@redhat.com>
13456 PR tree-optimization/84136
13457 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
13458 that the result of find_edge is non-NULL.
13460 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
13463 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
13464 storing integer register in SImode. Fix cost of 256 and 512
13465 byte aligned SSE register store.
13467 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
13469 * config/i386/i386.c (ix86_multiplication_cost): Fix
13470 multiplication cost for TARGET_AVX512DQ.
13472 2018-02-08 Marek Polacek <polacek@redhat.com>
13474 PR tree-optimization/84238
13475 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
13478 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
13480 PR tree-optimization/84265
13481 * tree-vect-stmts.c (vectorizable_store): Don't treat
13482 VMAT_CONTIGUOUS accesses as grouped.
13483 (vectorizable_load): Likewise.
13485 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
13487 PR tree-optimization/81635
13488 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
13489 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
13490 (test_round_for_mask): New functions.
13491 (wide_int_cc_tests): Call test_round_for_mask.
13492 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
13493 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
13494 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
13495 range returned by get_range_info.
13497 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
13500 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
13501 * symtab.c: Include builtins.h
13502 (symtab_node::output_to_lto_symbol_table_p): Move here
13503 from lto-streamer-out.c:output_symbol_p.
13504 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
13505 (output_symbol_p): Move all logic to symtab.c
13506 (produce_symtab): Update.
13508 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13510 * config/s390/s390-opts.h (enum indirect_branch): Define.
13511 * config/s390/s390-protos.h (s390_return_addr_from_memory)
13512 (s390_indirect_branch_via_thunk)
13513 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
13514 (enum s390_indirect_branch_type): Define.
13515 * config/s390/s390.c (struct s390_frame_layout, struct
13516 machine_function): Remove.
13517 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
13518 (indirect_branch_table_label_no, indirect_branch_table_name):
13520 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
13521 (enum s390_indirect_branch_option): Define.
13522 (s390_return_addr_from_memory): New function.
13523 (s390_handle_string_attribute): New function.
13524 (s390_attribute_table): Add new attribute handler.
13525 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
13526 (s390_indirect_branch_via_thunk): New function.
13527 (s390_indirect_branch_via_inline_thunk): New function.
13528 (s390_function_ok_for_sibcall): When jumping via thunk disallow
13529 sibling call optimization for non z10 compiles.
13530 (s390_emit_call): Force indirect branch target to be a single
13531 register. Add r1 clobber for non-z10 compiles.
13532 (s390_emit_epilogue): Emit return jump via return_use expander.
13533 (s390_reorg): Handle JUMP_INSNs as execute targets.
13534 (s390_option_override_internal): Perform validity checks for the
13535 new command line options.
13536 (s390_indirect_branch_attrvalue): New function.
13537 (s390_indirect_branch_settings): New function.
13538 (s390_set_current_function): Invoke s390_indirect_branch_settings.
13539 (s390_output_indirect_thunk_function): New function.
13540 (s390_code_end): Implement target hook.
13541 (s390_case_values_threshold): Implement target hook.
13542 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
13544 * config/s390/s390.h (struct s390_frame_layout)
13545 (struct machine_function): Move here from s390.c.
13546 (TARGET_INDIRECT_BRANCH_NOBP_RET)
13547 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
13548 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
13549 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
13550 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
13551 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
13552 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
13553 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
13554 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
13555 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
13556 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
13557 (mnemonic attribute): Add values which aren't recognized
13559 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
13560 pattern for branch conversion. Fix mnemonic attribute.
13561 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
13562 indirect branch via thunk if requested.
13563 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
13564 ("*indirect_jump"): Disable for branch conversion using out of
13566 ("indirect_jump_via_thunk<mode>_z10")
13567 ("indirect_jump_via_thunk<mode>")
13568 ("indirect_jump_via_inlinethunk<mode>_z10")
13569 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
13570 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
13571 ("casesi_jump_via_inlinethunk<mode>_z10")
13572 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
13573 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
13574 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
13575 ("*indirect2_jump"): Disable for branch conversion.
13576 ("casesi_jump"): Turn into expander and expand patterns for branch
13578 ("return_use"): New expander.
13579 ("*return"): Emit return via thunk and rename it to ...
13580 ("*return<mode>"): ... this one.
13581 * config/s390/s390.opt: Add new options and and enum for the
13584 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
13586 * lra-constraints.c (match_reload): Unconditionally use
13587 gen_lowpart_SUBREG, rather than selecting between that
13588 and equivalent gen_rtx_SUBREG code.
13590 2018-02-08 Richard Biener <rguenther@suse.de>
13592 PR tree-optimization/84233
13593 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
13594 changed flag instead of boguously re-using phi_inserted.
13596 2018-02-08 Martin Jambor <mjambor@suse.cz>
13598 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
13599 static local variables.
13601 2018-02-08 Richard Biener <rguenther@suse.de>
13603 PR tree-optimization/84278
13604 * tree-vect-stmts.c (vectorizable_store): When looking for
13605 smaller vector types to perform grouped strided loads/stores
13606 make sure the mode is supported by the target.
13607 (vectorizable_load): Likewise.
13609 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
13611 * config/aarch64/aarch64.c (aarch64_components_for_bb):
13612 Increase LDP/STP opportunities by adding adjacent callee-saves.
13614 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
13616 PR rtl-optimization/84068
13617 PR rtl-optimization/83459
13618 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
13620 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
13622 PR tree-optimization/84224
13623 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
13624 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
13625 non-zero arguments.
13627 2018-02-07 Iain Sandoe <iain@codesourcery.com>
13630 * config/rs6000/altivec.md (*restore_world): Remove LR use.
13631 * config/rs6000/predicates.md (restore_world_operation): Adjust op
13632 count, remove one USE.
13634 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13636 * doc/install.texi (Configuration): Document the
13637 --with-long-double-format={ibm,ieee} PowerPC configuration
13641 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
13642 Convert from define_expand to be define_insn_and_split. Rework
13643 float/double/_Float128 conversions to QI/HI/SImode to work with
13644 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
13645 conversions to QI/HImode types did a store and then a load to
13646 truncate the value. For conversions to VSX registers, don't split
13647 the insn, instead emit the code directly. Use the code iterator
13648 any_fix to combine signed and unsigned conversions.
13649 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
13650 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
13651 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
13652 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
13653 (fix_<mode>di2_hw): Likewise.
13654 (fixuns_<mode>di2_hw): Likewise.
13655 (fix_<mode>si2_hw): Likewise.
13656 (fixuns_<mode>si2_hw): Likewise.
13657 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
13658 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
13659 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
13660 fix<uns>_trunc<SFDF:mode>si2_p8.
13661 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
13662 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13663 (fix<uns>_<mode>_mem): Likewise.
13664 (fctiw<u>z_<mode>_mem): Likewise.
13665 (fix<uns>_<mode>_mem): Likewise.
13666 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
13667 the register allocator from doing a direct move to the GPRs to do
13668 a store, and instead use the ISA 3.0 store byte/half-word from
13669 vector register instruction. For IEEE 128-bit floating point,
13670 also optimize stores of 32-bit ints.
13671 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
13673 2018-02-07 Alan Hayward <alan.hayward@arm.com>
13675 * genextract.c (push_pathstr_operand): New function to support
13677 (walk_rtx): Call push_pathstr_operand.
13678 (print_path): Support [a-zA-Z].
13680 2018-02-07 Richard Biener <rguenther@suse.de>
13682 PR tree-optimization/84037
13683 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
13684 (cse_and_gimplify_to_preheader): Declare.
13685 (vect_get_place_in_interleaving_chain): Likewise.
13686 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
13688 (_loop_vec_info::~_loop_vec_info): Delete it.
13689 (cse_and_gimplify_to_preheader): New function.
13690 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
13691 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
13692 (vectorizable_load): Likewise. For grouped stores always base
13693 the IV on the first element.
13694 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
13695 condition before gimplifying.
13697 2018-02-07 Jakub Jelinek <jakub@redhat.com>
13699 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
13700 *DIV_EXPR and *MOD_EXPR.
13702 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
13705 * config/i386/i386.c (ix86_option_override_internal): Mask out
13706 the CF_SET bit when checking -fcf-protection.
13708 2018-02-07 Tom de Vries <tom@codesourcery.com>
13711 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
13714 2018-02-07 Richard Biener <rguenther@suse.de>
13716 PR tree-optimization/84204
13717 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
13720 PR tree-optimization/84205
13721 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
13722 special-case isl_ast_op_zdiv_r.
13724 PR tree-optimization/84223
13725 * graphite-scop-detection.c (gather_bbs::before_dom_children):
13726 Only add conditions from within the region.
13727 (gather_bbs::after_dom_children): Adjust.
13729 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
13732 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
13733 * config/avr/avr.md: Only post-reload split REG-REG moves if
13734 either register is GENERAL_REG_P.
13736 2018-02-07 Jakub Jelinek <jakub@redhat.com>
13738 PR tree-optimization/84235
13739 * tree-ssa-scopedtables.c
13740 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
13741 if the subtraction is performed in floating point type where NaNs are
13742 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
13743 build 1. Formatting fix.
13745 2018-02-06 Jakub Jelinek <jakub@redhat.com>
13748 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
13749 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
13750 and skip it regardless of bb boundaries. Use CALL_P macro,
13751 don't test INSN_P (insn) together with CALL_P or JUMP_P check
13752 unnecessarily, formatting fix.
13754 2018-02-06 Michael Collison <michael.collison@arm.com>
13756 * config/arm/thumb2.md:
13757 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
13758 (*thumb_mov_notscc): Ditto.
13760 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
13763 * config/rs6000/rs6000.md (su code attribute): Use "u" for
13764 unsigned_fix, not "s".
13766 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13768 * configure.ac (gcc_fn_eh_frame_ro): New function.
13769 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
13770 correct .eh_frame permissions.
13771 * configure: Regenerate.
13773 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
13775 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
13776 irrelevant options.
13778 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13780 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13781 Display warning message for -mno-speculate-indirect-jumps.
13783 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
13785 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
13787 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
13789 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
13791 PR tree-optimization/84225
13792 * tree-eh.c (find_trapping_overflow): Only call
13793 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
13795 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
13798 * config/i386/i386.c: Reimplement the check of possible options
13799 -mibt/-mshstk conbination. Change error messages.
13800 * doc/invoke.texi: Fix a typo: remove extra '='.
13802 2018-02-06 Marek Polacek <polacek@redhat.com>
13804 PR tree-optimization/84228
13805 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
13807 2018-02-06 Tamar Christina <tamar.christina@arm.com>
13810 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
13811 emitted arch directives.
13812 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
13813 __ARM_FEATURE_COPROC before changing architectures.
13815 2018-02-06 Richard Biener <rguenther@suse.de>
13817 * config/i386/i386.c (print_reg): Fix typo.
13818 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
13820 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
13822 * configure: Regenerate.
13824 2018-02-05 Martin Sebor <msebor@redhat.com>
13826 PR tree-optimization/83369
13827 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
13830 2018-02-05 Martin Liska <mliska@suse.cz>
13832 * doc/invoke.texi: Cherry-pick upstream r323995.
13834 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
13836 * ira.c (ira_init_register_move_cost): Adjust comment.
13838 2018-02-05 Martin Liska <mliska@suse.cz>
13840 PR gcov-profile/84137
13841 * doc/gcov.texi: Fix typo in documentation.
13843 2018-02-05 Martin Liska <mliska@suse.cz>
13845 PR gcov-profile/83879
13846 * doc/gcov.texi: Document necessity of --dynamic-list-data when
13847 using dlopen functionality.
13849 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
13851 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
13852 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
13853 _mm_maskz_range_ss, _mm_mask_range_round_ss,
13854 _mm_maskz_range_round_ss): New intrinsics.
13855 (__builtin_ia32_rangesd128_round)
13856 (__builtin_ia32_rangess128_round): Remove.
13857 (__builtin_ia32_rangesd128_mask_round,
13858 __builtin_ia32_rangess128_mask_round): New builtins.
13859 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
13860 __builtin_ia32_rangess128_round): Remove.
13861 (__builtin_ia32_rangesd128_mask_round,
13862 __builtin_ia32_rangess128_mask_round): New builtins.
13863 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
13864 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
13865 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
13866 "<round_saeonly_constraint>")): Changed to ...
13867 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
13868 "<round_saeonly_scalar_constraint>")): ... this.
13869 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
13870 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
13871 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
13872 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
13873 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
13875 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
13877 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
13879 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
13880 Remove all values except native, 8540 and 8548.
13882 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
13884 * config/i386/i386.c (ix86_output_function_return): Pass
13885 INVALID_REGNUM, instead of -1, as invalid register number to
13886 indirect_thunk_name and output_indirect_thunk.
13888 2018-02-02 Julia Koval <julia.koval@intel.com>
13890 * config.gcc: Add -march=icelake.
13891 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
13892 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
13893 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
13894 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
13895 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
13896 (processor_target_table): Add icelake.
13897 (ix86_option_override_internal): Handle new PTAs.
13898 (get_builtin_code_for_version): Handle icelake.
13899 (M_INTEL_COREI7_ICELAKE): New.
13900 (fold_builtin_cpu): Handle icelake.
13901 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
13902 * doc/invoke.texi: Add -march=icelake.
13904 2018-02-02 Julia Koval <julia.koval@intel.com>
13906 * config/i386/i386.c (ix86_option_override_internal): Change flags type
13907 to wide_int_bitmask.
13908 * wide-int-bitmask.h: New.
13910 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
13913 * config/i386/i386.md: Replace Pmode with word_mode in
13914 builtin_setjmp_setup and builtin_longjmp to support x32.
13916 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
13920 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
13922 (rs6000_supported_cpu_names): New static variable.
13923 (linux_cpu_translation_table): Likewise.
13924 (elf_platform) <cpu>: Define new static variable and use it.
13925 Translate kernel AT_PLATFORM name to canonical name if needed.
13926 Error if platform name is unknown.
13928 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
13931 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
13933 2018-02-01 Jeff Law <law@redhat.com>
13936 * config/i386/i386.c (release_scratch_register_on_entry): Add new
13937 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
13938 the scratch if RELEASE_VIA_POP is false.
13939 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
13940 If we have to save a temporary register, decrement SIZE appropriately.
13941 Pass new arguments to release_scratch_register_on_entry.
13942 (ix86_adjust_stack_and_probe): Likewise.
13943 (ix86_emit_probe_stack_range): Pass new arguments to
13944 release_scratch_register_on_entry.
13946 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
13948 PR rtl-optimization/84157
13949 * combine.c (change_zero_ext): Use REG_P predicate in
13950 front of HARD_REGISTER_P predicate.
13952 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
13954 * config/avr/avr.c (avr_option_override): Move disabling of
13955 -fdelete-null-pointer-checks to...
13956 * common/config/avr/avr-common.c (avr_option_optimization_table):
13959 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
13961 PR tree-optimization/81635
13962 * tree-data-ref.c (split_constant_offset_1): For types that
13963 wrap on overflow, try to use range info to prove that wrapping
13966 2018-02-01 Renlin Li <renlin.li@arm.com>
13969 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13970 TAILCALL_ADDR_REGS.
13971 (aarch64_register_move_cost): Likewise.
13972 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
13973 TAILCALL_ADDR_REGS.
13974 (REG_CLASS_NAMES): Likewise.
13975 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
13976 TAILCALL_ADDR_REGS. Remove IP registers.
13977 * config/aarch64/aarch64.md (Ucs): Update register constraint.
13979 2018-02-01 Richard Biener <rguenther@suse.de>
13981 * domwalk.h (dom_walker::dom_walker): Add additional constructor
13982 for specifying RPO order and allow NULL for that.
13983 * domwalk.c (dom_walker::dom_walker): Likewise.
13984 (dom_walker::walk): Handle NULL RPO order.
13985 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
13987 (rewrite_update_dom_walker): Likewise.
13988 (mark_def_dom_walker): Likewise.
13990 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
13992 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
13993 (aarch64_maybe_expand_sve_subreg_move): Declare.
13994 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
13995 * config/aarch64/predicates.md (aarch64_any_register_operand): New
13997 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
13998 that are semantically a reverse operation.
13999 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
14000 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
14001 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
14003 (aarch64_can_change_mode_class): For big-endian, forbid changes
14004 between two SVE modes if they have different element sizes.
14006 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
14008 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
14009 the TImode handling for big-endian targets.
14011 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
14013 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
14014 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
14016 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
14017 Remove BSWAP handing for big-endian targets and use the form of
14018 LD1RQ appropariate for the mode.
14020 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
14022 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
14023 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
14024 duplicated element.
14026 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
14029 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
14030 check for operands that need to go through aarch64_sve_reload_be.
14032 2018-02-01 Jakub Jelinek <jakub@redhat.com>
14034 PR tree-optimization/81661
14035 PR tree-optimization/84117
14036 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
14037 * tree-eh.c: Include gimplify.h.
14038 (find_trapping_overflow, replace_trapping_overflow,
14039 rewrite_to_non_trapping_overflow): New functions.
14040 * tree-vect-loop.c: Include tree-eh.h.
14041 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
14042 * tree-data-ref.c: Include tree-eh.h.
14043 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
14045 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
14047 PR rtl-optimization/84123
14048 * combine.c (change_zero_ext): Check if hard register satisfies
14049 can_change_dest_mode before calling gen_lowpart_SUBREG.
14051 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
14054 * ira.c (ira_init_register_move_cost): Remove assert.
14056 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
14058 PR rtl-optimization/84071
14059 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
14060 * doc/tm.texi: Regenerate.
14062 2018-01-31 Richard Biener <rguenther@suse.de>
14064 PR tree-optimization/84132
14065 * tree-data-ref.c (analyze_miv_subscript): Properly
14066 check whether evolution_function_is_affine_multivariate_p
14067 before calling gcd_of_steps_may_divide_p.
14069 2018-01-31 Julia Koval <julia.koval@intel.com>
14072 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
14073 * config/i386/i386.md (rdpid_rex64) New.
14074 (rdpid): Make 32bit only.
14076 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
14079 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
14080 an IDENTIFIER_NODE for FUNCTION_TYPE's.
14082 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
14085 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
14087 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
14089 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
14091 PR rtl-optimization/84071
14092 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
14093 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
14095 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
14097 * config/arc/arc.c (arc_handle_aux_attribute): New function.
14098 (arc_attribute_table): Add 'aux' attribute.
14099 (arc_in_small_data_p): Consider aux like variables.
14100 (arc_is_aux_reg_p): New function.
14101 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
14102 (arc_get_aux_arg): New function.
14103 (prepare_move_operands): Handle aux-register access.
14104 (arc_handle_aux_attribute): New function.
14105 * doc/extend.texi (ARC Variable attributes): Add subsection.
14107 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
14109 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
14110 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
14111 (arc_attribute_table): Add 'uncached' attribute.
14112 (arc_print_operand): Print '.di' flag for uncached memory
14114 (arc_in_small_data_p): Do not consider for small data the uncached
14116 (arc_is_uncached_mem_p): New function.
14117 * config/arc/predicates.md (compact_store_memory_operand): Check
14118 for uncached memory accesses.
14119 (nonvol_nonimm_operand): Likewise.
14120 * gcc/doc/extend.texi (ARC Type Attribute): New subsection.
14122 2018-01-31 Jakub Jelinek <jakub@redhat.com>
14125 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
14126 falign-loops=): Add Optimization flag.
14128 2018-01-30 Jeff Law <law@redhat.com>
14131 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
14132 INT_REGISTERS_SAVED. Check it prior to calling
14133 get_scratch_register_on_entry.
14134 (ix86_adjust_stack_and_probe): Similarly.
14135 (ix86_emit_probe_stack_range): Similarly.
14136 (ix86_expand_prologue): Corresponding changes.
14138 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14141 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
14142 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
14144 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
14147 * lra-constraints.c (curr_insn_transform): Process AND in the
14150 2018-01-30 Jakub Jelinek <jakub@redhat.com>
14152 PR rtl-optimization/83986
14153 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
14154 dependence against last_pending_memory_flush in addition to
14155 pending_jump_insns.
14157 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
14159 PR tree-optimization/81611
14160 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
14163 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14166 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
14169 2018-01-30 Richard Biener <rguenther@suse.de>
14170 Jakub Jelinek <jakub@redhat.com>
14172 PR tree-optimization/84111
14173 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
14174 inner loops added during recursion, as they don't have up-to-date
14177 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
14180 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
14181 (can_inline_edge_by_limits_p): ... here.
14182 (can_early_inline_edge_p, check_callers,
14183 update_caller_keys, update_callee_keys, recursive_inlining,
14184 add_new_edges_to_heap, speculation_useful_p,
14185 inline_small_functions,
14186 inline_small_functions, flatten_function,
14187 inline_to_all_callers_1): Update.
14189 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
14191 * profile-count.c (profile_count::combine_with_ipa_count): Handle
14194 2018-01-30 Richard Biener <rguenther@suse.de>
14196 PR tree-optimization/83008
14197 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
14198 invariant and constant vector uses in stmts when they need
14199 more than one stmt.
14201 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14204 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
14205 * configure: Regenerate.
14207 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
14209 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
14211 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
14212 Use gen_rtx_REG rather than gen_lowpart.
14214 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
14216 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
14217 rather than 0 when creating partial subregs.
14219 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
14221 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
14224 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
14227 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
14228 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
14229 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
14230 flags. This restores the settings used before the 2017-07-24.
14231 Turning off pre increment/decrement/modify allows IVOPTS to
14232 optimize DF/SF loops where the index is an int.
14234 2018-01-29 Richard Biener <rguenther@suse.de>
14235 Kelvin Nilsen <kelvin@gcc.gnu.org>
14238 * tree-vect-stmts.c (vectorizable_call): Don't call
14239 targetm.vectorize_builtin_md_vectorized_function if callee is
14242 2018-01-22 Carl Love <cel@us.ibm.com>
14244 * doc/extend.tex: Fix typo in second arg in
14245 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
14247 2018-01-29 Richard Biener <rguenther@suse.de>
14249 PR tree-optimization/84086
14250 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
14251 (flush_ssaname_freelist): When SSA names were released reset
14252 the SCEV hash table.
14254 2018-01-29 Richard Biener <rguenther@suse.de>
14256 PR tree-optimization/84057
14257 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
14258 removed paths when removing edges.
14260 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
14262 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
14263 -mfunction-return=@var{choice}.
14265 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
14267 PR diagnostic/84034
14268 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
14269 Handle CR like TAB.
14270 (layout::print_source_line): Likewise.
14271 (test_get_line_width_without_trailing_whitespace): Add test cases.
14273 2018-01-27 Jakub Jelinek <jakub@redhat.com>
14275 PR middle-end/84040
14276 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
14279 2018-01-26 Jim Wilson <jimw@sifive.com>
14281 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
14283 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
14286 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14288 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
14289 and CMP + SUB-immediate -> SUBS.
14291 2018-01-26 Martin Sebor <msebor@redhat.com>
14293 PR tree-optimization/83896
14294 * tree-ssa-strlen.c (get_string_len): Rename...
14295 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
14296 Avoid assuming length is constant.
14297 (handle_char_store): Use HOST_WIDE_INT for string length.
14299 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
14302 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
14303 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
14305 2018-01-26 Richard Biener <rguenther@suse.de>
14307 PR rtl-optimization/84003
14308 * dse.c (record_store): Only record redundant stores when
14309 the earlier store aliases at least all accesses the later one does.
14311 2018-01-26 Jakub Jelinek <jakub@redhat.com>
14313 PR rtl-optimization/83985
14314 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
14315 REG_CFA_RESTORE insns.
14316 (delete_unmarked_insns): Don't ignore separate shrink wrapping
14317 REG_CFA_RESTORE insns here.
14320 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
14321 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
14323 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
14325 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
14326 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
14327 (arc_init): Likewise.
14328 (arc_override_options): Likewise.
14329 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
14331 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
14333 * config/arc/arc.h (TARGET_DBNZ): Define.
14334 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
14335 properly set the tune attribute.
14336 (dbnz): Use TARGET_DBNZ guard.
14337 * config/arc/arc.opt (mtune): Add core3 option.
14339 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
14341 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
14342 recognize new pic like addresses.
14343 (arc_delegitimize_address): Clean up.
14345 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
14347 * config/arc/arc-arches.def: Option mrf16 valid for all
14349 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
14350 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
14351 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
14352 * config/arc/arc-tables.opt: Regenerate.
14353 * config/arc/arc.c (arc_conditional_register_usage): Handle
14354 reduced register file case.
14355 (arc_file_start): Set must have build attributes.
14356 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
14357 mrf16 option value.
14358 * config/arc/arc.opt (mrf16): Add new option.
14359 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
14360 * config/arc/genmultilib.awk: Handle new mrf16 option.
14361 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
14362 * config/arc/t-multilib: Regenerate.
14363 * doc/invoke.texi (ARC Options): Document mrf16 option.
14365 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
14367 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
14368 * config/arc/arc.c (arc_handle_secure_attribute): New function.
14369 (arc_attribute_table): Add 'secure_call' attribute.
14370 (arc_print_operand): Print secure call operand.
14371 (arc_function_ok_for_sibcall): Don't optimize tail calls when
14373 (arc_is_secure_call_p): New function. * config/arc/arc.md
14374 (call_i): Add support for sjli instruction.
14375 (call_value_i): Likewise.
14376 * config/arc/constraints.md (Csc): New constraint.
14378 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
14379 John Eric Martin <John.Martin@emmicro-us.com>
14381 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
14382 * config/arc/arc.c (_arc_jli_section): New struct.
14383 (arc_jli_section): New type.
14384 (rc_jli_sections): New static variable.
14385 (arc_handle_jli_attribute): New function.
14386 (arc_attribute_table): Add jli_always and jli_fixed attribute.
14387 (arc_file_end): New function.
14388 (TARGET_ASM_FILE_END): Define.
14389 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
14390 (arc_add_jli_section): New function.
14391 (jli_call_scan): Likewise.
14392 (arc_reorg): Call jli_call_scan.
14393 (arc_output_addsi): Remove 'S' from printing asm operand.
14394 (arc_is_jli_call_p): New function.
14395 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
14397 (movhi_insn): Likewise.
14398 (movsi_insn): Likewise.
14399 (movsi_set_cc_insn): Likewise.
14400 (loadqi_update): Likewise.
14401 (load_zeroextendqisi_update): Likewise.
14402 (load_signextendqisi_update): Likewise.
14403 (loadhi_update): Likewise.
14404 (load_zeroextendhisi_update): Likewise.
14405 (load_signextendhisi_update): Likewise.
14406 (loadsi_update): Likewise.
14407 (loadsf_update): Likewise.
14408 (movsicc_insn): Likewise.
14409 (bset_insn): Likewise.
14410 (bxor_insn): Likewise.
14411 (bclr_insn): Likewise.
14412 (bmsk_insn): Likewise.
14413 (bicsi3_insn): Likewise.
14414 (cmpsi_cc_c_insn): Likewise.
14415 (movsi_ne): Likewise.
14416 (movsi_cond_exec): Likewise.
14417 (clrsbsi2): Likewise.
14418 (norm_f): Likewise.
14428 (call_i): Remove 'S' asm letter, add jli instruction.
14429 (call_value_i): Likewise.
14430 * config/arc/arc.op (mjli-always): New option.
14431 * config/arc/constraints.md (Cji): New constraint.
14432 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
14434 (subsf3_fpx): Likewise.
14435 (mulsf3_fpx): Likewise.
14436 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
14438 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
14439 function attrbutes.
14440 * doc/invoke.texi (ARC): Document mjli-always option.
14442 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
14444 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
14445 avoid addition with 0 and use incw and decw where possible.
14447 2018-01-26 Richard Biener <rguenther@suse.de>
14449 PR tree-optimization/81082
14450 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
14451 association if it requires casting to unsigned.
14452 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
14453 from fold_plusminus_mult_expr to catch important cases late when
14454 range info is available.
14456 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14458 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
14459 * configure.ac (hidden_linkonce): New test.
14460 * configure: Regenerate.
14461 * config.in: Regenerate.
14463 2018-01-26 Julia Koval <julia.koval@intel.com>
14465 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
14466 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
14467 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
14468 _mm_mask_bitshuffle_epi64_mask): Fix type.
14469 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
14470 USI_FTYPE_V4DI_V4DI_USI): Remove.
14471 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
14472 __builtin_ia32_vpshufbitqmb256_mask,
14473 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
14474 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
14475 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
14477 2018-01-26 Alan Modra <amodra@gmail.com>
14480 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
14481 UNSPEC_VBPERMQ. Sort other unspecs.
14483 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
14485 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
14487 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
14489 PR middle-end/83055
14490 * predict.c (drop_profile): Do not push/pop cfun; update also
14492 (handle_missing_profiles): Fix logic looking for zero profiles.
14494 2018-01-25 Jakub Jelinek <jakub@redhat.com>
14496 PR middle-end/83977
14497 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
14498 on functions with #pragma omp declare simd or functions with simd
14500 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
14501 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
14502 Remove trailing \n from warning_at calls.
14504 2018-01-25 Tom de Vries <tom@codesourcery.com>
14507 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
14508 for neutered workers.
14510 2018-01-24 Joseph Myers <joseph@codesourcery.com>
14513 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
14514 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
14516 2018-01-24 Jeff Law <law@redhat.com>
14519 * i386.c (get_probe_interval): Move to earlier point.
14520 (ix86_compute_frame_layout): If -fstack-clash-protection and
14521 the frame is larger than the probe interval, then use pushes
14522 to save registers rather than reg->mem moves.
14523 (ix86_expand_prologue): Remove conditional for int_registers_saved
14526 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
14529 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
14530 min/max for never referenced object.
14532 2018-01-24 Jakub Jelinek <jakub@redhat.com>
14534 PR middle-end/83977
14535 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
14537 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
14538 attributes from DECL_ATTRIBUTES (decl) without affecting
14539 DECL_ATTRIBUTES (current_function_decl).
14540 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
14541 functions with non-NULL DECL_ABSTRACT_ORIGIN.
14543 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
14545 PR tree-optimization/83979
14546 * fold-const.c (fold_comparison): Use constant_boolean_node
14547 instead of boolean_{true,false}_node.
14549 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
14551 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
14554 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14556 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
14557 Simplify the clause that sets the length attribute.
14558 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
14559 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
14560 clause that sets the length attribute.
14561 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
14563 2018-01-24 Tom de Vries <tom@codesourcery.com>
14566 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
14567 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
14568 Add strict parameter.
14569 (prevent_branch_around_nothing): Insert dummy insn between branch to
14570 label and label with no ptx insn inbetween.
14571 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
14573 2018-01-24 Tom de Vries <tom@codesourcery.com>
14576 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
14577 for neutered threads in warp.
14578 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
14580 2018-01-24 Richard Biener <rguenther@suse.de>
14582 PR tree-optimization/83176
14583 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
14586 2018-01-24 Richard Biener <rguenther@suse.de>
14588 PR tree-optimization/82819
14589 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
14590 code generating pluses that are no-ops in the target precision.
14592 2018-01-24 Richard Biener <rguenther@suse.de>
14594 PR middle-end/84000
14595 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
14597 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
14599 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
14600 to merge probabilities.
14601 * predict.c (probably_never_executed): Also mark as cold functions
14602 with global 0 profile and guessed local profile.
14603 * profile-count.c (profile_probability::combine_with_count): New
14605 * profile-count.h (profile_probability::operator*,
14606 profile_probability::operator*=, profile_probability::operator/,
14607 profile_probability::operator/=): Reduce precision to adjusted
14608 and set value to guessed on contradictory divisions.
14609 (profile_probability::combine_with_freq): Remove.
14610 (profile_probability::combine_wiht_count): Declare.
14611 (profile_count::force_nonzero):: Set to adjusted.
14612 (profile_count::probability_in):: Set quality to adjusted.
14613 * tree-ssa-tail-merge.c (replace_block_by): Use
14614 combine_with_count.
14616 2018-01-23 Andrew Waterman <andrew@sifive.com>
14617 Jim Wilson <jimw@sifive.com>
14619 * config/riscv/riscv.c (riscv_stack_boundary): New.
14620 (riscv_option_override): Set riscv_stack_boundary. Handle
14621 riscv_preferred_stack_boundary_arg.
14622 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
14623 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
14624 (STACK_BOUNDARY): Set to riscv_stack_boundary.
14625 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
14626 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
14627 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
14629 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
14632 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
14633 of struct ix86_frame.
14634 (ix86_expand_epilogue): Likewise. Add a local variable for
14635 the reg_save_offset field in struct ix86_frame.
14637 2018-01-23 Bin Cheng <bin.cheng@arm.com>
14639 PR tree-optimization/82604
14640 * tree-loop-distribution.c (enum partition_kind): New enum item
14641 PKIND_PARTIAL_MEMSET.
14642 (partition_builtin_p): Support above new enum item.
14643 (generate_code_for_partition): Ditto.
14644 (compute_access_range): Differentiate cases that equality can be
14645 proven at all loops, the innermost loops or no loops.
14646 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
14647 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
14648 (finalize_partitions, distribute_loop): Don't fuse partition of
14649 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
14650 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
14651 parloop is enabled.
14653 2018-01-23 Martin Liska <mliska@suse.cz>
14655 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
14656 order to ignore the predictor.
14657 (PRED_POLYMORPHIC_CALL): Likewise.
14658 (PRED_RECURSIVE_CALL): Likewise.
14660 2018-01-23 Martin Liska <mliska@suse.cz>
14662 * tree-profile.c (tree_profiling): Print function header to
14663 aware reader which function we are working on.
14664 * value-prof.c (gimple_find_values_to_profile): Do not print
14665 not interesting value histograms.
14667 2018-01-23 Martin Liska <mliska@suse.cz>
14669 * profile-count.h (enum profile_quality): Add
14670 profile_uninitialized as the first value. Do not number values
14671 as they are zero based.
14672 (profile_count::verify): Update sanity check.
14673 (profile_probability::verify): Likewise.
14675 2018-01-23 Nathan Sidwell <nathan@acm.org>
14677 * doc/invoke.texi (ffor-scope): Deprecate.
14679 2018-01-23 David Malcolm <dmalcolm@redhat.com>
14681 PR tree-optimization/83510
14682 * domwalk.c (set_all_edges_as_executable): New function.
14683 (dom_walker::dom_walker): Convert bool param
14684 "skip_unreachable_blocks" to enum reachability. Move setup of
14685 edge flags to set_all_edges_as_executable and only do it when
14686 reachability is REACHABLE_BLOCKS.
14687 * domwalk.h (enum dom_walker::reachability): New enum.
14688 (dom_walker::dom_walker): Convert bool param
14689 "skip_unreachable_blocks" to enum reachability.
14690 (set_all_edges_as_executable): New decl.
14691 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
14692 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
14694 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
14695 but converting true to REACHABLE_BLOCKS.
14696 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
14698 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
14699 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
14700 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
14702 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
14703 if check_all_array_refs will be called.
14705 2018-01-23 David Malcolm <dmalcolm@redhat.com>
14707 * tree.c (selftest::test_location_wrappers): Add more test
14710 2018-01-23 David Malcolm <dmalcolm@redhat.com>
14712 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
14713 (selftest::test_bit_in_range): Likewise.
14715 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
14718 * doc/sourcebuild.texi (vect_float): Say that the selector
14719 only describes the situation when -funsafe-math-optimizations is on.
14720 (vect_float_strict): Document.
14722 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
14724 PR tree-optimization/83965
14725 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
14726 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
14727 instead of checking only for a reduction.
14728 (vect_recog_widen_sum_pattern): Likewise.
14730 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
14732 * predict.c (probably_never_executed): Only use precise profile info.
14733 (compute_function_frequency): Skip after inlining hack since we now
14734 have quality checking.
14736 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
14738 * profile-count.h (profile_probability::very_unlikely,
14739 profile_probability::unlikely, profile_probability::even): Set
14740 precision to guessed.
14742 2018-01-23 Richard Biener <rguenther@suse.de>
14744 PR tree-optimization/83963
14745 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
14746 Properly terminate dominator walk when crossing the exit edge not
14747 when visiting its source block.
14749 2018-01-23 Jakub Jelinek <jakub@redhat.com>
14752 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
14753 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
14755 2018-01-22 Jakub Jelinek <jakub@redhat.com>
14757 PR tree-optimization/83957
14758 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
14759 semicolon after for body surrounded by braces.
14761 PR tree-optimization/83081
14762 * profile-count.h (profile_probability::split): New method.
14763 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
14764 Use profile_probability::split.
14765 (do_compare_rtx_and_jump): Fix adjustment of probabilities
14766 when splitting a single conditional jump into 2.
14768 2018-01-22 David Malcolm <dmalcolm@redhat.com>
14770 PR tree-optimization/69452
14771 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
14774 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
14776 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
14777 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
14778 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
14780 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
14782 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
14784 * config/rl78/rl78.md (movdi): New define_expand.
14785 * config/rl78/rl78.c (rl78_split_movdi): New function.
14787 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
14790 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
14792 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
14793 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
14794 128-bit to produce an UNSPEC move to get the double word with the
14795 signbit and then a shift directly to do signbit.
14796 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
14797 implementation with a new version that just does either a direct
14798 move or a regular move. Move memory interface to separate insns.
14799 Move insns so they are next to the expander.
14800 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
14801 with signbit move. Split big and little endian case.
14802 (signbit<mode>2_dm_mem_le): Likewise.
14803 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
14804 (signbit<mode>2_dm2): Likewise.
14806 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
14808 * config/rl78/rl78.md (anddi3): New define_expand.
14810 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
14812 * config/rl78/rl78.md (umindi3): New define_expand.
14814 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
14816 * config/rl78/rl78.md (smindi3): New define_expand.
14818 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
14820 * config/rl78/rl78.md (smaxdi3): New define_expand.
14822 2018-01-22 Carl Love <cel@us.ibm.com>
14824 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
14825 LVX_V1TI): Add macro expansion.
14826 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
14827 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
14828 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
14829 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
14830 Change check to determine if the instruction is a byte reversing
14831 entry. Fix typo in comment.
14832 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
14833 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
14834 Add def_builtin calls for new builtins.
14835 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
14836 Add define_insn expansion.
14838 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
14840 * config/rl78/rl78.md (umaxdi3): New define_expand.
14842 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
14844 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
14845 for non-QImode registers.
14847 2018-01-22 Richard Biener <rguenther@suse.de>
14849 PR tree-optimization/83963
14850 * graphite-scop-detection.c (scop_detection::get_sese): Delay
14851 including the loop exit block.
14852 (scop_detection::merge_sese): Likewise.
14853 (scop_detection::add_scop): Do it here instead.
14855 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14857 * doc/sourcebuild.texi (arm_softfloat): Document.
14859 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
14862 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
14863 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
14864 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
14866 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14867 David Edelsohn <dje.gcc@gmail.com>
14870 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
14871 Change "crset eq" to "crset 2".
14872 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
14873 (*call_indirect_aix<mode>_nospec): Likewise.
14874 (*call_value_indirect_aix<mode>_nospec): Likewise.
14875 (*call_indirect_elfv2<mode>_nospec): Likewise.
14876 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
14877 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
14878 change assembly output from . to $.
14879 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
14880 (indirect_jump<mode>_nospec): Change assembly output from . to $.
14881 (*tablejump<mode>_internal1_nospec): Likewise.
14883 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
14886 * config/sh/sh_optimize_sett_clrt.cc:
14887 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
14889 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
14891 PR tree-optimization/83940
14892 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
14893 offset_dt to vect_constant_def rather than vect_unknown_def_type.
14894 (vect_check_load_store_mask): Add a mask_dt_out parameter and
14895 use it to pass back the definition type.
14896 (vect_check_store_rhs): Likewise rhs_dt_out.
14897 (vect_build_gather_load_calls): Add a mask_dt argument and use
14898 it instead of a call to vect_is_simple_use.
14899 (vectorizable_store): Update calls to vect_check_load_store_mask
14900 and vect_check_store_rhs. Use the dt returned by the latter instead
14901 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
14902 instead of calls to vect_is_simple_use. Pass the scalar rather
14903 than the vector operand to vect_is_simple_use when handling
14904 second and subsequent copies of an rhs value.
14905 (vectorizable_load): Update calls to vect_check_load_store_mask
14906 and vect_build_gather_load_calls. Use the cached mask_dt and
14907 gs_info.offset_dt instead of calls to vect_is_simple_use.
14909 2018-01-20 Jakub Jelinek <jakub@redhat.com>
14911 PR middle-end/83945
14912 * tree-emutls.c: Include gimplify.h.
14913 (lower_emutls_2): New function.
14914 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
14915 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
14916 it before further processing.
14919 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
14920 UINTVAL (trueop1) instead of INTVAL (op1).
14922 2018-01-19 Jakub Jelinek <jakub@redhat.com>
14926 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
14927 INCOMING_FRAME_SP_OFFSET if not defined.
14928 (scan_trace): Add ENTRY argument. If true and
14929 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
14930 emit a note to adjust the CFA offset.
14931 (create_cfi_notes): Adjust scan_trace callers.
14932 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
14933 INCOMING_FRAME_SP_OFFSET in the CIE.
14934 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
14935 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
14937 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
14938 * doc/tm.texi: Regenerated.
14940 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14942 PR rtl-optimization/83147
14943 * lra-constraints.c (remove_inheritance_pseudos): Use
14944 lra_substitute_pseudo_within_insn.
14946 2018-01-19 Tom de Vries <tom@codesourcery.com>
14947 Cesar Philippidis <cesar@codesourcery.com>
14950 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
14952 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
14955 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
14956 spaces for function labels.
14958 2018-01-19 Martin Liska <mliska@suse.cz>
14960 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
14961 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
14962 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
14963 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
14964 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
14965 (PRED_CONST_RETURN): Change from 69 to 65.
14966 (PRED_NULL_RETURN): Change from 91 to 71.
14967 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
14968 (PRED_LOOP_GUARD): Change from 66 to 73.
14970 2018-01-19 Martin Liska <mliska@suse.cz>
14972 * predict.c (predict_insn_def): Add new assert.
14973 (struct branch_predictor): Change type to signed integer.
14974 (test_prediction_value_range): Amend test to cover
14975 PROB_UNINITIALIZED.
14976 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
14977 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
14978 (PRED_LOOP_ITERATIONS_MAX): Likewise.
14979 (PRED_LOOP_IV_COMPARE): Likewise.
14980 * predict.h (PROB_UNINITIALIZED): Define new constant.
14982 2018-01-19 Martin Liska <mliska@suse.cz>
14984 * predict.c (dump_prediction): Add new format for
14985 analyze_brprob.py script which is enabled with -details
14987 * profile-count.h (precise_p): New function.
14989 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
14991 PR tree-optimization/83922
14992 * tree-vect-loop.c (vect_verify_full_masking): Return false if
14993 there are no statements that need masking.
14994 (vect_active_double_reduction_p): New function.
14995 (vect_analyze_loop_operations): Use it when handling phis that
14996 are not in the loop header.
14998 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
15000 PR tree-optimization/83914
15001 * tree-vect-loop.c (vectorizable_induction): Don't convert
15002 init_expr or apply the peeling adjustment for inductions
15003 that are nested within the vectorized loop.
15005 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15007 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
15010 2018-01-18 Jakub Jelinek <jakub@redhat.com>
15014 * function.h (gimplify_parameters): Add gimple_seq * argument.
15015 * function.c: Include gimple.h and options.h.
15016 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
15017 for the added local temporaries if needed.
15018 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
15019 if there are any parameter cleanups, wrap whole body into a
15020 try/finally with the cleanups.
15022 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
15025 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
15026 Use GET_MODE_CLASS for scalar floating point.
15028 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
15032 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
15033 Fix call of call_cgraph_insertion_hooks.
15035 2018-01-18 Martin Sebor <msebor@redhat.com>
15037 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
15039 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
15042 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
15045 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
15048 * common.opt: (-ffile-prefix-map): New option.
15049 * opts.c (common_handle_option): Defer it.
15050 * opts-global.c (handle_common_deferred_options): Handle it.
15051 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
15052 * file-prefix-map.h: New file.
15053 (remap_debug_filename, add_debug_prefix_map): ...here.
15054 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
15055 * final.c (debug_prefix_map, add_debug_prefix_map
15056 remap_debug_filename): Move to...
15057 * file-prefix-map.c: New file.
15058 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
15059 generalize, get rid of alloca(), use strrchr() instead of strchr().
15060 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
15061 Implement in terms of add_prefix_map().
15062 (remap_macro_filename, remap_debug_filename): Implement in term of
15064 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
15065 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
15066 * dbxout.c: Include file-prefix-map.h.
15067 * varasm.c: Likewise.
15068 * vmsdbgout.c: Likewise.
15069 * xcoffout.c: Likewise.
15070 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
15071 * doc/cppopts.texi (-fmacro-prefix-map): Document.
15072 * doc/invoke.texi (-ffile-prefix-map): Document.
15073 (-fdebug-prefix-map): Update description.
15075 2018-01-18 Martin Liska <mliska@suse.cz>
15077 * config/i386/i386.c (indirect_thunk_name): Document that also
15079 (output_indirect_thunk): Document why both instructions
15080 (pause and lfence) are generated.
15082 2018-01-18 Richard Biener <rguenther@suse.de>
15084 PR tree-optimization/83887
15085 * graphite-scop-detection.c
15086 (scop_detection::get_nearest_dom_with_single_entry): Remove.
15087 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
15088 (scop_detection::merge_sese): Re-implement with a flood-fill
15089 algorithm that properly finds a SESE region if it exists.
15091 2018-01-18 Jakub Jelinek <jakub@redhat.com>
15094 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
15095 pointer_diff optimizations use view_convert instead of convert.
15097 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15099 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
15100 Generate different code for -mno-speculate-indirect-jumps.
15101 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
15102 (*call_indirect_aix<mode>): Disable for
15103 -mno-speculate-indirect-jumps.
15104 (*call_indirect_aix<mode>_nospec): New define_insn.
15105 (*call_value_indirect_aix<mode>): Disable for
15106 -mno-speculate-indirect-jumps.
15107 (*call_value_indirect_aix<mode>_nospec): New define_insn.
15108 (*sibcall_nonlocal_sysv<mode>): Generate different code for
15109 -mno-speculate-indirect-jumps.
15110 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
15112 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
15114 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
15115 long double type, set the flags for noting the default long double
15116 type, even if we don't pass or return a long double type.
15118 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
15121 * ipa-inline.c (flatten_function): Do not overwrite final inlining
15124 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
15126 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
15127 support for merge[hl].
15128 (fold_mergehl_helper): New helper function.
15129 (tree-vector-builder.h): New #include for tree_vector_builder usage.
15130 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
15131 (altivec_vmrglw_direct): Add xxmrglw insn.
15133 2018-01-17 Andrew Waterman <andrew@sifive.com>
15135 * config/riscv/riscv.c (riscv_conditional_register_usage): If
15136 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
15138 2018-01-17 David Malcolm <dmalcolm@redhat.com>
15141 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
15142 call the lto_location_cache before reading the
15143 DECL_SOURCE_LOCATION of the types.
15145 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
15146 Richard Sandiford <richard.sandiford@linaro.org>
15148 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
15149 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
15150 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
15151 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
15152 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
15154 * config/aarch64/constraints.md (aarch64_movti_operand):
15156 * config/aarch64/predicates.md (Uti): Add new constraint.
15158 2018-01-17 Carl Love <cel@us.ibm.com>
15160 * config/rs6000/vsx.md (define_expand xl_len_r,
15161 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
15162 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
15164 (define_expand, define_insn): Move the shift left from the
15165 define_insn to the define_expand for lxvl and stxvl instructions.
15166 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
15167 and XL_LEN_R definitions to PURE.
15169 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
15171 * config/i386/i386.c (indirect_thunk_name): Declare regno
15172 as unsigned int. Compare regno with INVALID_REGNUM.
15173 (output_indirect_thunk): Ditto.
15174 (output_indirect_thunk_function): Ditto.
15175 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
15176 in the call to output_indirect_thunk_function.
15178 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
15180 PR middle-end/83884
15181 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
15182 rather than the size of inner_type to determine the stack slot size
15183 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
15185 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
15188 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
15191 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
15193 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
15194 endian Linux systems to optionally enable multilibs for selecting
15195 the long double type if the user configured an explicit type.
15196 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
15197 have no long double multilibs if not defined.
15198 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
15199 warn if the user used -mabi={ieee,ibm}longdouble and we built
15200 multilibs for long double.
15201 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
15202 appropriate multilib option.
15203 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
15205 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
15206 for building long double multilibs.
15207 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
15209 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
15211 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
15214 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
15216 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
15219 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
15222 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
15225 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
15227 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
15228 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
15230 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
15232 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
15233 different rtl trees depending on TARGET_64BIT.
15234 (rs6000_gen_lvx): Likewise.
15236 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
15238 * config/visium/visium.md (nop): Tweak comment.
15239 (hazard_nop): Likewise.
15241 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15243 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
15244 -mspeculate-indirect-jumps.
15245 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
15246 for -mno-speculate-indirect-jumps.
15247 (*call_indirect_elfv2<mode>_nospec): New define_insn.
15248 (*call_value_indirect_elfv2<mode>): Disable for
15249 -mno-speculate-indirect-jumps.
15250 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
15251 (indirect_jump): Emit different RTL for
15252 -mno-speculate-indirect-jumps.
15253 (*indirect_jump<mode>): Disable for
15254 -mno-speculate-indirect-jumps.
15255 (*indirect_jump<mode>_nospec): New define_insn.
15256 (tablejump): Emit different RTL for
15257 -mno-speculate-indirect-jumps.
15258 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
15259 (tablejumpsi_nospec): New define_expand.
15260 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
15261 (tablejumpdi_nospec): New define_expand.
15262 (*tablejump<mode>_internal1): Disable for
15263 -mno-speculate-indirect-jumps.
15264 (*tablejump<mode>_internal1_nospec): New define_insn.
15265 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
15268 2018-01-16 Artyom Skrobov tyomitch@gmail.com
15270 * caller-save.c (insert_save): Drop unnecessary parameter. All
15273 2018-01-16 Jakub Jelinek <jakub@redhat.com>
15274 Richard Biener <rguenth@suse.de>
15277 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
15278 return early, inline manually is_gimple_sizepos. Make sure if we
15279 call gimplify_expr we don't end up with a gimple constant.
15280 * tree.c (variably_modified_type_p): Don't return true for
15281 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
15282 * gimplify.h (is_gimple_sizepos): Remove.
15284 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
15286 PR tree-optimization/83857
15287 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
15288 vectorizable_live_operation for pure SLP statements.
15289 (vectorizable_live_operation): Handle PHIs.
15291 2018-01-16 Richard Biener <rguenther@suse.de>
15293 PR tree-optimization/83867
15294 * tree-vect-stmts.c (vect_transform_stmt): Precompute
15295 nested_in_vect_loop_p since the scalar stmt may get invalidated.
15297 2018-01-16 Jakub Jelinek <jakub@redhat.com>
15300 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
15301 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
15302 If off is not INTEGER_CST, issue a may not be aligned warning
15303 rather than isn't aligned. Use isn%'t rather than isn't.
15304 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
15306 <case MULT_EXPR>: Improve the case when bottom and one of the
15307 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
15308 operand, in that case check if the other operand is multiple of
15309 bottom divided by the INTEGER_CST operand.
15311 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
15314 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
15315 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
15316 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
15317 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
15318 * config/pa/pa.c (pa_function_arg_advance): Likewise.
15319 (pa_function_arg, pa_arg_partial_bytes): Likewise.
15320 (pa_function_arg_size): New function.
15322 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
15324 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
15325 in a separate statement.
15327 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
15329 PR tree-optimization/83847
15330 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
15331 group gathers and scatters.
15333 2018-01-16 Jakub Jelinek <jakub@redhat.com>
15335 PR rtl-optimization/86620
15336 * params.def (max-sched-ready-insns): Bump minimum value to 1.
15338 PR rtl-optimization/83213
15339 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
15340 to last if both are JUMP_INSNs.
15342 PR tree-optimization/83843
15343 * gimple-ssa-store-merging.c
15344 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
15345 store_immediate_info for bswap/nop orig_stores.
15347 2018-01-15 Andrew Waterman <andrew@sifive.com>
15349 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
15351 <UDIV>: Increase cost if !TARGET_DIV.
15353 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
15355 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
15356 (define_attr "cr_logical_3op"): New.
15357 (cceq_ior_compare): Adjust.
15358 (cceq_ior_compare_complement): Adjust.
15359 (*cceq_rev_compare): Adjust.
15360 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
15361 (is_cracked_insn): Adjust.
15362 (insn_must_be_first_in_group): Adjust.
15363 * config/rs6000/40x.md: Adjust.
15364 * config/rs6000/440.md: Adjust.
15365 * config/rs6000/476.md: Adjust.
15366 * config/rs6000/601.md: Adjust.
15367 * config/rs6000/603.md: Adjust.
15368 * config/rs6000/6xx.md: Adjust.
15369 * config/rs6000/7450.md: Adjust.
15370 * config/rs6000/7xx.md: Adjust.
15371 * config/rs6000/8540.md: Adjust.
15372 * config/rs6000/cell.md: Adjust.
15373 * config/rs6000/e300c2c3.md: Adjust.
15374 * config/rs6000/e500mc.md: Adjust.
15375 * config/rs6000/e500mc64.md: Adjust.
15376 * config/rs6000/e5500.md: Adjust.
15377 * config/rs6000/e6500.md: Adjust.
15378 * config/rs6000/mpc.md: Adjust.
15379 * config/rs6000/power4.md: Adjust.
15380 * config/rs6000/power5.md: Adjust.
15381 * config/rs6000/power6.md: Adjust.
15382 * config/rs6000/power7.md: Adjust.
15383 * config/rs6000/power8.md: Adjust.
15384 * config/rs6000/power9.md: Adjust.
15385 * config/rs6000/rs64.md: Adjust.
15386 * config/rs6000/titan.md: Adjust.
15388 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
15390 * config/i386/predicates.md (indirect_branch_operand): Rewrite
15391 ix86_indirect_branch_register logic.
15393 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
15395 * config/i386/constraints.md (Bs): Update
15396 ix86_indirect_branch_register check. Don't check
15397 ix86_indirect_branch_register with GOT_memory_operand.
15399 * config/i386/predicates.md (GOT_memory_operand): Don't check
15400 ix86_indirect_branch_register here.
15401 (GOT32_symbol_operand): Likewise.
15403 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
15405 * config/i386/predicates.md (constant_call_address_operand):
15406 Rewrite ix86_indirect_branch_register logic.
15407 (sibcall_insn_operand): Likewise.
15409 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
15411 * config/i386/constraints.md (Bs): Replace
15412 ix86_indirect_branch_thunk_register with
15413 ix86_indirect_branch_register.
15415 * config/i386/i386.md (indirect_jump): Likewise.
15416 (tablejump): Likewise.
15417 (*sibcall_memory): Likewise.
15418 (*sibcall_value_memory): Likewise.
15419 Peepholes of indirect call and jump via memory: Likewise.
15420 * config/i386/i386.opt: Likewise.
15421 * config/i386/predicates.md (indirect_branch_operand): Likewise.
15422 (GOT_memory_operand): Likewise.
15423 (call_insn_operand): Likewise.
15424 (sibcall_insn_operand): Likewise.
15425 (GOT32_symbol_operand): Likewise.
15427 2018-01-15 Jakub Jelinek <jakub@redhat.com>
15429 PR middle-end/83837
15430 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
15431 type rather than type addr's type points to.
15432 (expand_omp_atomic_mutex): Likewise.
15433 (expand_omp_atomic): Likewise.
15435 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
15438 * config/i386/i386.c (output_indirect_thunk_function): Use
15439 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
15440 for __x86_return_thunk.
15442 2018-01-15 Richard Biener <rguenther@suse.de>
15444 PR middle-end/83850
15445 * expmed.c (extract_bit_field_1): Fix typo.
15447 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15450 * config/arm/iterators.md (VF): New mode iterator.
15451 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
15452 Remove integer-related logic from pattern.
15453 (neon_vabd<mode>_3): Likewise.
15455 2018-01-15 Jakub Jelinek <jakub@redhat.com>
15457 PR middle-end/82694
15458 * common.opt (fstrict-overflow): No longer an alias.
15459 (fwrapv-pointer): New option.
15460 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
15461 also for pointer types based on flag_wrapv_pointer.
15462 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
15463 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
15464 opts->x_flag_wrapv got set.
15465 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
15466 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
15467 POINTER_TYPE_OVERFLOW_UNDEFINED.
15468 * match.pd: Likewise in address comparison pattern.
15469 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
15471 2018-01-15 Richard Biener <rguenther@suse.de>
15474 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
15475 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
15476 Reset type names to their identifier if their TYPE_DECL doesn't
15477 have linkage (and thus is used for ODR and devirt).
15478 (save_debug_info_for_decl): Remove.
15479 (save_debug_info_for_type): Likewise.
15480 (add_tree_to_fld_list): Adjust.
15481 * tree-pretty-print.c (dump_generic_node): Make dumping of
15482 type names more robust.
15484 2018-01-15 Richard Biener <rguenther@suse.de>
15486 * BASE-VER: Bump to 8.0.1.
15488 2018-01-14 Martin Sebor <msebor@redhat.com>
15491 * builtins.c (check_access): Avoid warning when the no-warning bit
15494 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
15496 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
15497 * ira-color (allocno_hard_regs_compare): Likewise.
15499 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
15502 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
15503 Use .pushsection/.popsection.
15505 2018-01-14 Martin Sebor <msebor@redhat.com>
15508 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
15510 2018-01-14 Jakub Jelinek <jakub@redhat.com>
15512 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
15513 entry from extra_headers.
15514 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
15515 extra_headers, make the list bitwise identical to the i?86-*-* one.
15517 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
15519 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
15520 -mcmodel=large with -mindirect-branch=thunk,
15521 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
15522 -mfunction-return=thunk-extern.
15523 * doc/invoke.texi: Document -mcmodel=large is incompatible with
15524 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
15525 -mfunction-return=thunk and -mfunction-return=thunk-extern.
15527 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
15529 * config/i386/i386.c (print_reg): Print the name of the full
15530 integer register without '%'.
15531 (ix86_print_operand): Handle 'V'.
15532 * doc/extend.texi: Document 'V' modifier.
15534 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
15536 * config/i386/constraints.md (Bs): Disallow memory operand for
15537 -mindirect-branch-register.
15539 * config/i386/predicates.md (indirect_branch_operand): Likewise.
15540 (GOT_memory_operand): Likewise.
15541 (call_insn_operand): Likewise.
15542 (sibcall_insn_operand): Likewise.
15543 (GOT32_symbol_operand): Likewise.
15544 * config/i386/i386.md (indirect_jump): Call convert_memory_address
15545 for -mindirect-branch-register.
15546 (tablejump): Likewise.
15547 (*sibcall_memory): Likewise.
15548 (*sibcall_value_memory): Likewise.
15549 Disallow peepholes of indirect call and jump via memory for
15550 -mindirect-branch-register.
15551 (*call_pop): Replace m with Bw.
15552 (*call_value_pop): Likewise.
15553 (*sibcall_pop_memory): Replace m with Bs.
15554 * config/i386/i386.opt (mindirect-branch-register): New option.
15555 * doc/invoke.texi: Document -mindirect-branch-register option.
15557 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
15559 * config/i386/i386-protos.h (ix86_output_function_return): New.
15560 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
15561 set function_return_type.
15562 (indirect_thunk_name): Add ret_p to indicate thunk for function
15564 (output_indirect_thunk_function): Pass false to
15565 indirect_thunk_name.
15566 (ix86_output_indirect_branch_via_reg): Likewise.
15567 (ix86_output_indirect_branch_via_push): Likewise.
15568 (output_indirect_thunk_function): Create alias for function
15569 return thunk if regno < 0.
15570 (ix86_output_function_return): New function.
15571 (ix86_handle_fndecl_attribute): Handle function_return.
15572 (ix86_attribute_table): Add function_return.
15573 * config/i386/i386.h (machine_function): Add
15574 function_return_type.
15575 * config/i386/i386.md (simple_return_internal): Use
15576 ix86_output_function_return.
15577 (simple_return_internal_long): Likewise.
15578 * config/i386/i386.opt (mfunction-return=): New option.
15579 (indirect_branch): Mention -mfunction-return=.
15580 * doc/extend.texi: Document function_return function attribute.
15581 * doc/invoke.texi: Document -mfunction-return= option.
15583 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
15585 * config/i386/i386-opts.h (indirect_branch): New.
15586 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
15587 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
15588 with local indirect jump when converting indirect call and jump.
15589 (ix86_set_indirect_branch_type): New.
15590 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
15591 (indirectlabelno): New.
15592 (indirect_thunk_needed): Likewise.
15593 (indirect_thunk_bnd_needed): Likewise.
15594 (indirect_thunks_used): Likewise.
15595 (indirect_thunks_bnd_used): Likewise.
15596 (INDIRECT_LABEL): Likewise.
15597 (indirect_thunk_name): Likewise.
15598 (output_indirect_thunk): Likewise.
15599 (output_indirect_thunk_function): Likewise.
15600 (ix86_output_indirect_branch_via_reg): Likewise.
15601 (ix86_output_indirect_branch_via_push): Likewise.
15602 (ix86_output_indirect_branch): Likewise.
15603 (ix86_output_indirect_jmp): Likewise.
15604 (ix86_code_end): Call output_indirect_thunk_function if needed.
15605 (ix86_output_call_insn): Call ix86_output_indirect_branch if
15607 (ix86_handle_fndecl_attribute): Handle indirect_branch.
15608 (ix86_attribute_table): Add indirect_branch.
15609 * config/i386/i386.h (machine_function): Add indirect_branch_type
15610 and has_local_indirect_jump.
15611 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
15613 (tablejump): Likewise.
15614 (*indirect_jump): Use ix86_output_indirect_jmp.
15615 (*tablejump_1): Likewise.
15616 (simple_return_indirect_internal): Likewise.
15617 * config/i386/i386.opt (mindirect-branch=): New option.
15618 (indirect_branch): New.
15621 (thunk-inline): Likewise.
15622 (thunk-extern): Likewise.
15623 * doc/extend.texi: Document indirect_branch function attribute.
15624 * doc/invoke.texi: Document -mindirect-branch= option.
15626 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
15629 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
15631 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
15633 * ipa-inline.c (want_inline_small_function_p): Return false if
15634 inlining has already failed with CIF_FINAL_ERROR.
15635 (update_caller_keys): Call want_inline_small_function_p before
15637 (update_callee_keys): Likewise.
15639 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
15641 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
15643 (rs6000_quadword_masked_address_p): Likewise.
15644 (quad_aligned_load_p): Likewise.
15645 (quad_aligned_store_p): Likewise.
15646 (const_load_sequence_p): Add comment to describe the outer-most loop.
15647 (mimic_memory_attributes_and_flags): New function.
15648 (rs6000_gen_stvx): Likewise.
15649 (replace_swapped_aligned_store): Likewise.
15650 (rs6000_gen_lvx): Likewise.
15651 (replace_swapped_aligned_load): Likewise.
15652 (replace_swapped_load_constant): Capitalize argument name in
15653 comment describing this function.
15654 (rs6000_analyze_swaps): Add a third pass to search for vector loads
15655 and stores that access quad-word aligned addresses and replace
15656 with stvx or lvx instructions when appropriate.
15657 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
15658 New function prototype.
15659 (rs6000_quadword_masked_address_p): Likewise.
15660 (rs6000_gen_lvx): Likewise.
15661 (rs6000_gen_stvx): Likewise.
15662 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
15663 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
15664 when memory address is aligned.
15665 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
15666 this split to select lvx instruction when memory address is aligned.
15667 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
15668 instruction when memory address is aligned.
15669 (*vsx_le_perm_load_v16qi): Likewise.
15670 (four unnamed splitters): Modify to select the stvx instruction
15671 when memory is aligned.
15673 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
15675 * predict.c (determine_unlikely_bbs): Handle correctly BBs
15676 which appears in the queue multiple times.
15678 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15679 Alan Hayward <alan.hayward@arm.com>
15680 David Sherwood <david.sherwood@arm.com>
15682 * tree-vectorizer.h (vec_lower_bound): New structure.
15683 (_loop_vec_info): Add check_nonzero and lower_bounds.
15684 (LOOP_VINFO_CHECK_NONZERO): New macro.
15685 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
15686 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
15687 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
15688 fields. Make seg_len the distance travelled, not including the
15690 (dr_direction_indicator): Declare.
15691 (dr_zero_step_indicator): Likewise.
15692 (dr_known_forward_stride_p): Likewise.
15693 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
15695 (runtime_alias_check_p): Allow runtime alias checks with
15697 (operator ==): Compare access_size and align.
15698 (prune_runtime_alias_test_list): Rework for new distinction between
15699 the access_size and seg_len.
15700 (create_intersect_range_checks_index): Likewise. Cope with polynomial
15702 (get_segment_min_max): New function.
15703 (create_intersect_range_checks): Use it.
15704 (dr_step_indicator): New function.
15705 (dr_direction_indicator): Likewise.
15706 (dr_zero_step_indicator): Likewise.
15707 (dr_known_forward_stride_p): Likewise.
15708 * tree-loop-distribution.c (data_ref_segment_size): Return
15709 DR_STEP * (niters - 1).
15710 (compute_alias_check_pairs): Update call to the dr_with_seg_len
15712 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
15713 (vect_preserves_scalar_order_p): New function, split out from...
15714 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
15715 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
15716 (vect_vfa_access_size): New function.
15717 (vect_vfa_align): Likewise.
15718 (vect_compile_time_alias): Take access_size_a and access_b arguments.
15719 (dump_lower_bound): New function.
15720 (vect_check_lower_bound): Likewise.
15721 (vect_small_gap_p): Likewise.
15722 (vectorizable_with_step_bound_p): Likewise.
15723 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
15724 depencies if the vectorization factor is 1. Convert the checks
15725 for nonzero steps into checks on the bounds of DR_STEP. Try using
15726 a bunds check for variable steps if the minimum required step is
15727 relatively small. Update calls to the dr_with_seg_len
15728 constructor and to vect_compile_time_alias.
15729 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
15731 (vect_loop_versioning): Call it.
15732 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
15734 (vect_estimate_min_profitable_iters): Account for any bounds checks.
15736 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15737 Alan Hayward <alan.hayward@arm.com>
15738 David Sherwood <david.sherwood@arm.com>
15740 * doc/sourcebuild.texi (vect_scatter_store): Document.
15741 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
15743 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
15745 * genopinit.c (main): Add supports_vec_scatter_store and
15746 supports_vec_scatter_store_cached to target_optabs.
15747 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
15748 IFN_MASK_SCATTER_STORE.
15749 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
15751 * internal-fn.h (internal_store_fn_p): Declare.
15752 (internal_fn_stored_value_index): Likewise.
15753 * internal-fn.c (scatter_store_direct): New macro.
15754 (expand_scatter_store_optab_fn): New function.
15755 (direct_scatter_store_optab_supported_p): New macro.
15756 (internal_store_fn_p): New function.
15757 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
15758 IFN_MASK_SCATTER_STORE.
15759 (internal_fn_mask_index): Likewise.
15760 (internal_fn_stored_value_index): New function.
15761 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
15762 for scatter stores.
15763 * optabs-query.h (supports_vec_scatter_store_p): Declare.
15764 * optabs-query.c (supports_vec_scatter_store_p): New function.
15765 * tree-vectorizer.h (vect_get_store_rhs): Declare.
15766 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
15767 true for scatter stores.
15768 (vect_gather_scatter_fn_p): Handle scatter stores too.
15769 (vect_check_gather_scatter): Consider using scatter stores if
15770 supports_vec_scatter_store_p.
15771 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
15772 scatter stores too.
15773 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
15774 internal_fn_stored_value_index.
15775 (check_load_store_masking): Handle scatter stores too.
15776 (vect_get_store_rhs): Make public.
15777 (vectorizable_call): Use internal_store_fn_p.
15778 (vectorizable_store): Handle scatter store internal functions.
15779 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
15780 when deciding whether the end of the group has been reached.
15781 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
15782 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
15783 (mask_scatter_store<mode>): New insns.
15785 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15786 Alan Hayward <alan.hayward@arm.com>
15787 David Sherwood <david.sherwood@arm.com>
15789 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
15790 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
15791 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
15793 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
15794 Use vect_truncate_gather_scatter_offset if we can't treat the
15795 operation as a normal gather load or scatter store.
15796 (get_group_load_store_type): Take the gather_scatter_info
15797 as argument. Try using a gather load or scatter store for
15798 single-element groups.
15799 (get_load_store_type): Update calls to get_group_load_store_type
15800 and vect_use_strided_gather_scatters_p.
15802 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15803 Alan Hayward <alan.hayward@arm.com>
15804 David Sherwood <david.sherwood@arm.com>
15806 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
15807 optional tree argument.
15808 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
15810 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
15811 but continue to use the current value as a fallback.
15812 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
15813 to compare the updates.
15814 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
15815 (get_load_store_type): Use it when handling a strided access.
15816 (vect_get_strided_load_store_ops): New function.
15817 (vect_get_data_ptr_increment): Likewise.
15818 (vectorizable_load): Handle strided gather loads. Always pass
15819 a step to vect_create_data_ref_ptr and bump_vector_ptr.
15821 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15822 Alan Hayward <alan.hayward@arm.com>
15823 David Sherwood <david.sherwood@arm.com>
15825 * doc/md.texi (gather_load@var{m}): Document.
15826 (mask_gather_load@var{m}): Likewise.
15827 * genopinit.c (main): Add supports_vec_gather_load and
15828 supports_vec_gather_load_cached to target_optabs.
15829 * optabs-tree.c (init_tree_optimization_optabs): Use
15830 ggc_cleared_alloc to allocate target_optabs.
15831 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
15832 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
15834 * internal-fn.h (internal_load_fn_p): Declare.
15835 (internal_gather_scatter_fn_p): Likewise.
15836 (internal_fn_mask_index): Likewise.
15837 (internal_gather_scatter_fn_supported_p): Likewise.
15838 * internal-fn.c (gather_load_direct): New macro.
15839 (expand_gather_load_optab_fn): New function.
15840 (direct_gather_load_optab_supported_p): New macro.
15841 (direct_internal_fn_optab): New function.
15842 (internal_load_fn_p): Likewise.
15843 (internal_gather_scatter_fn_p): Likewise.
15844 (internal_fn_mask_index): Likewise.
15845 (internal_gather_scatter_fn_supported_p): Likewise.
15846 * optabs-query.c (supports_at_least_one_mode_p): New function.
15847 (supports_vec_gather_load_p): Likewise.
15848 * optabs-query.h (supports_vec_gather_load_p): Declare.
15849 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
15850 and memory_type field.
15851 (NUM_PATTERNS): Bump to 15.
15852 * tree-vect-data-refs.c: Include internal-fn.h.
15853 (vect_gather_scatter_fn_p): New function.
15854 (vect_describe_gather_scatter_call): Likewise.
15855 (vect_check_gather_scatter): Try using internal functions for
15856 gather loads. Recognize existing calls to a gather load function.
15857 (vect_analyze_data_refs): Consider using gather loads if
15858 supports_vec_gather_load_p.
15859 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
15860 (vect_get_gather_scatter_offset_type): Likewise.
15861 (vect_convert_mask_for_vectype): Likewise.
15862 (vect_add_conversion_to_patterm): Likewise.
15863 (vect_try_gather_scatter_pattern): Likewise.
15864 (vect_recog_gather_scatter_pattern): New pattern recognizer.
15865 (vect_vect_recog_func_ptrs): Add it.
15866 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
15867 internal_fn_mask_index and internal_gather_scatter_fn_p.
15868 (check_load_store_masking): Take the gather_scatter_info as an
15869 argument and handle gather loads.
15870 (vect_get_gather_scatter_ops): New function.
15871 (vectorizable_call): Check internal_load_fn_p.
15872 (vectorizable_load): Likewise. Handle gather load internal
15874 (vectorizable_store): Update call to check_load_store_masking.
15875 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
15876 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
15877 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
15878 (aarch64_gather_scale_operand_d): New predicates.
15879 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
15880 (mask_gather_load<mode>): New insns.
15882 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15883 Alan Hayward <alan.hayward@arm.com>
15884 David Sherwood <david.sherwood@arm.com>
15886 * optabs.def (fold_left_plus_optab): New optab.
15887 * doc/md.texi (fold_left_plus_@var{m}): Document.
15888 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
15889 * internal-fn.c (fold_left_direct): Define.
15890 (expand_fold_left_optab_fn): Likewise.
15891 (direct_fold_left_optab_supported_p): Likewise.
15892 * fold-const-call.c (fold_const_fold_left): New function.
15893 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
15894 * tree-parloops.c (valid_reduction_p): New function.
15895 (gather_scalar_reductions): Use it.
15896 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
15897 (vect_finish_replace_stmt): Declare.
15898 * tree-vect-loop.c (fold_left_reduction_fn): New function.
15899 (needs_fold_left_reduction_p): New function, split out from...
15900 (vect_is_simple_reduction): ...here. Accept reductions that
15901 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
15902 (vect_force_simple_reduction): Also store the reduction type in
15903 the assignment's STMT_VINFO_REDUC_TYPE.
15904 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
15905 (merge_with_identity): New function.
15906 (vect_expand_fold_left): Likewise.
15907 (vectorize_fold_left_reduction): Likewise.
15908 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
15909 scalar phi in place for it. Check for target support and reject
15910 cases that would reassociate the operation. Defer the transform
15911 phase to vectorize_fold_left_reduction.
15912 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
15913 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
15914 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
15916 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15918 * tree-if-conv.c (predicate_mem_writes): Remove redundant
15919 call to ifc_temp_var.
15921 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15922 Alan Hayward <alan.hayward@arm.com>
15923 David Sherwood <david.sherwood@arm.com>
15925 * target.def (legitimize_address_displacement): Take the original
15926 offset as a poly_int.
15927 * targhooks.h (default_legitimize_address_displacement): Update
15929 * targhooks.c (default_legitimize_address_displacement): Likewise.
15930 * doc/tm.texi: Regenerate.
15931 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
15932 as an argument, moving assert of ad->disp == ad->disp_term to...
15933 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
15934 Try calling targetm.legitimize_address_displacement before expanding
15935 the address rather than afterwards, and adjust for the new interface.
15936 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
15937 Match the new hook interface. Handle SVE addresses.
15938 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
15939 new hook interface.
15941 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15943 * Makefile.in (OBJS): Add early-remat.o.
15944 * target.def (select_early_remat_modes): New hook.
15945 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
15946 * doc/tm.texi: Regenerate.
15947 * targhooks.h (default_select_early_remat_modes): Declare.
15948 * targhooks.c (default_select_early_remat_modes): New function.
15949 * timevar.def (TV_EARLY_REMAT): New timevar.
15950 * passes.def (pass_early_remat): New pass.
15951 * tree-pass.h (make_pass_early_remat): Declare.
15952 * early-remat.c: New file.
15953 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
15955 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
15957 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15958 Alan Hayward <alan.hayward@arm.com>
15959 David Sherwood <david.sherwood@arm.com>
15961 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
15962 vfm1 with a bound_epilog parameter.
15963 (vect_do_peeling): Update calls accordingly, and move the prologue
15964 call earlier in the function. Treat the base bound_epilog as 0 for
15965 fully-masked loops and retain vf - 1 for other loops. Add 1 to
15966 this base when peeling for gaps.
15967 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
15968 with fully-masked loops.
15969 (vect_estimate_min_profitable_iters): Handle the single peeled
15970 iteration in that case.
15972 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15973 Alan Hayward <alan.hayward@arm.com>
15974 David Sherwood <david.sherwood@arm.com>
15976 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
15977 single-element interleaving even if the size is not a power of 2.
15978 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
15979 accesses for single-element interleaving if the group size is
15982 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
15983 Alan Hayward <alan.hayward@arm.com>
15984 David Sherwood <david.sherwood@arm.com>
15986 * doc/md.texi (fold_extract_last_@var{m}): Document.
15987 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
15988 * optabs.def (fold_extract_last_optab): New optab.
15989 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
15990 * internal-fn.c (fold_extract_direct): New macro.
15991 (expand_fold_extract_optab_fn): Likewise.
15992 (direct_fold_extract_optab_supported_p): Likewise.
15993 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
15994 * tree-vect-loop.c (vect_model_reduction_cost): Handle
15995 EXTRACT_LAST_REDUCTION.
15996 (get_initial_def_for_reduction): Do not create an initial vector
15997 for EXTRACT_LAST_REDUCTION reductions.
15998 (vectorizable_reduction): Leave the scalar phi in place for
15999 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
16000 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
16001 epilogue code for EXTRACT_LAST_REDUCTION and defer the
16002 transform phase to vectorizable_condition.
16003 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
16005 (vect_finish_stmt_generation): ...here.
16006 (vect_finish_replace_stmt): New function.
16007 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
16008 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
16010 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
16012 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16013 Alan Hayward <alan.hayward@arm.com>
16014 David Sherwood <david.sherwood@arm.com>
16016 * doc/md.texi (extract_last_@var{m}): Document.
16017 * optabs.def (extract_last_optab): New optab.
16018 * internal-fn.def (EXTRACT_LAST): New internal function.
16019 * internal-fn.c (cond_unary_direct): New macro.
16020 (expand_cond_unary_optab_fn): Likewise.
16021 (direct_cond_unary_optab_supported_p): Likewise.
16022 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
16023 loops using EXTRACT_LAST.
16024 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
16025 (extract_last_<mode>): ...this optab.
16026 (vec_extract<mode><Vel>): Update accordingly.
16028 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16029 Alan Hayward <alan.hayward@arm.com>
16030 David Sherwood <david.sherwood@arm.com>
16032 * target.def (empty_mask_is_expensive): New hook.
16033 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
16034 * doc/tm.texi: Regenerate.
16035 * targhooks.h (default_empty_mask_is_expensive): Declare.
16036 * targhooks.c (default_empty_mask_is_expensive): New function.
16037 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
16038 if the target says that empty masks are expensive.
16039 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
16041 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
16043 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16044 Alan Hayward <alan.hayward@arm.com>
16045 David Sherwood <david.sherwood@arm.com>
16047 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
16048 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
16049 (vect_use_loop_mask_for_alignment_p): New function.
16050 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
16051 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
16052 niters_skip argument. Make sure that the first niters_skip elements
16053 of the first iteration are inactive.
16054 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
16055 Update call to vect_set_loop_masks_directly.
16056 (get_misalign_in_elems): New function, split out from...
16057 (vect_gen_prolog_loop_niters): ...here.
16058 (vect_update_init_of_dr): Take a code argument that specifies whether
16059 the adjustment should be added or subtracted.
16060 (vect_update_init_of_drs): Likewise.
16061 (vect_prepare_for_masked_peels): New function.
16062 (vect_do_peeling): Skip prologue peeling if we're using a mask
16063 instead. Update call to vect_update_inits_of_drs.
16064 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
16066 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
16067 alignment. Do not include the number of peeled iterations in
16068 the minimum threshold in that case.
16069 (vectorizable_induction): Adjust the start value down by
16070 LOOP_VINFO_MASK_SKIP_NITERS iterations.
16071 (vect_transform_loop): Call vect_prepare_for_masked_peels.
16072 Take the number of skipped iterations into account when calculating
16074 * tree-vect-stmts.c (vect_gen_while_not): New function.
16076 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16077 Alan Hayward <alan.hayward@arm.com>
16078 David Sherwood <david.sherwood@arm.com>
16080 * doc/sourcebuild.texi (vect_fully_masked): Document.
16081 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
16082 default value to 0.
16083 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
16085 (vect_analyze_loop_2): ...here. Don't check the vectorization
16086 factor against the number of loop iterations if the loop is
16089 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16090 Alan Hayward <alan.hayward@arm.com>
16091 David Sherwood <david.sherwood@arm.com>
16093 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
16094 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
16095 (dump_groups): Update accordingly.
16096 (iv_use::mem_type): New member variable.
16097 (address_p): New function.
16098 (record_use): Add a mem_type argument and initialize the new
16100 (record_group_use): Add a mem_type argument. Use address_p.
16101 Remove obsolete null checks of base_object. Update call to record_use.
16102 (find_interesting_uses_op): Update call to record_group_use.
16103 (find_interesting_uses_cond): Likewise.
16104 (find_interesting_uses_address): Likewise.
16105 (get_mem_type_for_internal_fn): New function.
16106 (find_address_like_use): Likewise.
16107 (find_interesting_uses_stmt): Try find_address_like_use before
16108 calling find_interesting_uses_op.
16109 (addr_offset_valid_p): Use the iv mem_type field as the type
16110 of the addressed memory.
16111 (add_autoinc_candidates): Likewise.
16112 (get_address_cost): Likewise.
16113 (split_small_address_groups_p): Use address_p.
16114 (split_address_groups): Likewise.
16115 (add_iv_candidate_for_use): Likewise.
16116 (autoinc_possible_for_pair): Likewise.
16117 (rewrite_groups): Likewise.
16118 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
16119 (determine_group_iv_cost): Update after split of USE_ADDRESS.
16120 (get_alias_ptr_type_for_ptr_address): New function.
16121 (rewrite_use_address): Rewrite address uses in calls that were
16122 identified by find_address_like_use.
16124 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16125 Alan Hayward <alan.hayward@arm.com>
16126 David Sherwood <david.sherwood@arm.com>
16128 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
16130 * gimple-expr.h (is_gimple_addressable: Likewise.
16131 * gimple-expr.c (is_gimple_address): Likewise.
16132 * internal-fn.c (expand_call_mem_ref): New function.
16133 (expand_mask_load_optab_fn): Use it.
16134 (expand_mask_store_optab_fn): Likewise.
16136 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16137 Alan Hayward <alan.hayward@arm.com>
16138 David Sherwood <david.sherwood@arm.com>
16140 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
16141 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
16142 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
16143 (cond_umax@var{mode}): Document.
16144 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
16145 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
16146 (cond_umin_optab, cond_umax_optab): New optabs.
16147 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
16148 (COND_IOR, COND_XOR): New internal functions.
16149 * internal-fn.h (get_conditional_internal_fn): Declare.
16150 * internal-fn.c (cond_binary_direct): New macro.
16151 (expand_cond_binary_optab_fn): Likewise.
16152 (direct_cond_binary_optab_supported_p): Likewise.
16153 (get_conditional_internal_fn): New function.
16154 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
16155 Cope with reduction statements that are vectorized as calls rather
16157 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
16158 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
16159 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
16160 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
16161 (UNSPEC_COND_EOR): New unspecs.
16162 (optab): Add mappings for them.
16163 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
16164 (sve_int_op, sve_fp_op): New int attributes.
16166 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16167 Alan Hayward <alan.hayward@arm.com>
16168 David Sherwood <david.sherwood@arm.com>
16170 * optabs.def (while_ult_optab): New optab.
16171 * doc/md.texi (while_ult@var{m}@var{n}): Document.
16172 * internal-fn.def (WHILE_ULT): New internal function.
16173 * internal-fn.h (direct_internal_fn_supported_p): New override
16174 that takes two types as argument.
16175 * internal-fn.c (while_direct): New macro.
16176 (expand_while_optab_fn): New function.
16177 (convert_optab_supported_p): Likewise.
16178 (direct_while_optab_supported_p): New macro.
16179 * wide-int.h (wi::udiv_ceil): New function.
16180 * tree-vectorizer.h (rgroup_masks): New structure.
16181 (vec_loop_masks): New typedef.
16182 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
16183 and fully_masked_p.
16184 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
16185 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
16186 (vect_max_vf): New function.
16187 (slpeel_make_loop_iterate_ntimes): Delete.
16188 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
16189 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
16190 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
16191 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
16192 internal-fn.h, stor-layout.h and optabs-query.h.
16193 (vect_set_loop_mask): New function.
16194 (add_preheader_seq): Likewise.
16195 (add_header_seq): Likewise.
16196 (interleave_supported_p): Likewise.
16197 (vect_maybe_permute_loop_masks): Likewise.
16198 (vect_set_loop_masks_directly): Likewise.
16199 (vect_set_loop_condition_masked): Likewise.
16200 (vect_set_loop_condition_unmasked): New function, split out from
16201 slpeel_make_loop_iterate_ntimes.
16202 (slpeel_make_loop_iterate_ntimes): Rename to..
16203 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
16204 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
16205 (vect_do_peeling): Update call accordingly.
16206 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
16208 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
16209 mask_compare_type, can_fully_mask_p and fully_masked_p.
16210 (release_vec_loop_masks): New function.
16211 (_loop_vec_info): Use it to free the loop masks.
16212 (can_produce_all_loop_masks_p): New function.
16213 (vect_get_max_nscalars_per_iter): Likewise.
16214 (vect_verify_full_masking): Likewise.
16215 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
16216 retries, and free the mask rgroups before retrying. Check loop-wide
16217 reasons for disallowing fully-masked loops. Make the final decision
16218 about whether use a fully-masked loop or not.
16219 (vect_estimate_min_profitable_iters): Do not assume that peeling
16220 for the number of iterations will be needed for fully-masked loops.
16221 (vectorizable_reduction): Disable fully-masked loops.
16222 (vectorizable_live_operation): Likewise.
16223 (vect_halve_mask_nunits): New function.
16224 (vect_double_mask_nunits): Likewise.
16225 (vect_record_loop_mask): Likewise.
16226 (vect_get_loop_mask): Likewise.
16227 (vect_transform_loop): Handle the case in which the final loop
16228 iteration might handle a partial vector. Call vect_set_loop_condition
16229 instead of slpeel_make_loop_iterate_ntimes.
16230 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
16231 (check_load_store_masking): New function.
16232 (prepare_load_store_mask): Likewise.
16233 (vectorizable_store): Handle fully-masked loops.
16234 (vectorizable_load): Likewise.
16235 (supportable_widening_operation): Use vect_halve_mask_nunits for
16237 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
16238 (vect_gen_while): New function.
16239 * config/aarch64/aarch64.md (umax<mode>3): New expander.
16240 (aarch64_uqdec<mode>): New insn.
16242 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16243 Alan Hayward <alan.hayward@arm.com>
16244 David Sherwood <david.sherwood@arm.com>
16246 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
16247 (reduc_xor_scal_optab): New optabs.
16248 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
16249 (reduc_xor_scal_@var{m}): Document.
16250 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
16251 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
16252 internal functions.
16253 * fold-const-call.c (fold_const_call): Handle them.
16254 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
16255 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
16256 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
16257 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
16258 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
16259 (UNSPEC_XORV): New unspecs.
16260 (optab): Add entries for them.
16261 (BITWISEV): New int iterator.
16262 (bit_reduc_op): New int attributes.
16264 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16265 Alan Hayward <alan.hayward@arm.com>
16266 David Sherwood <david.sherwood@arm.com>
16268 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
16269 * internal-fn.def (VEC_SHL_INSERT): New internal function.
16270 * optabs.def (vec_shl_insert_optab): New optab.
16271 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
16272 (duplicate_and_interleave): Likewise.
16273 * tree-vect-loop.c: Include internal-fn.h.
16274 (neutral_op_for_slp_reduction): New function, split out from
16275 get_initial_defs_for_reduction.
16276 (get_initial_def_for_reduction): Handle option 2 for variable-length
16277 vectors by loading the neutral value into a vector and then shifting
16278 the initial value into element 0.
16279 (get_initial_defs_for_reduction): Replace the code argument with
16280 the neutral value calculated by neutral_op_for_slp_reduction.
16281 Use gimple_build_vector for constant-length vectors.
16282 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
16283 but the first group_size elements have a neutral value.
16284 Use duplicate_and_interleave otherwise.
16285 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
16286 Update call to get_initial_defs_for_reduction. Handle SLP
16287 reductions for variable-length vectors by creating one vector
16288 result for each scalar result, with the elements associated
16289 with other scalar results stubbed out with the neutral value.
16290 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
16291 Require IFN_VEC_SHL_INSERT for double reductions on
16292 variable-length vectors, or SLP reductions that have
16293 a neutral value. Require can_duplicate_and_interleave_p
16294 support for variable-length unchained SLP reductions if there
16295 is no neutral value, such as for MIN/MAX reductions. Also require
16296 the number of vector elements to be a multiple of the number of
16297 SLP statements when doing variable-length unchained SLP reductions.
16298 Update call to vect_create_epilog_for_reduction.
16299 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
16300 and remove initial values.
16301 (duplicate_and_interleave): Make public.
16302 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
16303 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
16305 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16306 Alan Hayward <alan.hayward@arm.com>
16307 David Sherwood <david.sherwood@arm.com>
16309 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
16310 (can_duplicate_and_interleave_p): New function.
16311 (vect_get_and_check_slp_defs): Take the vector of statements
16312 rather than just the current one. Remove excess parentheses.
16313 Restriction rejectinon of vect_constant_def and vect_external_def
16314 for variable-length vectors to boolean types, or types for which
16315 can_duplicate_and_interleave_p is false.
16316 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
16317 (duplicate_and_interleave): New function.
16318 (vect_get_constant_vectors): Use gimple_build_vector for
16319 constant-length vectors and suitable variable-length constant
16320 vectors. Use duplicate_and_interleave for other variable-length
16321 vectors. Don't defer the update when inserting new statements.
16323 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16324 Alan Hayward <alan.hayward@arm.com>
16325 David Sherwood <david.sherwood@arm.com>
16327 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
16328 min_profitable_iters doesn't go negative.
16330 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16331 Alan Hayward <alan.hayward@arm.com>
16332 David Sherwood <david.sherwood@arm.com>
16334 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
16335 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
16336 * optabs.def (vec_mask_load_lanes_optab): New optab.
16337 (vec_mask_store_lanes_optab): Likewise.
16338 * internal-fn.def (MASK_LOAD_LANES): New internal function.
16339 (MASK_STORE_LANES): Likewise.
16340 * internal-fn.c (mask_load_lanes_direct): New macro.
16341 (mask_store_lanes_direct): Likewise.
16342 (expand_mask_load_optab_fn): Handle masked operations.
16343 (expand_mask_load_lanes_optab_fn): New macro.
16344 (expand_mask_store_optab_fn): Handle masked operations.
16345 (expand_mask_store_lanes_optab_fn): New macro.
16346 (direct_mask_load_lanes_optab_supported_p): Likewise.
16347 (direct_mask_store_lanes_optab_supported_p): Likewise.
16348 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
16350 (vect_load_lanes_supported): Likewise.
16351 * tree-vect-data-refs.c (strip_conversion): New function.
16352 (can_group_stmts_p): Likewise.
16353 (vect_analyze_data_ref_accesses): Use it instead of checking
16354 for a pair of assignments.
16355 (vect_store_lanes_supported): Take a masked_p parameter.
16356 (vect_load_lanes_supported): Likewise.
16357 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
16358 vect_store_lanes_supported and vect_load_lanes_supported.
16359 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
16360 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
16361 parameter. Don't allow gaps for masked accesses.
16362 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
16363 and vect_load_lanes_supported.
16364 (get_load_store_type): Take a masked_p parameter and update
16365 call to get_group_load_store_type.
16366 (vectorizable_store): Update call to get_load_store_type.
16367 Handle IFN_MASK_STORE_LANES.
16368 (vectorizable_load): Update call to get_load_store_type.
16369 Handle IFN_MASK_LOAD_LANES.
16371 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16372 Alan Hayward <alan.hayward@arm.com>
16373 David Sherwood <david.sherwood@arm.com>
16375 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
16377 * config/aarch64/aarch64-protos.h
16378 (aarch64_sve_struct_memory_operand_p): Declare.
16379 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
16380 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
16381 (VPRED, vpred): Handle SVE structure modes.
16382 * config/aarch64/constraints.md (Utx): New constraint.
16383 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
16384 (aarch64_sve_struct_nonimmediate_operand): New predicates.
16385 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
16386 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
16387 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
16388 structure modes. Split into pieces after RA.
16389 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
16390 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
16392 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
16393 SVE structure modes.
16394 (aarch64_classify_address): Likewise.
16395 (sizetochar): Move earlier in file.
16396 (aarch64_print_operand): Handle SVE register lists.
16397 (aarch64_array_mode): New function.
16398 (aarch64_sve_struct_memory_operand_p): Likewise.
16399 (TARGET_ARRAY_MODE): Redefine.
16401 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16402 Alan Hayward <alan.hayward@arm.com>
16403 David Sherwood <david.sherwood@arm.com>
16405 * target.def (array_mode): New target hook.
16406 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
16407 * doc/tm.texi: Regenerate.
16408 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
16409 * hooks.c (hook_optmode_mode_uhwi_none): New function.
16410 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
16411 targetm.array_mode.
16412 * stor-layout.c (mode_for_array): Likewise. Support polynomial
16415 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16416 Alan Hayward <alan.hayward@arm.com>
16417 David Sherwood <david.sherwood@arm.com>
16419 * fold-const.c (fold_binary_loc): Check the argument types
16420 rather than the result type when testing for a vector operation.
16422 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16424 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
16425 * doc/tm.texi: Regenerate.
16427 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16428 Alan Hayward <alan.hayward@arm.com>
16429 David Sherwood <david.sherwood@arm.com>
16431 * doc/invoke.texi (-msve-vector-bits=): Document new option.
16432 (sve): Document new AArch64 extension.
16433 * doc/md.texi (w): Extend the description of the AArch64
16434 constraint to include SVE vectors.
16435 (Upl, Upa): Document new AArch64 predicate constraints.
16436 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
16438 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
16439 (msve-vector-bits=): New option.
16440 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
16441 SVE when these are disabled.
16442 (sve): New extension.
16443 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
16444 modes. Adjust their number of units based on aarch64_sve_vg.
16445 (MAX_BITSIZE_MODE_ANY_MODE): Define.
16446 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
16447 aarch64_addr_query_type.
16448 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
16449 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
16450 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
16451 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
16452 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
16453 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
16454 (aarch64_simd_imm_zero_p): Delete.
16455 (aarch64_check_zero_based_sve_index_immediate): Declare.
16456 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
16457 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
16458 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
16459 (aarch64_sve_float_mul_immediate_p): Likewise.
16460 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
16461 rather than an rtx.
16462 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
16463 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
16464 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
16465 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
16466 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
16467 (aarch64_regmode_natural_size): Likewise.
16468 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
16469 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
16471 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
16472 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
16473 for VG and the SVE predicate registers.
16474 (V_ALIASES): Add a "z"-prefixed alias.
16475 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
16476 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
16477 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
16478 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
16479 (REG_CLASS_NAMES): Add entries for them.
16480 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
16481 and the predicate registers.
16482 (aarch64_sve_vg): Declare.
16483 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
16484 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
16485 (REGMODE_NATURAL_SIZE): Define.
16486 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
16488 * config/aarch64/aarch64.c: Include cfgrtl.h.
16489 (simd_immediate_info): Add a constructor for series vectors,
16490 and an associated step field.
16491 (aarch64_sve_vg): New variable.
16492 (aarch64_dbx_register_number): Handle VG and the predicate registers.
16493 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
16494 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
16495 (VEC_ANY_DATA, VEC_STRUCT): New constants.
16496 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
16497 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
16498 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
16499 (aarch64_get_mask_mode): New functions.
16500 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
16501 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
16502 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
16503 predicate modes and predicate registers. Explicitly restrict
16504 GPRs to modes of 16 bytes or smaller. Only allow FP registers
16505 to store a vector mode if it is recognized by
16506 aarch64_classify_vector_mode.
16507 (aarch64_regmode_natural_size): New function.
16508 (aarch64_hard_regno_caller_save_mode): Return the original mode
16510 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
16511 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
16512 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
16513 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
16515 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
16516 does not overlap dest if the function is frame-related. Handle
16518 (aarch64_split_add_offset): New function.
16519 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
16520 them aarch64_add_offset.
16521 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
16522 and update call to aarch64_sub_sp.
16523 (aarch64_add_cfa_expression): New function.
16524 (aarch64_expand_prologue): Pass extra temporary registers to the
16525 functions above. Handle the case in which we need to emit new
16526 DW_CFA_expressions for registers that were originally saved
16527 relative to the stack pointer, but now have to be expressed
16528 relative to the frame pointer.
16529 (aarch64_output_mi_thunk): Pass extra temporary registers to the
16531 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
16532 IP0 and IP1 values for SVE frames.
16533 (aarch64_expand_vec_series): New function.
16534 (aarch64_expand_sve_widened_duplicate): Likewise.
16535 (aarch64_expand_sve_const_vector): Likewise.
16536 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
16537 Handle SVE constants. Use emit_move_insn to move a force_const_mem
16538 into the register, rather than emitting a SET directly.
16539 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
16540 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
16541 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
16542 (offset_9bit_signed_scaled_p): New functions.
16543 (aarch64_replicate_bitmask_imm): New function.
16544 (aarch64_bitmask_imm): Use it.
16545 (aarch64_cannot_force_const_mem): Reject expressions involving
16546 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
16547 (aarch64_classify_index): Handle SVE indices, by requiring
16548 a plain register index with a scale that matches the element size.
16549 (aarch64_classify_address): Handle SVE addresses. Assert that
16550 the mode of the address is VOIDmode or an integer mode.
16551 Update call to aarch64_classify_symbol.
16552 (aarch64_classify_symbolic_expression): Update call to
16553 aarch64_classify_symbol.
16554 (aarch64_const_vec_all_in_range_p): New function.
16555 (aarch64_print_vector_float_operand): Likewise.
16556 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
16557 "vN" for FP registers with SVE modes. Handle (const ...) vectors
16558 and the FP immediates 1.0 and 0.5.
16559 (aarch64_print_address_internal): Handle SVE addresses.
16560 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
16561 (aarch64_regno_regclass): Handle predicate registers.
16562 (aarch64_secondary_reload): Handle big-endian reloads of SVE
16564 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
16565 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
16566 (aarch64_convert_sve_vector_bits): New function.
16567 (aarch64_override_options): Use it to handle -msve-vector-bits=.
16568 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
16569 rather than an rtx.
16570 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
16571 Handle SVE vector and predicate modes. Accept VL-based constants
16572 that need only one temporary register, and VL offsets that require
16573 no temporary registers.
16574 (aarch64_conditional_register_usage): Mark the predicate registers
16575 as fixed if SVE isn't available.
16576 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
16577 Return true for SVE vector and predicate modes.
16578 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
16579 rather than an unsigned int. Handle SVE modes.
16580 (aarch64_preferred_simd_mode): Update call accordingly. Handle
16582 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
16584 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
16585 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
16586 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
16587 (aarch64_sve_float_mul_immediate_p): New functions.
16588 (aarch64_sve_valid_immediate): New function.
16589 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
16590 Explicitly reject structure modes. Check for INDEX constants.
16591 Handle PTRUE and PFALSE constants.
16592 (aarch64_check_zero_based_sve_index_immediate): New function.
16593 (aarch64_simd_imm_zero_p): Delete.
16594 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
16595 vector modes. Accept constants in the range of CNT[BHWD].
16596 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
16597 ask for an Advanced SIMD mode.
16598 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
16599 (aarch64_simd_vector_alignment): Handle SVE predicates.
16600 (aarch64_vectorize_preferred_vector_alignment): New function.
16601 (aarch64_simd_vector_alignment_reachable): Use it instead of
16603 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
16604 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
16606 (MAX_VECT_LEN): Delete.
16607 (expand_vec_perm_d): Add a vec_flags field.
16608 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
16609 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
16610 (aarch64_evpc_ext): Don't apply a big-endian lane correction
16612 (aarch64_evpc_rev): Rename to...
16613 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
16614 (aarch64_evpc_rev_global): New function.
16615 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
16616 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
16618 (aarch64_evpc_sve_tbl): New function.
16619 (aarch64_expand_vec_perm_const_1): Update after rename of
16620 aarch64_evpc_rev. Handle SVE permutes too, trying
16621 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
16622 than aarch64_evpc_tbl.
16623 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
16624 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
16625 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
16626 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
16627 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
16628 (aarch64_expand_sve_vcond): New functions.
16629 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
16630 of aarch64_vector_mode_p.
16631 (aarch64_dwarf_poly_indeterminate_value): New function.
16632 (aarch64_compute_pressure_classes): Likewise.
16633 (aarch64_can_change_mode_class): Likewise.
16634 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
16635 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
16636 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
16637 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
16638 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
16639 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
16640 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
16641 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
16643 (Dn, Dl, Dr): Accept const as well as const_vector.
16644 (Dz): Likewise. Compare against CONST0_RTX.
16645 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
16646 of "vector" where appropriate.
16647 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
16648 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
16649 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
16650 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
16651 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
16652 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
16653 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
16654 (v_int_equiv): Extend to SVE modes.
16655 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
16657 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
16658 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
16659 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
16660 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
16661 (SVE_COND_FP_CMP): New int iterators.
16662 (perm_hilo): Handle the new unpack unspecs.
16663 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
16665 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
16666 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
16667 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
16668 (aarch64_equality_operator, aarch64_constant_vector_operand)
16669 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
16670 (aarch64_sve_nonimmediate_operand): Likewise.
16671 (aarch64_sve_general_operand): Likewise.
16672 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
16673 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
16674 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
16675 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
16676 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
16677 (aarch64_sve_float_arith_immediate): Likewise.
16678 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
16679 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
16680 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
16681 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
16682 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
16683 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
16684 (aarch64_sve_float_arith_operand): Likewise.
16685 (aarch64_sve_float_arith_with_sub_operand): Likewise.
16686 (aarch64_sve_float_mul_operand): Likewise.
16687 (aarch64_sve_vec_perm_operand): Likewise.
16688 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
16689 (aarch64_mov_operand): Accept const_poly_int and const_vector.
16690 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
16691 as well as const_vector.
16692 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
16693 in file. Use CONST0_RTX and CONSTM1_RTX.
16694 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
16695 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
16696 Use aarch64_simd_imm_zero.
16697 * config/aarch64/aarch64-sve.md: New file.
16698 * config/aarch64/aarch64.md: Include it.
16699 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
16700 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
16701 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
16702 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
16703 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
16704 (sve): New attribute.
16705 (enabled): Disable instructions with the sve attribute unless
16707 (movqi, movhi): Pass CONST_POLY_INT operaneds through
16708 aarch64_expand_mov_immediate.
16709 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
16710 CNT[BHSD] immediates.
16711 (movti): Split CONST_POLY_INT moves into two halves.
16712 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
16713 Split additions that need a temporary here if the destination
16714 is the stack pointer.
16715 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
16716 (*add<mode>3_poly_1): New instruction.
16717 (set_clobber_cc): New expander.
16719 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16721 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
16722 parameter and use it instead of GET_MODE_SIZE (innermode). Use
16723 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
16724 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
16725 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
16726 Change innermode from fixed_mode_size to machine_mode.
16727 (simplify_subreg): Update call accordingly. Handle a constant-sized
16728 subreg of a variable-length CONST_VECTOR.
16730 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
16731 Alan Hayward <alan.hayward@arm.com>
16732 David Sherwood <david.sherwood@arm.com>
16734 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
16735 (add_offset_to_base): New function, split out from...
16736 (create_mem_ref): ...here. When handling a scale other than 1,
16737 check first whether the address is valid without the offset.
16738 Add it into the base if so, leaving the index and scale as-is.
16740 2018-01-12 Jakub Jelinek <jakub@redhat.com>
16743 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
16744 fold_for_warn before checking if arg2 is INTEGER_CST.
16746 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
16748 * config/rs6000/predicates.md (load_multiple_operation): Delete.
16749 (store_multiple_operation): Delete.
16750 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
16751 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
16752 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
16753 guarded by TARGET_STRING.
16754 (rs6000_output_load_multiple): Delete.
16755 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16756 OPTION_MASK_STRING / TARGET_STRING handling.
16757 (print_operand) <'N', 'O'>: Add comment that these are unused now.
16758 (const rs6000_opt_masks) <"string">: Change mask to 0.
16759 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
16760 (MASK_STRING): Delete.
16761 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
16763 (load_multiple): Delete.
16770 (store_multiple): Delete.
16777 (movmemsi_8reg): Delete.
16778 (corresponding unnamed define_insn): Delete.
16779 (movmemsi_6reg): Delete.
16780 (corresponding unnamed define_insn): Delete.
16781 (movmemsi_4reg): Delete.
16782 (corresponding unnamed define_insn): Delete.
16783 (movmemsi_2reg): Delete.
16784 (corresponding unnamed define_insn): Delete.
16785 (movmemsi_1reg): Delete.
16786 (corresponding unnamed define_insn): Delete.
16787 * config/rs6000/rs6000.opt (mno-string): New.
16788 (mstring): Replace by deprecation warning stub.
16789 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
16791 2018-01-12 Jakub Jelinek <jakub@redhat.com>
16793 * regrename.c (regrename_do_replace): If replacing the same
16794 reg multiple times, try to reuse last created gen_raw_REG.
16797 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
16798 main to workaround a bug in GDB.
16800 2018-01-12 Tom de Vries <tom@codesourcery.com>
16803 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
16805 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
16807 PR rtl-optimization/80481
16808 * ira-color.c (get_cap_member): New function.
16809 (allocnos_conflict_by_live_ranges_p): Use it.
16810 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
16811 (setup_slot_coalesced_allocno_live_ranges): Ditto.
16813 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
16816 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
16817 (*saddl_se_1): Ditto.
16818 (*ssubsi_1): Ditto.
16819 (*ssubl_se_1): Ditto.
16821 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
16823 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
16824 rather than wi::to_widest for DR_INITs.
16825 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
16826 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
16827 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
16829 (vect_analyze_group_access_1): Note that here.
16831 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
16833 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
16834 polynomial type sizes.
16836 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
16838 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
16839 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
16840 (gimple_add_tmp_var): Likewise.
16842 2018-01-12 Martin Liska <mliska@suse.cz>
16844 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
16845 (gimple_alloc_sizes): Likewise.
16846 (dump_gimple_statistics): Use PRIu64 in printf format.
16847 * gimple.h: Change uint64_t to int.
16849 2018-01-12 Martin Liska <mliska@suse.cz>
16851 * tree-core.h: Use uint64_t instead of int.
16852 * tree.c (tree_node_counts): Likewise.
16853 (tree_node_sizes): Likewise.
16854 (dump_tree_statistics): Use PRIu64 in printf format.
16856 2018-01-12 Martin Liska <mliska@suse.cz>
16858 * Makefile.in: As qsort_chk is implemented in vec.c, add
16859 vec.o to linkage of gencfn-macros.
16860 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
16861 passing the info to record_node_allocation_statistics.
16862 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
16864 * ggc-common.c (struct ggc_usage): Add operator== and use
16865 it in operator< and compare function.
16866 * mem-stats.h (struct mem_usage): Likewise.
16867 * vec.c (struct vec_usage): Remove operator< and compare
16868 function. Can be simply inherited.
16870 2018-01-12 Martin Jambor <mjambor@suse.cz>
16873 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
16874 * tree-ssa-math-opts.c: Include domwalk.h.
16875 (convert_mult_to_fma_1): New function.
16876 (fma_transformation_info): New type.
16877 (fma_deferring_state): Likewise.
16878 (cancel_fma_deferring): New function.
16879 (result_of_phi): Likewise.
16880 (last_fma_candidate_feeds_initial_phi): Likewise.
16881 (convert_mult_to_fma): Added deferring logic, split actual
16882 transformation to convert_mult_to_fma_1.
16883 (math_opts_dom_walker): New type.
16884 (math_opts_dom_walker::after_dom_children): New method, body moved
16885 here from pass_optimize_widening_mul::execute, added deferring logic
16887 (pass_optimize_widening_mul::execute): Moved most of code to
16888 math_opts_dom_walker::after_dom_children.
16889 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
16890 * config/i386/i386.c (ix86_option_override_internal): Added
16891 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
16893 2018-01-12 Richard Biener <rguenther@suse.de>
16896 * dwarf2out.c (gen_variable_die): Do not reset old_die for
16897 inline instance vars.
16899 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
16902 * config/rx/rx.c (rx_is_restricted_memory_address):
16903 Handle SUBREG case.
16905 2018-01-12 Richard Biener <rguenther@suse.de>
16907 PR tree-optimization/80846
16908 * target.def (split_reduction): New target hook.
16909 * targhooks.c (default_split_reduction): New function.
16910 * targhooks.h (default_split_reduction): Declare.
16911 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
16912 target requests first reduce vectors by combining low and high
16914 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
16915 (get_vectype_for_scalar_type_and_size): Export.
16916 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
16917 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
16918 * doc/tm.texi: Regenerate.
16919 * config/i386/i386.c (ix86_split_reduction): Implement
16920 TARGET_VECTORIZE_SPLIT_REDUCTION.
16922 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
16925 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
16926 in PIC mode except for TARGET_VXWORKS_RTP.
16927 * config/sparc/sparc.c: Include cfgrtl.h.
16928 (TARGET_INIT_PIC_REG): Define.
16929 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
16930 (sparc_pic_register_p): New predicate.
16931 (sparc_legitimate_address_p): Use it.
16932 (sparc_legitimize_pic_address): Likewise.
16933 (sparc_delegitimize_address): Likewise.
16934 (sparc_mode_dependent_address_p): Likewise.
16935 (gen_load_pcrel_sym): Remove 4th parameter.
16936 (load_got_register): Adjust call to above. Remove obsolete stuff.
16937 (sparc_expand_prologue): Do not call load_got_register here.
16938 (sparc_flat_expand_prologue): Likewise.
16939 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
16940 (sparc_use_pseudo_pic_reg): New function.
16941 (sparc_init_pic_reg): Likewise.
16942 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
16943 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
16945 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
16947 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
16948 Add item for branch_cost.
16950 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
16952 PR rtl-optimization/83565
16953 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
16954 not extend the result to a larger mode for rotate operations.
16955 (num_sign_bit_copies1): Likewise.
16957 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16960 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
16962 Use values-Xc.o for -pedantic.
16963 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
16965 2018-01-12 Martin Liska <mliska@suse.cz>
16968 * ipa-devirt.c (final_warning_record::grow_type_warnings):
16970 (possible_polymorphic_call_targets): Use it.
16971 (ipa_devirt): Likewise.
16973 2018-01-12 Martin Liska <mliska@suse.cz>
16975 * profile-count.h (enum profile_quality): Use 0 as invalid
16976 enum value of profile_quality.
16978 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
16980 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
16981 -mext-string options.
16983 2018-01-12 Richard Biener <rguenther@suse.de>
16985 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
16986 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
16987 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
16989 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
16991 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
16993 * configure.ac (--with-long-double-format): Add support for the
16994 configuration option to change the default long double format on
16996 * config.gcc (powerpc*-linux*-*): Likewise.
16997 * configure: Regenerate.
16998 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
16999 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
17000 used without modification.
17002 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17004 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
17005 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
17006 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
17007 MISC_BUILTIN_SPEC_BARRIER.
17008 (rs6000_init_builtins): Likewise.
17009 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
17011 (speculation_barrier): New define_insn.
17012 * doc/extend.texi: Document __builtin_speculation_barrier.
17014 2018-01-11 Jakub Jelinek <jakub@redhat.com>
17017 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
17018 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
17019 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
17021 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
17022 integral modes instead of "ss" and "sd".
17023 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
17024 vectors with 32-bit and 64-bit elements.
17025 (vecdupssescalarmodesuffix): New mode attribute.
17026 (vec_dup<mode>): Use it.
17028 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
17031 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
17032 frame if argument is passed on stack.
17034 2018-01-11 Jakub Jelinek <jakub@redhat.com>
17037 * ree.c (combine_reaching_defs): Optimize also
17038 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
17039 reg2=any_extend(exp); reg1=reg2;, formatting fix.
17041 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
17043 PR middle-end/83189
17044 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
17046 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
17048 PR middle-end/83718
17049 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
17050 after they are computed.
17052 2018-01-11 Bin Cheng <bin.cheng@arm.com>
17054 PR tree-optimization/83695
17055 * gimple-loop-linterchange.cc
17056 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
17057 reset cached scev information after interchange.
17058 (pass_linterchange::execute): Remove call to scev_reset_htab.
17060 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17062 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
17063 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
17064 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
17065 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
17066 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
17067 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
17068 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
17069 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
17070 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
17071 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
17072 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
17073 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
17074 (V_lane_reg): Likewise.
17075 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
17077 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
17078 (vfmal_lane_low<mode>_intrinsic,
17079 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
17080 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
17081 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
17082 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
17083 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
17084 vfmsl_lane_high<mode>_intrinsic): New define_insns.
17086 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17088 * config/arm/arm-cpus.in (fp16fml): New feature.
17089 (ALL_SIMD): Add fp16fml.
17090 (armv8.2-a): Add fp16fml as an option.
17091 (armv8.3-a): Likewise.
17092 (armv8.4-a): Add fp16fml as part of fp16.
17093 * config/arm/arm.h (TARGET_FP16FML): Define.
17094 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
17096 * config/arm/arm-modes.def (V2HF): Define.
17097 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
17098 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
17099 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
17100 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
17101 vfmsl_low, vfmsl_high): New set of builtins.
17102 * config/arm/iterators.md (PLUSMINUS): New code iterator.
17103 (vfml_op): New code attribute.
17104 (VFMLHALVES): New int iterator.
17105 (VFML, VFMLSEL): New mode attributes.
17106 (V_reg): Define mapping for V2HF.
17107 (V_hi, V_lo): New mode attributes.
17108 (VF_constraint): Likewise.
17109 (vfml_half, vfml_half_selector): New int attributes.
17110 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
17112 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
17113 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
17115 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
17116 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
17117 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
17118 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
17120 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
17121 Document new effective target and option set.
17123 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17125 * config/arm/arm-cpus.in (armv8_4): New feature.
17126 (ARMv8_4a): New fgroup.
17127 (armv8.4-a): New arch.
17128 * config/arm/arm-tables.opt: Regenerate.
17129 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
17130 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
17131 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
17132 Add matching rules for -march=armv8.4-a and extensions.
17133 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
17135 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
17138 * config/rx/rx.md (BW): New mode attribute.
17139 (sync_lock_test_and_setsi): Add mode suffix to insn output.
17141 2018-01-11 Richard Biener <rguenther@suse.de>
17143 PR tree-optimization/83435
17144 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
17145 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
17146 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
17148 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
17149 Alan Hayward <alan.hayward@arm.com>
17150 David Sherwood <david.sherwood@arm.com>
17152 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
17154 (aarch64_classify_address): Initialize it. Track polynomial offsets.
17155 (aarch64_print_address_internal): Use it to check for a zero offset.
17157 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
17158 Alan Hayward <alan.hayward@arm.com>
17159 David Sherwood <david.sherwood@arm.com>
17161 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
17162 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
17163 Return a poly_int64 rather than a HOST_WIDE_INT.
17164 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
17165 rather than a HOST_WIDE_INT.
17166 * config/aarch64/aarch64.h (aarch64_frame): Protect with
17167 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
17168 hard_fp_offset, frame_size, initial_adjust, callee_offset and
17169 final_offset from HOST_WIDE_INT to poly_int64.
17170 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
17171 to_constant when getting the number of units in an Advanced SIMD
17173 (aarch64_builtin_vectorized_function): Check for a constant number
17175 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
17177 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
17178 attribute instead of GET_MODE_NUNITS.
17179 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
17180 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
17181 GET_MODE_SIZE for fixed-size registers.
17182 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
17183 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
17184 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
17185 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
17186 (aarch64_print_operand, aarch64_print_address_internal)
17187 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
17188 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
17189 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
17190 Handle polynomial GET_MODE_SIZE.
17191 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
17192 wider than SImode without modification.
17193 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
17194 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
17195 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
17196 passing and returning SVE modes.
17197 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
17198 rather than GEN_INT.
17199 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
17200 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
17201 (aarch64_allocate_and_probe_stack_space): Likewise.
17202 (aarch64_layout_frame): Cope with polynomial offsets.
17203 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
17204 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
17205 polynomial offsets.
17206 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
17207 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
17208 poly_int64 rather than a HOST_WIDE_INT.
17209 (aarch64_get_separate_components, aarch64_process_components)
17210 (aarch64_expand_prologue, aarch64_expand_epilogue)
17211 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
17212 (aarch64_anchor_offset): New function, split out from...
17213 (aarch64_legitimize_address): ...here.
17214 (aarch64_builtin_vectorization_cost): Handle polynomial
17215 TYPE_VECTOR_SUBPARTS.
17216 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
17218 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
17219 number of elements from the PARALLEL rather than the mode.
17220 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
17221 rather than GET_MODE_BITSIZE.
17222 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
17223 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
17224 (aarch64_expand_vec_perm_const_1): Handle polynomial
17225 d->perm.length () and d->perm elements.
17226 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
17227 Apply to_constant to d->perm elements.
17228 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
17229 polynomial CONST_VECTOR_NUNITS.
17230 (aarch64_move_pointer): Take amount as a poly_int64 rather
17232 (aarch64_progress_pointer): Avoid temporary variable.
17233 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
17234 the mode attribute instead of GET_MODE.
17236 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
17237 Alan Hayward <alan.hayward@arm.com>
17238 David Sherwood <david.sherwood@arm.com>
17240 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
17241 x exists before using it.
17242 (aarch64_add_constant_internal): Rename to...
17243 (aarch64_add_offset_1): ...this. Replace regnum with separate
17244 src and dest rtxes. Handle the case in which they're different,
17245 including when the offset is zero. Replace scratchreg with an rtx.
17246 Use 2 additions if there is no spare register into which we can
17247 move a 16-bit constant.
17248 (aarch64_add_constant): Delete.
17249 (aarch64_add_offset): Replace reg with separate src and dest
17250 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
17251 Use aarch64_add_offset_1.
17252 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
17253 an rtx rather than an int. Take the delta as a poly_int64
17254 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
17255 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
17256 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
17257 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
17258 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
17259 and aarch64_add_sp.
17260 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
17261 aarch64_add_constant.
17263 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
17265 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
17266 Use scalar_float_mode.
17268 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
17270 * config/aarch64/aarch64-simd.md
17271 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
17272 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
17273 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
17274 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
17275 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
17276 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
17277 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
17278 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
17279 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
17280 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
17282 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17285 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
17286 targ_options->x_arm_arch_string is non NULL.
17288 2018-01-11 Tamar Christina <tamar.christina@arm.com>
17290 * config/aarch64/aarch64.h
17291 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
17293 2018-01-11 Sudakshina Das <sudi.das@arm.com>
17296 * expmed.c (emit_store_flag_force): Swap if const op0
17297 and change VOIDmode to mode of op0.
17299 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
17301 PR rtl-optimization/83761
17302 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
17303 than bytes to mode_for_size.
17305 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
17307 PR middle-end/83189
17308 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
17309 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
17312 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
17314 PR middle-end/83575
17315 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
17316 when in layout mode.
17317 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
17318 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
17321 2018-01-10 Michael Collison <michael.collison@arm.com>
17323 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
17324 * config/aarch64/aarch64-option-extension.def: Add
17325 AARCH64_OPT_EXTENSION of 'fp16fml'.
17326 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
17327 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
17328 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
17329 * config/aarch64/constraints.md (Ui7): New constraint.
17330 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
17331 (VFMLA_SEL_W): Ditto.
17334 (VFMLA16_LOW): New int iterator.
17335 (VFMLA16_HIGH): Ditto.
17336 (UNSPEC_FMLAL): New unspec.
17337 (UNSPEC_FMLSL): Ditto.
17338 (UNSPEC_FMLAL2): Ditto.
17339 (UNSPEC_FMLSL2): Ditto.
17340 (f16mac): New code attribute.
17341 * config/aarch64/aarch64-simd-builtins.def
17342 (aarch64_fmlal_lowv2sf): Ditto.
17343 (aarch64_fmlsl_lowv2sf): Ditto.
17344 (aarch64_fmlalq_lowv4sf): Ditto.
17345 (aarch64_fmlslq_lowv4sf): Ditto.
17346 (aarch64_fmlal_highv2sf): Ditto.
17347 (aarch64_fmlsl_highv2sf): Ditto.
17348 (aarch64_fmlalq_highv4sf): Ditto.
17349 (aarch64_fmlslq_highv4sf): Ditto.
17350 (aarch64_fmlal_lane_lowv2sf): Ditto.
17351 (aarch64_fmlsl_lane_lowv2sf): Ditto.
17352 (aarch64_fmlal_laneq_lowv2sf): Ditto.
17353 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
17354 (aarch64_fmlalq_lane_lowv4sf): Ditto.
17355 (aarch64_fmlsl_lane_lowv4sf): Ditto.
17356 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
17357 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
17358 (aarch64_fmlal_lane_highv2sf): Ditto.
17359 (aarch64_fmlsl_lane_highv2sf): Ditto.
17360 (aarch64_fmlal_laneq_highv2sf): Ditto.
17361 (aarch64_fmlsl_laneq_highv2sf): Ditto.
17362 (aarch64_fmlalq_lane_highv4sf): Ditto.
17363 (aarch64_fmlsl_lane_highv4sf): Ditto.
17364 (aarch64_fmlalq_laneq_highv4sf): Ditto.
17365 (aarch64_fmlsl_laneq_highv4sf): Ditto.
17366 * config/aarch64/aarch64-simd.md:
17367 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
17368 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
17369 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
17370 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
17371 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
17372 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
17373 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
17374 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
17375 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
17376 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
17377 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
17378 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
17379 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
17380 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
17381 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
17382 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
17383 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
17384 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
17385 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
17386 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
17387 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
17388 (vfmlsl_low_u32): Ditto.
17389 (vfmlalq_low_u32): Ditto.
17390 (vfmlslq_low_u32): Ditto.
17391 (vfmlal_high_u32): Ditto.
17392 (vfmlsl_high_u32): Ditto.
17393 (vfmlalq_high_u32): Ditto.
17394 (vfmlslq_high_u32): Ditto.
17395 (vfmlal_lane_low_u32): Ditto.
17396 (vfmlsl_lane_low_u32): Ditto.
17397 (vfmlal_laneq_low_u32): Ditto.
17398 (vfmlsl_laneq_low_u32): Ditto.
17399 (vfmlalq_lane_low_u32): Ditto.
17400 (vfmlslq_lane_low_u32): Ditto.
17401 (vfmlalq_laneq_low_u32): Ditto.
17402 (vfmlslq_laneq_low_u32): Ditto.
17403 (vfmlal_lane_high_u32): Ditto.
17404 (vfmlsl_lane_high_u32): Ditto.
17405 (vfmlal_laneq_high_u32): Ditto.
17406 (vfmlsl_laneq_high_u32): Ditto.
17407 (vfmlalq_lane_high_u32): Ditto.
17408 (vfmlslq_lane_high_u32): Ditto.
17409 (vfmlalq_laneq_high_u32): Ditto.
17410 (vfmlslq_laneq_high_u32): Ditto.
17411 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
17412 (AARCH64_FL_FOR_ARCH8_4): New.
17413 (AARCH64_ISA_F16FML): New ISA flag.
17414 (TARGET_F16FML): New feature flag for fp16fml.
17415 (doc/invoke.texi): Document new fp16fml option.
17417 2018-01-10 Michael Collison <michael.collison@arm.com>
17419 * config/aarch64/aarch64-builtins.c:
17420 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
17421 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
17422 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
17423 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
17424 (AARCH64_ISA_SHA3): New ISA flag.
17425 (TARGET_SHA3): New feature flag for sha3.
17426 * config/aarch64/iterators.md (sha512_op): New int attribute.
17427 (CRYPTO_SHA512): New int iterator.
17428 (UNSPEC_SHA512H): New unspec.
17429 (UNSPEC_SHA512H2): Ditto.
17430 (UNSPEC_SHA512SU0): Ditto.
17431 (UNSPEC_SHA512SU1): Ditto.
17432 * config/aarch64/aarch64-simd-builtins.def
17433 (aarch64_crypto_sha512hqv2di): New builtin.
17434 (aarch64_crypto_sha512h2qv2di): Ditto.
17435 (aarch64_crypto_sha512su0qv2di): Ditto.
17436 (aarch64_crypto_sha512su1qv2di): Ditto.
17437 (aarch64_eor3qv8hi): Ditto.
17438 (aarch64_rax1qv2di): Ditto.
17439 (aarch64_xarqv2di): Ditto.
17440 (aarch64_bcaxqv8hi): Ditto.
17441 * config/aarch64/aarch64-simd.md:
17442 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
17443 (aarch64_crypto_sha512su0qv2di): Ditto.
17444 (aarch64_crypto_sha512su1qv2di): Ditto.
17445 (aarch64_eor3qv8hi): Ditto.
17446 (aarch64_rax1qv2di): Ditto.
17447 (aarch64_xarqv2di): Ditto.
17448 (aarch64_bcaxqv8hi): Ditto.
17449 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
17450 (vsha512h2q_u64): Ditto.
17451 (vsha512su0q_u64): Ditto.
17452 (vsha512su1q_u64): Ditto.
17453 (veor3q_u16): Ditto.
17454 (vrax1q_u64): Ditto.
17455 (vxarq_u64): Ditto.
17456 (vbcaxq_u16): Ditto.
17457 * config/arm/types.md (crypto_sha512): New type attribute.
17458 (crypto_sha3): Ditto.
17459 (doc/invoke.texi): Document new sha3 option.
17461 2018-01-10 Michael Collison <michael.collison@arm.com>
17463 * config/aarch64/aarch64-builtins.c:
17464 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
17465 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
17466 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
17467 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
17468 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
17469 (AARCH64_ISA_SM4): New ISA flag.
17470 (TARGET_SM4): New feature flag for sm4.
17471 * config/aarch64/aarch64-simd-builtins.def
17472 (aarch64_sm3ss1qv4si): Ditto.
17473 (aarch64_sm3tt1aq4si): Ditto.
17474 (aarch64_sm3tt1bq4si): Ditto.
17475 (aarch64_sm3tt2aq4si): Ditto.
17476 (aarch64_sm3tt2bq4si): Ditto.
17477 (aarch64_sm3partw1qv4si): Ditto.
17478 (aarch64_sm3partw2qv4si): Ditto.
17479 (aarch64_sm4eqv4si): Ditto.
17480 (aarch64_sm4ekeyqv4si): Ditto.
17481 * config/aarch64/aarch64-simd.md:
17482 (aarch64_sm3ss1qv4si): Ditto.
17483 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
17484 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
17485 (aarch64_sm4eqv4si): Ditto.
17486 (aarch64_sm4ekeyqv4si): Ditto.
17487 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
17488 (sm3part_op): Ditto.
17489 (CRYPTO_SM3TT): Ditto.
17490 (CRYPTO_SM3PART): Ditto.
17491 (UNSPEC_SM3SS1): New unspec.
17492 (UNSPEC_SM3TT1A): Ditto.
17493 (UNSPEC_SM3TT1B): Ditto.
17494 (UNSPEC_SM3TT2A): Ditto.
17495 (UNSPEC_SM3TT2B): Ditto.
17496 (UNSPEC_SM3PARTW1): Ditto.
17497 (UNSPEC_SM3PARTW2): Ditto.
17498 (UNSPEC_SM4E): Ditto.
17499 (UNSPEC_SM4EKEY): Ditto.
17500 * config/aarch64/constraints.md (Ui2): New constraint.
17501 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
17502 * config/arm/types.md (crypto_sm3): New type attribute.
17503 (crypto_sm4): Ditto.
17504 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
17505 (vsm3tt1aq_u32): Ditto.
17506 (vsm3tt1bq_u32): Ditto.
17507 (vsm3tt2aq_u32): Ditto.
17508 (vsm3tt2bq_u32): Ditto.
17509 (vsm3partw1q_u32): Ditto.
17510 (vsm3partw2q_u32): Ditto.
17511 (vsm4eq_u32): Ditto.
17512 (vsm4ekeyq_u32): Ditto.
17513 (doc/invoke.texi): Document new sm4 option.
17515 2018-01-10 Michael Collison <michael.collison@arm.com>
17517 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
17518 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
17519 (AARCH64_FL_FOR_ARCH8_4): New.
17520 (AARCH64_FL_V8_4): New flag.
17521 (doc/invoke.texi): Document new armv8.4-a option.
17523 2018-01-10 Michael Collison <michael.collison@arm.com>
17525 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
17526 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
17527 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
17528 * config/aarch64/aarch64-option-extension.def: Add
17529 AARCH64_OPT_EXTENSION of 'sha2'.
17530 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
17531 (crypto): Disable sha2 and aes if crypto disabled.
17532 (crypto): Enable aes and sha2 if enabled.
17533 (simd): Disable sha2 and aes if simd disabled.
17534 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
17536 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
17537 (TARGET_SHA2): New feature flag for sha2.
17538 (TARGET_AES): New feature flag for aes.
17539 * config/aarch64/aarch64-simd.md:
17540 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
17541 conditional on TARGET_AES.
17542 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
17543 (aarch64_crypto_sha1hsi): Make pattern conditional
17545 (aarch64_crypto_sha1hv4si): Ditto.
17546 (aarch64_be_crypto_sha1hv4si): Ditto.
17547 (aarch64_crypto_sha1su1v4si): Ditto.
17548 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
17549 (aarch64_crypto_sha1su0v4si): Ditto.
17550 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
17551 (aarch64_crypto_sha256su0v4si): Ditto.
17552 (aarch64_crypto_sha256su1v4si): Ditto.
17553 (doc/invoke.texi): Document new aes and sha2 options.
17555 2018-01-10 Martin Sebor <msebor@redhat.com>
17557 PR tree-optimization/83781
17558 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
17561 2018-01-11 Martin Sebor <msebor@gmail.com>
17562 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17564 PR tree-optimization/83501
17565 PR tree-optimization/81703
17567 * tree-ssa-strlen.c (get_string_cst): Rename...
17568 (get_string_len): ...to this. Handle global constants.
17569 (handle_char_store): Adjust.
17571 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
17572 Jim Wilson <jimw@sifive.com>
17574 * config/riscv/riscv-protos.h (riscv_output_return): New.
17575 * config/riscv/riscv.c (struct machine_function): New naked_p field.
17576 (riscv_attribute_table, riscv_output_return),
17577 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
17578 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
17579 (riscv_compute_frame_info): Only compute frame->mask if not a naked
17581 (riscv_expand_prologue): Add early return for naked function.
17582 (riscv_expand_epilogue): Likewise.
17583 (riscv_function_ok_for_sibcall): Return false for naked function.
17584 (riscv_set_current_function): New.
17585 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
17586 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
17587 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
17588 * doc/extend.texi (RISC-V Function Attributes): New.
17590 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
17592 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
17593 check for 128-bit long double before checking TCmode.
17594 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
17595 128-bit long doubles before checking TFmode or TCmode.
17596 (FLOAT128_IBM_P): Likewise.
17598 2018-01-10 Martin Sebor <msebor@redhat.com>
17600 PR tree-optimization/83671
17601 * builtins.c (c_strlen): Unconditionally return zero for the empty
17603 Use -Warray-bounds for warnings.
17604 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
17605 for non-constant array indices with COMPONENT_REF, arrays of
17606 arrays, and pointers to arrays.
17607 (gimple_fold_builtin_strlen): Determine and set length range for
17608 non-constant character arrays.
17610 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
17612 PR middle-end/81897
17613 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
17616 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
17618 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
17620 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
17623 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
17624 VECTOR_MEM_ALTIVEC_OR_VSX_P.
17625 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
17626 indexed_or_indirect_operand predicate.
17627 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
17628 (*vsx_le_perm_load_v8hi): Likewise.
17629 (*vsx_le_perm_load_v16qi): Likewise.
17630 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
17631 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
17632 (*vsx_le_perm_store_v8hi): Likewise.
17633 (*vsx_le_perm_store_v16qi): Likewise.
17634 (eight unnamed splitters): Likewise.
17636 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
17638 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
17639 * config/rs6000/emmintrin.h: Likewise.
17640 * config/rs6000/mmintrin.h: Likewise.
17641 * config/rs6000/xmmintrin.h: Likewise.
17643 2018-01-10 David Malcolm <dmalcolm@redhat.com>
17646 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
17648 * tree.c (tree_nop_conversion): Return true for location wrapper
17650 (maybe_wrap_with_location): New function.
17651 (selftest::check_strip_nops): New function.
17652 (selftest::test_location_wrappers): New function.
17653 (selftest::tree_c_tests): Call it.
17654 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
17655 (maybe_wrap_with_location): New decl.
17656 (EXPR_LOCATION_WRAPPER_P): New macro.
17657 (location_wrapper_p): New inline function.
17658 (tree_strip_any_location_wrapper): New inline function.
17660 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
17663 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
17664 stack_realign_offset for the largest alignment of stack slot
17666 (ix86_find_max_used_stack_alignment): New function.
17667 (ix86_finalize_stack_frame_flags): Use it. Set
17668 max_used_stack_alignment if we don't realign stack.
17669 * config/i386/i386.h (machine_function): Add
17670 max_used_stack_alignment.
17672 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
17674 * config/arm/arm.opt (-mbranch-cost): New option.
17675 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
17678 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
17681 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
17682 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
17684 2018-01-10 Richard Biener <rguenther@suse.de>
17687 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
17688 early out so it also covers the case where we have a non-NULL
17691 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
17693 PR tree-optimization/83753
17694 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
17695 for non-strided grouped accesses if the number of elements is 1.
17697 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
17700 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
17701 * i386.h (TARGET_USE_GATHER): Define.
17702 * x86-tune.def (X86_TUNE_USE_GATHER): New.
17704 2018-01-10 Martin Liska <mliska@suse.cz>
17707 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
17708 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
17710 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
17711 CLEANUP_NO_PARTITIONING is not set.
17713 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
17715 * doc/rtl.texi: Remove documentation of (const ...) wrappers
17716 for vectors, as a partial revert of r254296.
17717 * rtl.h (const_vec_p): Delete.
17718 (const_vec_duplicate_p): Don't test for vector CONSTs.
17719 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
17720 * expmed.c (make_tree): Likewise.
17723 * common.md (E, F): Use CONSTANT_P instead of checking for
17725 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
17726 checking for CONST_VECTOR.
17728 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
17730 PR middle-end/83575
17731 * predict.c (force_edge_cold): Handle in more sane way edges
17732 with no prediction.
17734 2018-01-09 Carl Love <cel@us.ibm.com>
17736 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
17738 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
17739 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
17740 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
17741 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
17742 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
17743 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
17744 * config/rs6000/rs6000-protos.h: Add extern defition for
17745 rs6000_generate_float2_double_code.
17746 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
17748 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
17749 (float2_v2df): Add define_expand.
17751 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
17754 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
17755 op_mode in the force_to_mode call.
17757 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
17759 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
17760 instead of checking each element individually.
17761 (aarch64_evpc_uzp): Likewise.
17762 (aarch64_evpc_zip): Likewise.
17763 (aarch64_evpc_ext): Likewise.
17764 (aarch64_evpc_rev): Likewise.
17765 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
17766 instead of checking each element individually. Return true without
17768 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
17769 whether all selected elements come from the same input, instead of
17770 checking each element individually. Remove calls to gen_rtx_REG,
17771 start_sequence and end_sequence and instead assert that no rtl is
17774 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
17776 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
17777 order of HIGH and CONST checks.
17779 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
17781 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
17782 if the destination isn't an SSA_NAME.
17784 2018-01-09 Richard Biener <rguenther@suse.de>
17786 PR tree-optimization/83668
17787 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
17789 (canonicalize_loop_form): ... here, renamed from ...
17790 (canonicalize_loop_closed_ssa_form): ... this and amended to
17791 swap successor edges for loop exit blocks to make us use
17792 the RPO order we need for initial schedule generation.
17794 2018-01-09 Joseph Myers <joseph@codesourcery.com>
17796 PR tree-optimization/64811
17797 * match.pd: When optimizing comparisons with Inf, avoid
17798 introducing or losing exceptions from comparisons with NaN.
17800 2018-01-09 Martin Liska <mliska@suse.cz>
17803 * asan.c (shadow_mem_size): Add gcc_assert.
17805 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
17807 Don't save registers in main().
17810 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
17811 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
17812 * config/avr/avr.c (avr_set_current_function): Don't error if
17813 naked, OS_task or OS_main are specified at the same time.
17814 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
17816 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
17818 * common/config/avr/avr-common.c (avr_option_optimization_table):
17819 Switch on -mmain-is-OS_task for optimizing compilations.
17821 2018-01-09 Richard Biener <rguenther@suse.de>
17823 PR tree-optimization/83572
17824 * graphite.c: Include cfganal.h.
17825 (graphite_transform_loops): Connect infinite loops to exit
17826 and remove fake edges at the end.
17828 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
17830 * ipa-inline.c (edge_badness): Revert accidental checkin.
17832 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
17835 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
17836 symbols; not inline clones.
17838 2018-01-09 Jakub Jelinek <jakub@redhat.com>
17841 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
17842 hard registers. Formatting fixes.
17844 PR preprocessor/83722
17845 * gcc.c (try_generate_repro): Pass
17846 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
17847 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
17850 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
17851 Kito Cheng <kito.cheng@gmail.com>
17853 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
17854 (riscv_leaf_function_p): Delete.
17855 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
17857 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17859 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
17861 (do_ifelse): New function.
17862 (do_isel): New function.
17863 (do_sub3): New function.
17864 (do_add3): New function.
17865 (do_load_mask_compare): New function.
17866 (do_overlap_load_compare): New function.
17867 (expand_compare_loop): New function.
17868 (expand_block_compare): Call expand_compare_loop() when appropriate.
17869 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
17870 option description.
17871 (-mblock-compare-inline-loop-limit): New option.
17873 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17876 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
17877 Reverse order of second and third operands in first alternative.
17878 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
17879 of first and second elements in UNSPEC_VPERMR vector.
17880 (altivec_expand_vec_perm_le): Likewise.
17882 2018-01-08 Jeff Law <law@redhat.com>
17884 PR rtl-optimizatin/81308
17885 * tree-switch-conversion.c (cfg_altered): New file scoped static.
17886 (process_switch): If group_case_labels makes a change, then set
17888 (pass_convert_switch::execute): If a switch is converted, then
17889 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
17891 PR rtl-optimization/81308
17892 * recog.c (split_all_insns): Conditionally cleanup the CFG after
17895 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
17897 PR target/83663 - Revert r255946
17898 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
17899 generation for cases where splatting a value is not useful.
17900 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
17901 across a vec_duplicate and a paradoxical subreg forming a vector
17902 mode to a vec_concat.
17904 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17906 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
17907 -march=armv8.3-a variants.
17908 * config/arm/t-multilib: Likewise.
17909 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
17911 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17913 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
17915 (cceq_ior_compare_complement): Give it a name so I can use it, and
17916 change boolean_or_operator predicate to boolean_operator so it can
17917 be used to generate a crand.
17918 (eqne): New code iterator.
17919 (bd/bd_neg): New code_attrs.
17920 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
17921 a single define_insn.
17922 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
17923 decrement (bdnzt/bdnzf/bdzt/bdzf).
17924 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
17925 with the new names of the branch decrement patterns, and added the
17926 names of the branch decrement conditional patterns.
17928 2018-01-08 Richard Biener <rguenther@suse.de>
17930 PR tree-optimization/83563
17931 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
17934 2018-01-08 Richard Biener <rguenther@suse.de>
17936 PR middle-end/83713
17937 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
17939 2018-01-08 Richard Biener <rguenther@suse.de>
17941 PR tree-optimization/83685
17942 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
17943 references to abnormals.
17945 2018-01-08 Richard Biener <rguenther@suse.de>
17948 * dwarf2out.c (output_indirect_strings): Handle empty
17949 skeleton_debug_str_hash.
17950 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
17952 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
17954 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
17955 (emit_store_direct): Likewise.
17956 (arc_trampoline_adjust_address): Likewise.
17957 (arc_asm_trampoline_template): New function.
17958 (arc_initialize_trampoline): Use asm_trampoline_template.
17959 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
17960 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
17961 * config/arc/arc.md (flush_icache): Delete pattern.
17963 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
17965 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
17966 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
17969 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
17972 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
17973 by not USED_FOR_TARGET.
17974 (make_pass_resolve_sw_modes): Likewise.
17976 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
17978 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
17981 2018-01-08 Richard Biener <rguenther@suse.de>
17983 PR middle-end/83580
17984 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
17986 2018-01-08 Richard Biener <rguenther@suse.de>
17988 PR middle-end/83517
17989 * match.pd ((t * 2) / 2) -> t): Add missing :c.
17991 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
17993 PR middle-end/81897
17994 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
17995 basic blocks with a small number of successors.
17996 (convert_control_dep_chain_into_preds): Improve handling of
17998 (dump_predicates): Split apart into...
17999 (dump_pred_chain): ...here...
18000 (dump_pred_info): ...and here.
18001 (can_one_predicate_be_invalidated_p): Add debugging printfs.
18002 (can_chain_union_be_invalidated_p): Improve check for invalidation
18004 (uninit_uses_cannot_happen): Avoid unnecessary if
18005 convert_control_dep_chain_into_preds yielded nothing.
18007 2018-01-06 Martin Sebor <msebor@redhat.com>
18009 PR tree-optimization/83640
18010 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
18011 subtracting negative offset from size.
18012 (builtin_access::overlap): Adjust offset bounds of the access to fall
18013 within the size of the object if possible.
18015 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
18017 PR rtl-optimization/83699
18018 * expmed.c (extract_bit_field_1): Restrict the vector usage of
18019 extract_bit_field_as_subreg to cases in which the extracted
18020 value is also a vector.
18022 * lra-constraints.c (process_alt_operands): Test for the equivalence
18023 substitutions when detecting a possible reload cycle.
18025 2018-01-06 Jakub Jelinek <jakub@redhat.com>
18028 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
18029 by default if flag_selective_schedling{,2}. Formatting fixes.
18031 PR rtl-optimization/83682
18032 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
18033 if it has non-VECTOR_MODE element mode.
18034 (vec_duplicate_p): Likewise.
18036 PR middle-end/83694
18037 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
18038 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
18040 2018-01-05 Jakub Jelinek <jakub@redhat.com>
18043 * config/i386/i386-builtin.def
18044 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
18045 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
18046 Require also OPTION_MASK_ISA_AVX512F in addition to
18047 OPTION_MASK_ISA_GFNI.
18048 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
18049 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
18050 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
18051 to OPTION_MASK_ISA_GFNI.
18052 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
18053 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
18054 OPTION_MASK_ISA_AVX512BW.
18055 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
18056 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
18057 addition to OPTION_MASK_ISA_GFNI.
18058 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
18059 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
18060 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
18061 to OPTION_MASK_ISA_GFNI.
18062 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
18063 a requirement for all ISAs rather than any of them with a few
18065 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
18067 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
18068 bitmasks to be enabled with 3 exceptions, instead of requiring any
18069 enabled ISA with lots of exceptions.
18070 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
18071 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
18072 Change avx512bw in isa attribute to avx512f.
18073 * config/i386/sgxintrin.h: Add license boilerplate.
18074 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
18075 to __AVX512F__ and __AVX512VL to __AVX512VL__.
18076 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
18077 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
18079 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
18080 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
18081 temporarily sse2 rather than sse if not enabled already.
18084 * config/i386/sse.md (VI248_VLBW): Rename to ...
18085 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
18086 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
18087 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
18088 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
18089 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
18090 mode iterator instead of VI248_VLBW.
18092 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
18094 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
18095 (record_modified): Skip clobbers; add debug output.
18096 (param_change_prob): Use sreal frequencies.
18098 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
18100 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
18101 punt for user-aligned variables.
18103 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
18105 * tree-chrec.c (chrec_contains_symbols): Return true for
18108 2018-01-05 Sudakshina Das <sudi.das@arm.com>
18111 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
18112 of (x|y) == x for BICS pattern.
18114 2018-01-05 Jakub Jelinek <jakub@redhat.com>
18116 PR tree-optimization/83605
18117 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
18118 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
18121 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
18123 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
18124 * config/epiphany/rtems.h: New file.
18126 2018-01-04 Jakub Jelinek <jakub@redhat.com>
18127 Uros Bizjak <ubizjak@gmail.com>
18130 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
18131 QIreg_operand instead of register_operand predicate.
18132 * config/i386/i386.c (ix86_rop_should_change_byte_p,
18133 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
18134 comments instead of -fmitigate[-_]rop.
18136 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18139 * cgraphunit.c (symbol_table::compile): Switch to text_section
18140 before calling assembly_start debug hook.
18141 * run-rtl-passes.c (run_rtl_passes): Likewise.
18144 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
18146 * tree-vrp.c (extract_range_from_binary_expr_1): Check
18147 range_int_cst_p rather than !symbolic_range_p before calling
18148 extract_range_from_multiplicative_op_1.
18150 2018-01-04 Jeff Law <law@redhat.com>
18152 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
18153 redundant test in assertion.
18155 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
18157 * doc/rtl.texi: Document machine_mode wrapper classes.
18159 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
18161 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
18162 using tree_to_uhwi.
18164 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
18166 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
18167 the VEC_PERM_EXPR fold to fail.
18169 2018-01-04 Jakub Jelinek <jakub@redhat.com>
18172 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
18173 to switched_sections.
18175 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
18178 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
18179 test for d.testing.
18181 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
18184 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
18185 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
18187 2018-01-04 Jakub Jelinek <jakub@redhat.com>
18190 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
18191 is BLKmode and bitpos not zero or mode change is needed.
18193 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
18196 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
18199 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
18202 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
18203 instead of MULT rtx. Update all corresponding splitters.
18204 (*saddl_se): Ditto.
18205 (*ssub<modesuffix>): Ditto.
18206 (*ssubl_se): Ditto.
18207 (*cmp_sadd_di): Update split patterns.
18208 (*cmp_sadd_si): Ditto.
18209 (*cmp_sadd_sidi): Ditto.
18210 (*cmp_ssub_di): Ditto.
18211 (*cmp_ssub_si): Ditto.
18212 (*cmp_ssub_sidi): Ditto.
18213 * config/alpha/predicates.md (const23_operand): New predicate.
18214 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
18215 Look for ASHIFT, not MULT inner operand.
18216 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
18218 2018-01-04 Martin Liska <mliska@suse.cz>
18220 PR gcov-profile/83669
18221 * gcov.c (output_intermediate_file): Add version to intermediate
18223 * doc/gcov.texi: Document new field 'version' in intermediate
18224 file format. Fix location of '-k' option of gcov command.
18226 2018-01-04 Martin Liska <mliska@suse.cz>
18229 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
18231 2018-01-04 Jakub Jelinek <jakub@redhat.com>
18233 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
18235 2018-01-03 Martin Sebor <msebor@redhat.com>
18237 PR tree-optimization/83655
18238 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
18239 checking calls with invalid arguments.
18241 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18243 * tree-vect-stmts.c (vect_get_store_rhs): New function.
18244 (vectorizable_mask_load_store): Delete.
18245 (vectorizable_call): Return false for masked loads and stores.
18246 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
18247 instead of gimple_assign_rhs1.
18248 (vectorizable_load): Handle IFN_MASK_LOAD.
18249 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
18251 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18253 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
18255 (vectorizable_mask_load_store): ...here.
18256 (vectorizable_load): ...and here.
18258 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18260 * tree-vect-stmts.c (vect_build_all_ones_mask)
18261 (vect_build_zero_merge_argument): New functions, split out from...
18262 (vectorizable_load): ...here.
18264 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18266 * tree-vect-stmts.c (vect_check_store_rhs): New function,
18268 (vectorizable_mask_load_store): ...here.
18269 (vectorizable_store): ...and here.
18271 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18273 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
18275 (vectorizable_mask_load_store): ...here.
18277 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18279 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
18280 (vect_model_store_cost): Take a vec_load_store_type instead of a
18282 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
18283 (vect_model_store_cost): Take a vec_load_store_type instead of a
18285 (vectorizable_mask_load_store): Update accordingly.
18286 (vectorizable_store): Likewise.
18287 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
18289 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18291 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
18292 IFN_MASK_LOAD calls here rather than...
18293 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
18295 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18296 Alan Hayward <alan.hayward@arm.com>
18297 David Sherwood <david.sherwood@arm.com>
18299 * expmed.c (extract_bit_field_1): For vector extracts,
18300 fall back to extract_bit_field_as_subreg if vec_extract
18303 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18304 Alan Hayward <alan.hayward@arm.com>
18305 David Sherwood <david.sherwood@arm.com>
18307 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
18308 they are variable or constant sized.
18309 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
18310 slots for constant-sized data.
18312 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18313 Alan Hayward <alan.hayward@arm.com>
18314 David Sherwood <david.sherwood@arm.com>
18316 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
18317 handling COND_EXPRs with boolean comparisons, try to find a better
18318 basis for the mask type than the boolean itself.
18320 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18322 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
18323 is calculated and how it can be overridden.
18324 * genmodes.c (max_bitsize_mode_any_mode): New variable.
18325 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
18327 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
18330 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18331 Alan Hayward <alan.hayward@arm.com>
18332 David Sherwood <david.sherwood@arm.com>
18334 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
18335 Remove the mode argument.
18336 (aarch64_simd_valid_immediate): Remove the mode and inverse
18338 * config/aarch64/iterators.md (bitsize): New iterator.
18339 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
18340 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
18341 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
18342 aarch64_simd_valid_immediate.
18343 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
18344 (aarch64_reg_or_bic_imm): Likewise.
18345 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
18346 with an insn_type enum and msl with a modifier_type enum.
18347 Replace element_width with a scalar_mode. Change the shift
18348 to unsigned int. Add constructors for scalar_float_mode and
18349 scalar_int_mode elements.
18350 (aarch64_vect_float_const_representable_p): Delete.
18351 (aarch64_can_const_movi_rtx_p)
18352 (aarch64_simd_scalar_immediate_valid_for_move)
18353 (aarch64_simd_make_constant): Update call to
18354 aarch64_simd_valid_immediate.
18355 (aarch64_advsimd_valid_immediate_hs): New function.
18356 (aarch64_advsimd_valid_immediate): Likewise.
18357 (aarch64_simd_valid_immediate): Remove mode and inverse
18358 arguments. Rewrite to use the above. Use const_vec_duplicate_p
18359 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
18360 and aarch64_float_const_representable_p on the result.
18361 (aarch64_output_simd_mov_immediate): Remove mode argument.
18362 Update call to aarch64_simd_valid_immediate and use of
18363 simd_immediate_info.
18364 (aarch64_output_scalar_simd_mov_immediate): Update call
18367 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18368 Alan Hayward <alan.hayward@arm.com>
18369 David Sherwood <david.sherwood@arm.com>
18371 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
18372 (mode_nunits): Likewise CONST_MODE_NUNITS.
18373 * machmode.def (ADJUST_NUNITS): Document.
18374 * genmodes.c (mode_data::need_nunits_adj): New field.
18375 (blank_mode): Update accordingly.
18376 (adj_nunits): New variable.
18377 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
18379 (emit_mode_size_inline): Set need_bytesize_adj for all modes
18380 listed in adj_nunits.
18381 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
18382 listed in adj_nunits. Don't emit case statements for such modes.
18383 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
18384 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
18385 nothing if adj_nunits is nonnull.
18386 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
18387 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
18388 (emit_mode_fbit): Update use of print_maybe_const_decl.
18389 (emit_move_size): Likewise. Treat the array as non-const
18391 (emit_mode_adjustments): Handle adj_nunits.
18393 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18395 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
18396 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
18397 (VECTOR_MODES): Use it.
18398 (make_vector_modes): Take the prefix as an argument.
18400 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18401 Alan Hayward <alan.hayward@arm.com>
18402 David Sherwood <david.sherwood@arm.com>
18404 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
18405 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
18406 for MODE_VECTOR_BOOL.
18407 * machmode.def (VECTOR_BOOL_MODE): Document.
18408 * genmodes.c (VECTOR_BOOL_MODE): New macro.
18409 (make_vector_bool_mode): New function.
18410 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
18412 * lto-streamer-in.c (lto_input_mode_table): Likewise.
18413 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
18415 * stor-layout.c (int_mode_for_mode): Likewise.
18416 * tree.c (build_vector_type_for_mode): Likewise.
18417 * varasm.c (output_constant_pool_2): Likewise.
18418 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
18419 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
18420 for MODE_VECTOR_BOOL.
18421 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
18422 of mode class checks.
18423 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
18424 instead of a list of mode class checks.
18425 (expand_vector_scalar_condition): Likewise.
18426 (type_for_widest_vector_mode): Handle BImode as an inner mode.
18428 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18429 Alan Hayward <alan.hayward@arm.com>
18430 David Sherwood <david.sherwood@arm.com>
18432 * machmode.h (mode_size): Change from unsigned short to
18434 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
18435 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
18436 or if measurement_type is not polynomial.
18437 (fixed_size_mode::includes_p): Check for constant-sized modes.
18438 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
18439 return a poly_uint16 rather than an unsigned short.
18440 (emit_mode_size): Change the type of mode_size from unsigned short
18441 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
18442 (emit_mode_adjustments): Cope with polynomial vector sizes.
18443 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
18445 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
18447 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
18448 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
18449 * caller-save.c (setup_save_areas): Likewise.
18450 (replace_reg_with_saved_mem): Likewise.
18451 * calls.c (emit_library_call_value_1): Likewise.
18452 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
18453 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
18454 (gen_lowpart_for_combine): Likewise.
18455 * convert.c (convert_to_integer_1): Likewise.
18456 * cse.c (equiv_constant, cse_insn): Likewise.
18457 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
18458 (cselib_subst_to_values): Likewise.
18459 * dce.c (word_dce_process_block): Likewise.
18460 * df-problems.c (df_word_lr_mark_ref): Likewise.
18461 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
18462 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
18463 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
18464 (rtl_for_decl_location): Likewise.
18465 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
18466 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
18467 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
18468 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
18469 (expand_expr_real_1): Likewise.
18470 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
18471 (pad_below): Likewise.
18472 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
18473 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
18474 * ira.c (get_subreg_tracking_sizes): Likewise.
18475 * ira-build.c (ira_create_allocno_objects): Likewise.
18476 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
18477 (ira_sort_regnos_for_alter_reg): Likewise.
18478 * ira-costs.c (record_operand_costs): Likewise.
18479 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
18480 (resolve_simple_move): Likewise.
18481 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
18482 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
18483 (lra_constraints): Likewise.
18484 (CONST_POOL_OK_P): Reject variable-sized modes.
18485 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
18486 (add_pseudo_to_slot, lra_spill): Likewise.
18487 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18488 * optabs-query.c (get_best_extraction_insn): Likewise.
18489 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
18490 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
18491 (expand_mult_highpart, valid_multiword_target_p): Likewise.
18492 * recog.c (offsettable_address_addr_space_p): Likewise.
18493 * regcprop.c (maybe_mode_change): Likewise.
18494 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
18495 * regrename.c (build_def_use): Likewise.
18496 * regstat.c (dump_reg_info): Likewise.
18497 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
18498 (find_reloads, find_reloads_subreg_address): Likewise.
18499 * reload1.c (eliminate_regs_1): Likewise.
18500 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
18501 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
18502 (simplify_binary_operation_1, simplify_subreg): Likewise.
18503 * targhooks.c (default_function_arg_padding): Likewise.
18504 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
18505 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
18506 (verify_gimple_assign_ternary): Likewise.
18507 * tree-inline.c (estimate_move_cost): Likewise.
18508 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
18509 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
18510 (get_address_cost_ainc): Likewise.
18511 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
18512 (vect_supportable_dr_alignment): Likewise.
18513 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
18514 (vectorizable_reduction): Likewise.
18515 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
18516 (vectorizable_operation, vectorizable_load): Likewise.
18517 * tree.c (build_same_sized_truth_vector_type): Likewise.
18518 * valtrack.c (cleanup_auto_inc_dec): Likewise.
18519 * var-tracking.c (emit_note_insn_var_location): Likewise.
18520 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
18521 (ADDR_VEC_ALIGN): Likewise.
18523 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18524 Alan Hayward <alan.hayward@arm.com>
18525 David Sherwood <david.sherwood@arm.com>
18527 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
18529 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
18530 or if measurement_type is polynomial.
18531 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
18532 * combine.c (make_extraction): Likewise.
18533 * dse.c (find_shift_sequence): Likewise.
18534 * dwarf2out.c (mem_loc_descriptor): Likewise.
18535 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
18536 (extract_bit_field, extract_low_bits): Likewise.
18537 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
18538 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
18539 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
18540 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
18541 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
18542 * reload.c (find_reloads): Likewise.
18543 * reload1.c (alter_reg): Likewise.
18544 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
18545 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
18546 * tree-if-conv.c (predicate_mem_writes): Likewise.
18547 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
18548 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
18549 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
18550 * valtrack.c (dead_debug_insert_temp): Likewise.
18551 * varasm.c (mergeable_constant_section): Likewise.
18552 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
18554 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18555 Alan Hayward <alan.hayward@arm.com>
18556 David Sherwood <david.sherwood@arm.com>
18558 * expr.c (expand_assignment): Cope with polynomial mode sizes
18559 when assigning to a CONCAT.
18561 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18562 Alan Hayward <alan.hayward@arm.com>
18563 David Sherwood <david.sherwood@arm.com>
18565 * machmode.h (mode_precision): Change from unsigned short to
18567 (mode_to_precision): Return a poly_uint16 rather than an unsigned
18569 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
18570 or if measurement_type is not polynomial.
18571 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
18572 in which the mode is already known to be a scalar_int_mode.
18573 * genmodes.c (emit_mode_precision): Change the type of mode_precision
18574 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
18576 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
18577 for GET_MODE_PRECISION.
18578 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
18579 for GET_MODE_PRECISION.
18580 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
18582 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
18583 (expand_field_assignment, make_extraction): Likewise.
18584 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
18585 (get_last_value): Likewise.
18586 * convert.c (convert_to_integer_1): Likewise.
18587 * cse.c (cse_insn): Likewise.
18588 * expr.c (expand_expr_real_1): Likewise.
18589 * lra-constraints.c (simplify_operand_subreg): Likewise.
18590 * optabs-query.c (can_atomic_load_p): Likewise.
18591 * optabs.c (expand_atomic_load): Likewise.
18592 (expand_atomic_store): Likewise.
18593 * ree.c (combine_reaching_defs): Likewise.
18594 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
18595 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
18596 * tree.h (type_has_mode_precision_p): Likewise.
18597 * ubsan.c (instrument_si_overflow): Likewise.
18599 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18600 Alan Hayward <alan.hayward@arm.com>
18601 David Sherwood <david.sherwood@arm.com>
18603 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
18604 polynomial numbers of units.
18605 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
18606 (valid_vector_subparts_p): New function.
18607 (build_vector_type): Remove temporary shim and take the number
18608 of units as a poly_uint64 rather than an int.
18609 (build_opaque_vector_type): Take the number of units as a
18610 poly_uint64 rather than an int.
18611 * tree.c (build_vector_from_ctor): Handle polynomial
18612 TYPE_VECTOR_SUBPARTS.
18613 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
18614 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
18615 (build_vector_from_val): If the number of units is variable,
18616 use build_vec_duplicate_cst for constant operands and
18617 VEC_DUPLICATE_EXPR otherwise.
18618 (make_vector_type): Remove temporary is_constant ().
18619 (build_vector_type, build_opaque_vector_type): Take the number of
18620 units as a poly_uint64 rather than an int.
18621 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
18623 * cfgexpand.c (expand_debug_expr): Likewise.
18624 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
18625 (store_constructor, expand_expr_real_1): Likewise.
18626 (const_scalar_mask_from_tree): Likewise.
18627 * fold-const-call.c (fold_const_reduction): Likewise.
18628 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
18629 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
18630 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
18631 (fold_relational_const): Likewise.
18632 (native_interpret_vector): Likewise. Change the size from an
18633 int to an unsigned int.
18634 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
18635 TYPE_VECTOR_SUBPARTS.
18636 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
18637 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
18638 duplicating a non-constant operand into a variable-length vector.
18639 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
18640 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
18641 * ipa-icf.c (sem_variable::equals): Likewise.
18642 * match.pd: Likewise.
18643 * omp-simd-clone.c (simd_clone_subparts): Likewise.
18644 * print-tree.c (print_node): Likewise.
18645 * stor-layout.c (layout_type): Likewise.
18646 * targhooks.c (default_builtin_vectorization_cost): Likewise.
18647 * tree-cfg.c (verify_gimple_comparison): Likewise.
18648 (verify_gimple_assign_binary): Likewise.
18649 (verify_gimple_assign_ternary): Likewise.
18650 (verify_gimple_assign_single): Likewise.
18651 * tree-pretty-print.c (dump_generic_node): Likewise.
18652 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
18653 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
18654 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
18655 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
18656 (vect_shift_permute_load_chain): Likewise.
18657 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
18658 (expand_vector_condition, optimize_vector_constructor): Likewise.
18659 (lower_vec_perm, get_compute_type): Likewise.
18660 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
18661 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
18662 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
18663 (vect_recog_mask_conversion_pattern): Likewise.
18664 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
18665 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
18666 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
18667 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
18668 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
18669 (vectorizable_shift, vectorizable_operation, vectorizable_store)
18670 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
18671 (supportable_widening_operation): Likewise.
18672 (supportable_narrowing_operation): Likewise.
18673 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
18675 * varasm.c (output_constant): Likewise.
18677 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18678 Alan Hayward <alan.hayward@arm.com>
18679 David Sherwood <david.sherwood@arm.com>
18681 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
18682 so that both the length == 3 and length != 3 cases set up their
18683 own permute vectors. Add comments explaining why we know the
18684 number of elements is constant.
18685 (vect_permute_load_chain): Likewise.
18687 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18688 Alan Hayward <alan.hayward@arm.com>
18689 David Sherwood <david.sherwood@arm.com>
18691 * machmode.h (mode_nunits): Change from unsigned char to
18693 (ONLY_FIXED_SIZE_MODES): New macro.
18694 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
18695 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
18696 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
18698 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
18699 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
18700 or if measurement_type is not polynomial.
18701 * genmodes.c (ZERO_COEFFS): New macro.
18702 (emit_mode_nunits_inline): Make mode_nunits_inline return a
18704 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
18705 Use ZERO_COEFFS when emitting initializers.
18706 * data-streamer.h (bp_pack_poly_value): New function.
18707 (bp_unpack_poly_value): Likewise.
18708 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
18709 for GET_MODE_NUNITS.
18710 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
18711 for GET_MODE_NUNITS.
18712 * tree.c (make_vector_type): Remove temporary shim and make
18713 the real function take the number of units as a poly_uint64
18714 rather than an int.
18715 (build_vector_type_for_mode): Handle polynomial nunits.
18716 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
18717 * emit-rtl.c (const_vec_series_p_1): Likewise.
18718 (gen_rtx_CONST_VECTOR): Likewise.
18719 * fold-const.c (test_vec_duplicate_folding): Likewise.
18720 * genrecog.c (validate_pattern): Likewise.
18721 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
18722 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
18723 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
18724 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
18725 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
18726 * rtlanal.c (subreg_get_info): Likewise.
18727 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
18728 (vect_grouped_load_supported): Likewise.
18729 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
18730 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
18731 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18732 (simplify_const_unary_operation, simplify_binary_operation_1)
18733 (simplify_const_binary_operation, simplify_ternary_operation)
18734 (test_vector_ops_duplicate, test_vector_ops): Likewise.
18735 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
18736 instead of CONST_VECTOR_NUNITS.
18737 * varasm.c (output_constant_pool_2): Likewise.
18738 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
18739 explicit-encoded elements in the XVEC for variable-length vectors.
18741 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18743 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
18745 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18746 Alan Hayward <alan.hayward@arm.com>
18747 David Sherwood <david.sherwood@arm.com>
18749 * coretypes.h (fixed_size_mode): Declare.
18750 (fixed_size_mode_pod): New typedef.
18751 * builtins.h (target_builtins::x_apply_args_mode)
18752 (target_builtins::x_apply_result_mode): Change type to
18753 fixed_size_mode_pod.
18754 * builtins.c (apply_args_size, apply_result_size, result_vector)
18755 (expand_builtin_apply_args_1, expand_builtin_apply)
18756 (expand_builtin_return): Update accordingly.
18758 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18760 * cse.c (hash_rtx_cb): Hash only the encoded elements.
18761 * cselib.c (cselib_hash_rtx): Likewise.
18762 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
18763 CONST_VECTOR encoding.
18765 2018-01-03 Jakub Jelinek <jakub@redhat.com>
18766 Jeff Law <law@redhat.com>
18769 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
18770 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
18771 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
18772 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
18775 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
18776 explicitly probe *sp in a noreturn function if there were any callee
18777 register saves or frame pointer is needed.
18779 2018-01-03 Jakub Jelinek <jakub@redhat.com>
18782 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
18783 BLKmode for ternary, binary or unary expressions.
18786 * var-tracking.c (delete_vta_debug_insn): New inline function.
18787 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
18788 insns from get_insns () to NULL instead of each bb separately.
18789 Use delete_vta_debug_insn. No longer static.
18790 (vt_debug_insns_local, variable_tracking_main_1): Adjust
18791 delete_vta_debug_insns callers.
18792 * rtl.h (delete_vta_debug_insns): Declare.
18793 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
18794 instead of variable_tracking_main.
18796 2018-01-03 Martin Sebor <msebor@redhat.com>
18798 PR tree-optimization/83603
18799 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
18800 arguments past the endof the argument list in functions declared
18801 without a prototype.
18802 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
18803 Avoid checking when arguments are null.
18805 2018-01-03 Martin Sebor <msebor@redhat.com>
18808 * doc/extend.texi (attribute const): Fix a typo.
18809 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
18810 issuing -Wsuggest-attribute for void functions.
18812 2018-01-03 Martin Sebor <msebor@redhat.com>
18814 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
18815 offset_int::from instead of wide_int::to_shwi.
18816 (maybe_diag_overlap): Remove assertion.
18817 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
18818 * gimple-ssa-sprintf.c (format_directive): Same.
18819 (parse_directive): Same.
18820 (sprintf_dom_walker::compute_format_length): Same.
18821 (try_substitute_return_value): Same.
18823 2018-01-03 Jeff Law <law@redhat.com>
18825 PR middle-end/83654
18826 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
18827 non-constant residual for zero at runtime and avoid probing in
18828 that case. Reorganize code for trailing problem to mirror handling
18831 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18833 PR tree-optimization/83501
18834 * tree-ssa-strlen.c (get_string_cst): New.
18835 (handle_char_store): Call get_string_cst.
18837 2018-01-03 Martin Liska <mliska@suse.cz>
18839 PR tree-optimization/83593
18840 * tree-ssa-strlen.c: Include tree-cfg.h.
18841 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
18842 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
18843 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
18845 (strlen_dom_walker::before_dom_children): Call
18846 gimple_purge_dead_eh_edges. Dump tranformation with details
18848 (strlen_dom_walker::before_dom_children): Update call by adding
18849 new argument cleanup_eh.
18850 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
18852 2018-01-03 Martin Liska <mliska@suse.cz>
18855 * cif-code.def (VARIADIC_THUNK): New enum value.
18856 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
18859 2018-01-03 Jan Beulich <jbeulich@suse.com>
18861 * sse.md (mov<mode>_internal): Tighten condition for when to use
18862 vmovdqu<ssescalarsize> for TI and OI modes.
18864 2018-01-03 Jakub Jelinek <jakub@redhat.com>
18866 Update copyright years.
18868 2018-01-03 Martin Liska <mliska@suse.cz>
18871 * ipa-visibility.c (function_and_variable_visibility): Skip
18872 functions with noipa attribure.
18874 2018-01-03 Jakub Jelinek <jakub@redhat.com>
18876 * gcc.c (process_command): Update copyright notice dates.
18877 * gcov-dump.c (print_version): Ditto.
18878 * gcov.c (print_version): Ditto.
18879 * gcov-tool.c (print_version): Ditto.
18880 * gengtype.c (create_file): Ditto.
18881 * doc/cpp.texi: Bump @copying's copyright year.
18882 * doc/cppinternals.texi: Ditto.
18883 * doc/gcc.texi: Ditto.
18884 * doc/gccint.texi: Ditto.
18885 * doc/gcov.texi: Ditto.
18886 * doc/install.texi: Ditto.
18887 * doc/invoke.texi: Ditto.
18889 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18891 * vector-builder.h (vector_builder::m_full_nelts): Change from
18892 unsigned int to poly_uint64.
18893 (vector_builder::full_nelts): Update prototype accordingly.
18894 (vector_builder::new_vector): Likewise.
18895 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
18896 (vector_builder::operator ==): Likewise.
18897 (vector_builder::finalize): Likewise.
18898 * int-vector-builder.h (int_vector_builder::int_vector_builder):
18899 Take the number of elements as a poly_uint64 rather than an
18901 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
18902 from unsigned int to poly_uint64.
18903 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
18904 (vec_perm_indices::new_vector): Likewise.
18905 (vec_perm_indices::length): Likewise.
18906 (vec_perm_indices::nelts_per_input): Likewise.
18907 (vec_perm_indices::input_nelts): Likewise.
18908 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
18909 number of elements per input as a poly_uint64 rather than an
18910 unsigned int. Use the original encoding for variable-length
18911 vectors, rather than clamping each individual element.
18912 For the second and subsequent elements in each pattern,
18913 clamp the step and base before clamping their sum.
18914 (vec_perm_indices::series_p): Handle polynomial element counts.
18915 (vec_perm_indices::all_in_range_p): Likewise.
18916 (vec_perm_indices_to_tree): Likewise.
18917 (vec_perm_indices_to_rtx): Likewise.
18918 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18919 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
18920 (tree_vector_builder::new_binary_operation): Handle polynomial
18921 element counts. Return false if we need to know the number
18922 of elements at compile time.
18923 * fold-const.c (fold_vec_perm): Punt if the number of elements
18924 isn't known at compile time.
18926 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18928 * vec-perm-indices.h (vec_perm_builder): Change element type
18929 from HOST_WIDE_INT to poly_int64.
18930 (vec_perm_indices::element_type): Update accordingly.
18931 (vec_perm_indices::clamp): Handle polynomial element_types.
18932 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
18933 (vec_perm_indices::all_in_range_p): Likewise.
18934 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
18936 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
18937 polynomial vec_perm_indices element types.
18938 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
18939 * fold-const.c (fold_vec_perm): Likewise.
18940 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
18941 * tree-vect-generic.c (lower_vec_perm): Likewise.
18942 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18943 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
18944 element type to HOST_WIDE_INT.
18946 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18947 Alan Hayward <alan.hayward@arm.com>
18948 David Sherwood <david.sherwood@arm.com>
18950 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
18951 rather than an int. Use plus_constant.
18952 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
18953 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
18955 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18956 Alan Hayward <alan.hayward@arm.com>
18957 David Sherwood <david.sherwood@arm.com>
18959 * calls.c (emit_call_1, expand_call): Change struct_value_size from
18960 a HOST_WIDE_INT to a poly_int64.
18962 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18963 Alan Hayward <alan.hayward@arm.com>
18964 David Sherwood <david.sherwood@arm.com>
18966 * calls.c (load_register_parameters): Cope with polynomial
18967 mode sizes. Require a constant size for BLKmode parameters
18968 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
18969 forces a parameter to be padded at the lsb end in order to
18970 fill a complete number of words, require the parameter size
18971 to be ordered wrt UNITS_PER_WORD.
18973 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18974 Alan Hayward <alan.hayward@arm.com>
18975 David Sherwood <david.sherwood@arm.com>
18977 * reload1.c (spill_stack_slot_width): Change element type
18978 from unsigned int to poly_uint64_pod.
18979 (alter_reg): Treat mode sizes as polynomial.
18981 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18982 Alan Hayward <alan.hayward@arm.com>
18983 David Sherwood <david.sherwood@arm.com>
18985 * reload.c (complex_word_subreg_p): New function.
18986 (reload_inner_reg_of_subreg, push_reload): Use it.
18988 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18989 Alan Hayward <alan.hayward@arm.com>
18990 David Sherwood <david.sherwood@arm.com>
18992 * lra-constraints.c (process_alt_operands): Reject matched
18993 operands whose sizes aren't ordered.
18994 (match_reload): Refer to this check here.
18996 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
18997 Alan Hayward <alan.hayward@arm.com>
18998 David Sherwood <david.sherwood@arm.com>
19000 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
19001 that the mode size is in the set {1, 2, 4, 8, 16}.
19003 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19004 Alan Hayward <alan.hayward@arm.com>
19005 David Sherwood <david.sherwood@arm.com>
19007 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
19008 Use plus_constant instead of gen_rtx_PLUS.
19010 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19011 Alan Hayward <alan.hayward@arm.com>
19012 David Sherwood <david.sherwood@arm.com>
19014 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
19015 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
19016 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
19017 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
19018 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
19019 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
19020 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
19021 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
19022 * config/i386/i386.c (ix86_push_rounding): ...this new function.
19023 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
19025 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
19026 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
19027 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
19028 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
19029 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
19030 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
19031 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
19032 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
19033 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
19034 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
19036 * expr.c (emit_move_resolve_push): Treat the input and result
19037 of PUSH_ROUNDING as a poly_int64.
19038 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
19039 (emit_push_insn): Likewise.
19040 * lra-eliminations.c (mark_not_eliminable): Likewise.
19041 * recog.c (push_operand): Likewise.
19042 * reload1.c (elimination_effects): Likewise.
19043 * rtlanal.c (nonzero_bits1): Likewise.
19044 * calls.c (store_one_arg): Likewise. Require the padding to be
19045 known at compile time.
19047 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19048 Alan Hayward <alan.hayward@arm.com>
19049 David Sherwood <david.sherwood@arm.com>
19051 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
19052 Use plus_constant instead of gen_rtx_PLUS.
19054 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19055 Alan Hayward <alan.hayward@arm.com>
19056 David Sherwood <david.sherwood@arm.com>
19058 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
19059 rather than an int.
19061 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19062 Alan Hayward <alan.hayward@arm.com>
19063 David Sherwood <david.sherwood@arm.com>
19065 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
19066 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
19067 via stack temporaries. Treat the mode size as polynomial too.
19069 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19070 Alan Hayward <alan.hayward@arm.com>
19071 David Sherwood <david.sherwood@arm.com>
19073 * expr.c (expand_expr_real_2): When handling conversions involving
19074 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
19075 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
19076 as a poly_uint64 too.
19078 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19079 Alan Hayward <alan.hayward@arm.com>
19080 David Sherwood <david.sherwood@arm.com>
19082 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
19084 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19085 Alan Hayward <alan.hayward@arm.com>
19086 David Sherwood <david.sherwood@arm.com>
19088 * combine.c (can_change_dest_mode): Handle polynomial
19089 REGMODE_NATURAL_SIZE.
19090 * expmed.c (store_bit_field_1): Likewise.
19091 * expr.c (store_constructor): Likewise.
19092 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
19093 and polynomial REGMODE_NATURAL_SIZE.
19094 (gen_lowpart_common): Likewise.
19095 * reginfo.c (record_subregs_of_mode): Likewise.
19096 * rtlanal.c (read_modify_subreg_p): Likewise.
19098 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19099 Alan Hayward <alan.hayward@arm.com>
19100 David Sherwood <david.sherwood@arm.com>
19102 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
19103 numbers of elements.
19105 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19106 Alan Hayward <alan.hayward@arm.com>
19107 David Sherwood <david.sherwood@arm.com>
19109 * match.pd: Cope with polynomial numbers of vector elements.
19111 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19112 Alan Hayward <alan.hayward@arm.com>
19113 David Sherwood <david.sherwood@arm.com>
19115 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
19116 in a POINTER_PLUS_EXPR.
19118 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19119 Alan Hayward <alan.hayward@arm.com>
19120 David Sherwood <david.sherwood@arm.com>
19122 * omp-simd-clone.c (simd_clone_subparts): New function.
19123 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
19124 (ipa_simd_modify_function_body): Likewise.
19126 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19127 Alan Hayward <alan.hayward@arm.com>
19128 David Sherwood <david.sherwood@arm.com>
19130 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
19131 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
19132 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
19133 (expand_vector_condition, vector_element): Likewise.
19134 (subparts_gt): New function.
19135 (get_compute_type): Use subparts_gt.
19136 (count_type_subparts): Delete.
19137 (expand_vector_operations_1): Use subparts_gt instead of
19138 count_type_subparts.
19140 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19141 Alan Hayward <alan.hayward@arm.com>
19142 David Sherwood <david.sherwood@arm.com>
19144 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
19145 (vect_compile_time_alias): ...this new function. Do the calculation
19146 on poly_ints rather than trees.
19147 (vect_prune_runtime_alias_test_list): Update call accordingly.
19149 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19150 Alan Hayward <alan.hayward@arm.com>
19151 David Sherwood <david.sherwood@arm.com>
19153 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
19155 (vect_schedule_slp_instance): Likewise.
19157 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19158 Alan Hayward <alan.hayward@arm.com>
19159 David Sherwood <david.sherwood@arm.com>
19161 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
19162 constant and extern definitions for variable-length vectors.
19163 (vect_get_constant_vectors): Note that the number of units
19164 is known to be constant.
19166 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19167 Alan Hayward <alan.hayward@arm.com>
19168 David Sherwood <david.sherwood@arm.com>
19170 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
19171 of units as polynomial. Choose between WIDE and NARROW based
19174 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19175 Alan Hayward <alan.hayward@arm.com>
19176 David Sherwood <david.sherwood@arm.com>
19178 * tree-vect-stmts.c (simd_clone_subparts): New function.
19179 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
19181 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19182 Alan Hayward <alan.hayward@arm.com>
19183 David Sherwood <david.sherwood@arm.com>
19185 * tree-vect-stmts.c (vectorizable_call): Treat the number of
19186 vectors as polynomial. Use build_index_vector for
19187 IFN_GOMP_SIMD_LANE.
19189 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19190 Alan Hayward <alan.hayward@arm.com>
19191 David Sherwood <david.sherwood@arm.com>
19193 * tree-vect-stmts.c (get_load_store_type): Treat the number of
19194 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
19195 for variable-length vectors.
19196 (vectorizable_mask_load_store): Treat the number of units as
19197 polynomial, asserting that it is constant if the condition has
19198 already been enforced.
19199 (vectorizable_store, vectorizable_load): Likewise.
19201 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19202 Alan Hayward <alan.hayward@arm.com>
19203 David Sherwood <david.sherwood@arm.com>
19205 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
19206 of units as polynomial. Punt if we can't tell at compile time
19207 which vector contains the final result.
19209 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19210 Alan Hayward <alan.hayward@arm.com>
19211 David Sherwood <david.sherwood@arm.com>
19213 * tree-vect-loop.c (vectorizable_induction): Treat the number
19214 of units as polynomial. Punt on SLP inductions. Use an integer
19215 VEC_SERIES_EXPR for variable-length integer reductions. Use a
19216 cast of such a series for variable-length floating-point
19219 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19220 Alan Hayward <alan.hayward@arm.com>
19221 David Sherwood <david.sherwood@arm.com>
19223 * tree.h (build_index_vector): Declare.
19224 * tree.c (build_index_vector): New function.
19225 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
19226 of units as polynomial, forcibly converting it to a constant if
19227 vectorizable_reduction has already enforced the condition.
19228 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
19229 to create a {1,2,3,...} vector.
19230 (vectorizable_reduction): Treat the number of units as polynomial.
19231 Choose vectype_in based on the largest scalar element size rather
19232 than the smallest number of units. Enforce the restrictions
19235 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19236 Alan Hayward <alan.hayward@arm.com>
19237 David Sherwood <david.sherwood@arm.com>
19239 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
19240 number of units as polynomial.
19242 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19243 Alan Hayward <alan.hayward@arm.com>
19244 David Sherwood <david.sherwood@arm.com>
19246 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
19247 * target.def (autovectorize_vector_sizes): Return the vector sizes
19248 by pointer, using vector_sizes rather than a bitmask.
19249 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
19250 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
19251 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
19253 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
19254 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
19255 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
19256 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
19257 * omp-general.c (omp_max_vf): Likewise.
19258 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19259 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
19260 * tree-vect-loop.c (vect_analyze_loop): Likewise.
19261 * tree-vect-slp.c (vect_slp_bb): Likewise.
19262 * doc/tm.texi: Regenerate.
19263 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
19265 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
19266 the vector size as a poly_uint64 rather than an unsigned int.
19267 (current_vector_size): Change from an unsigned int to a poly_uint64.
19268 (get_vectype_for_scalar_type): Update accordingly.
19269 * tree.h (build_truth_vector_type): Take the size and number of
19270 units as a poly_uint64 rather than an unsigned int.
19271 (build_vector_type): Add a temporary overload that takes
19272 the number of units as a poly_uint64 rather than an unsigned int.
19273 * tree.c (make_vector_type): Likewise.
19274 (build_truth_vector_type): Take the number of units as a poly_uint64
19275 rather than an unsigned int.
19277 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19278 Alan Hayward <alan.hayward@arm.com>
19279 David Sherwood <david.sherwood@arm.com>
19281 * target.def (get_mask_mode): Take the number of units and length
19282 as poly_uint64s rather than unsigned ints.
19283 * targhooks.h (default_get_mask_mode): Update accordingly.
19284 * targhooks.c (default_get_mask_mode): Likewise.
19285 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19286 * doc/tm.texi: Regenerate.
19288 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19289 Alan Hayward <alan.hayward@arm.com>
19290 David Sherwood <david.sherwood@arm.com>
19292 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
19293 * omp-general.c (omp_max_vf): Likewise.
19294 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
19295 (expand_omp_simd): Handle polynomial safelen.
19296 * omp-low.c (omplow_simd_context): Add a default constructor.
19297 (omplow_simd_context::max_vf): Change from int to poly_uint64.
19298 (lower_rec_simd_input_clauses): Update accordingly.
19299 (lower_rec_input_clauses): Likewise.
19301 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19302 Alan Hayward <alan.hayward@arm.com>
19303 David Sherwood <david.sherwood@arm.com>
19305 * tree-vectorizer.h (vect_nunits_for_cost): New function.
19306 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
19307 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
19308 (vect_analyze_slp_cost): Likewise.
19309 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
19310 (vect_model_load_cost): Likewise.
19312 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19313 Alan Hayward <alan.hayward@arm.com>
19314 David Sherwood <david.sherwood@arm.com>
19316 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
19317 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
19318 from an unsigned int * to a poly_uint64_pod *.
19319 (calculate_unrolling_factor): New function.
19320 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
19322 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19323 Alan Hayward <alan.hayward@arm.com>
19324 David Sherwood <david.sherwood@arm.com>
19326 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
19327 from an unsigned int to a poly_uint64.
19328 (_loop_vec_info::slp_unrolling_factor): Likewise.
19329 (_loop_vec_info::vectorization_factor): Change from an int
19331 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
19332 (vect_get_num_vectors): New function.
19333 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
19334 (vect_get_num_copies): Use vect_get_num_vectors.
19335 (vect_analyze_data_ref_dependences): Change max_vf from an int *
19336 to an unsigned int *.
19337 (vect_analyze_data_refs): Change min_vf from an int * to a
19339 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
19340 than an unsigned HOST_WIDE_INT.
19341 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
19342 (vect_analyze_data_ref_dependence): Change max_vf from an int *
19343 to an unsigned int *.
19344 (vect_analyze_data_ref_dependences): Likewise.
19345 (vect_compute_data_ref_alignment): Handle polynomial vf.
19346 (vect_enhance_data_refs_alignment): Likewise.
19347 (vect_prune_runtime_alias_test_list): Likewise.
19348 (vect_shift_permute_load_chain): Likewise.
19349 (vect_supportable_dr_alignment): Likewise.
19350 (dependence_distance_ge_vf): Take the vectorization factor as a
19351 poly_uint64 rather than an unsigned HOST_WIDE_INT.
19352 (vect_analyze_data_refs): Change min_vf from an int * to a
19354 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
19355 vfm1 as a poly_uint64 rather than an int. Make the same change
19356 for the returned bound_scalar.
19357 (vect_gen_vector_loop_niters): Handle polynomial vf.
19358 (vect_do_peeling): Likewise. Update call to
19359 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
19360 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
19362 * tree-vect-loop.c (vect_determine_vectorization_factor)
19363 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
19364 (vect_get_known_peeling_cost): Likewise.
19365 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
19366 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
19367 (vect_transform_loop): Likewise. Use the lowest possible VF when
19368 updating the upper bounds of the loop.
19369 (vect_min_worthwhile_factor): Make static. Return an unsigned int
19370 rather than an int.
19371 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
19372 polynomial unroll factors.
19373 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
19374 (vect_make_slp_decision): Likewise.
19375 (vect_supported_load_permutation_p): Likewise, and polynomial
19377 (vect_analyze_slp_cost): Handle polynomial vf.
19378 (vect_slp_analyze_node_operations): Likewise.
19379 (vect_slp_analyze_bb_1): Likewise.
19380 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
19381 than an unsigned HOST_WIDE_INT.
19382 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
19383 (vectorizable_load): Handle polynomial vf.
19384 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
19386 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
19388 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19389 Alan Hayward <alan.hayward@arm.com>
19390 David Sherwood <david.sherwood@arm.com>
19392 * match.pd: Handle bit operations involving three constants
19393 and try to fold one pair.
19395 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19397 * tree-vect-loop-manip.c: Include gimple-fold.h.
19398 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
19399 niters_maybe_zero parameters. Handle other cases besides a step of 1.
19400 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
19401 Add a path that uses a step of VF instead of 1, but disable it
19403 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
19404 and niters_no_overflow parameters. Update calls to
19405 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
19406 Create a new SSA name if the latter choses to use a ste other
19407 than zero, and return it via niters_vector_mult_vf_var.
19408 * tree-vect-loop.c (vect_transform_loop): Update calls to
19409 vect_do_peeling, vect_gen_vector_loop_niters and
19410 slpeel_make_loop_iterate_ntimes.
19411 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
19412 (vect_gen_vector_loop_niters): Update declarations after above changes.
19414 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
19416 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
19417 128-bit round to integer instructions.
19418 (ceil<mode>2): Likewise.
19419 (btrunc<mode>2): Likewise.
19420 (round<mode>2): Likewise.
19422 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19424 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
19425 unaligned VSX load/store on P8/P9.
19426 (expand_block_clear): Allow the use of unaligned VSX
19427 load/store on P8/P9.
19429 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19431 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
19433 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
19434 swap associated with both a load and a store.
19436 2018-01-02 Andrew Waterman <andrew@sifive.com>
19438 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
19439 * config/riscv/riscv.md (clear_cache): Use it.
19441 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
19443 * web.c: Remove out-of-date comment.
19445 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19447 * expr.c (fixup_args_size_notes): Check that any existing
19448 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
19449 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
19450 (emit_single_push_insn): ...here.
19452 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19454 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
19455 (const_vector_encoded_nelts): New function.
19456 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
19457 (const_vector_int_elt, const_vector_elt): Declare.
19458 * emit-rtl.c (const_vector_int_elt_1): New function.
19459 (const_vector_elt): Likewise.
19460 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
19461 of CONST_VECTOR_ELT.
19463 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19465 * expr.c: Include rtx-vector-builder.h.
19466 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
19467 directly on the tree encoding.
19468 (const_vector_from_tree): Likewise.
19469 * optabs.c: Include rtx-vector-builder.h.
19470 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
19471 sequence of "u" values.
19472 * vec-perm-indices.c: Include rtx-vector-builder.h.
19473 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
19474 directly on the vec_perm_indices encoding.
19476 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19478 * doc/rtl.texi (const_vector): Describe new encoding scheme.
19479 * Makefile.in (OBJS): Add rtx-vector-builder.o.
19480 * rtx-vector-builder.h: New file.
19481 * rtx-vector-builder.c: Likewise.
19482 * rtl.h (rtx_def::u2): Add a const_vector field.
19483 (CONST_VECTOR_NPATTERNS): New macro.
19484 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
19485 (CONST_VECTOR_DUPLICATE_P): Likewise.
19486 (CONST_VECTOR_STEPPED_P): Likewise.
19487 (CONST_VECTOR_ENCODED_ELT): Likewise.
19488 (const_vec_duplicate_p): Check for a duplicated vector encoding.
19489 (unwrap_const_vec_duplicate): Likewise.
19490 (const_vec_series_p): Check for a non-duplicated vector encoding.
19491 Say that the function only returns true for integer vectors.
19492 * emit-rtl.c: Include rtx-vector-builder.h.
19493 (gen_const_vec_duplicate_1): Delete.
19494 (gen_const_vector): Call gen_const_vec_duplicate instead of
19495 gen_const_vec_duplicate_1.
19496 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
19497 (gen_const_vec_duplicate): Use rtx_vector_builder.
19498 (gen_const_vec_series): Likewise.
19499 (gen_rtx_CONST_VECTOR): Likewise.
19500 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
19501 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
19502 Build a new vector rather than modifying a CONST_VECTOR in-place.
19503 (handle_special_swappables): Update call accordingly.
19504 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
19505 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
19506 Build a new vector rather than modifying a CONST_VECTOR in-place.
19507 (handle_special_swappables): Update call accordingly.
19509 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19511 * simplify-rtx.c (simplify_const_binary_operation): Use
19512 CONST_VECTOR_ELT instead of XVECEXP.
19514 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19516 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
19517 the selector elements to be different from the data elements
19518 if the selector is a VECTOR_CST.
19519 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
19520 ssizetype for the selector.
19522 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19524 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
19525 before testing each element individually.
19526 * tree-vect-generic.c (lower_vec_perm): Likewise.
19528 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19530 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
19531 * selftest-run-tests.c (selftest::run_tests): Call it.
19532 * vector-builder.h (vector_builder::operator ==): New function.
19533 (vector_builder::operator !=): Likewise.
19534 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
19535 (vec_perm_indices::all_from_input_p): New function.
19536 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
19537 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
19538 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
19539 instead of reading the VECTOR_CST directly. Detect whether both
19540 vector inputs are the same before constructing the vec_perm_indices,
19541 and update the number of inputs argument accordingly. Use the
19542 utility functions added above. Only construct sel2 if we need to.
19544 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19546 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
19547 the broadcast of the low byte.
19548 (expand_mult_highpart): Use an explicit encoding for the permutes.
19549 * optabs-query.c (can_mult_highpart_p): Likewise.
19550 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
19551 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
19552 (vectorizable_bswap): Likewise.
19553 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
19554 explicit encoding for the power-of-2 permutes.
19555 (vect_permute_store_chain): Likewise.
19556 (vect_grouped_load_supported): Likewise.
19557 (vect_permute_load_chain): Likewise.
19559 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19561 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
19562 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
19563 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
19564 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19565 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
19566 (vect_gen_perm_mask_any): Likewise.
19568 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19570 * int-vector-builder.h: New file.
19571 * vec-perm-indices.h: Include int-vector-builder.h.
19572 (vec_perm_indices): Redefine as an int_vector_builder.
19573 (auto_vec_perm_indices): Delete.
19574 (vec_perm_builder): Redefine as a stand-alone class.
19575 (vec_perm_indices::vec_perm_indices): New function.
19576 (vec_perm_indices::clamp): Likewise.
19577 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
19578 (vec_perm_indices::new_vector): New function.
19579 (vec_perm_indices::new_expanded_vector): Update for new
19580 vec_perm_indices class.
19581 (vec_perm_indices::rotate_inputs): New function.
19582 (vec_perm_indices::all_in_range_p): Operate directly on the
19583 encoded form, without computing elided elements.
19584 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
19585 encoding. Update for new vec_perm_indices class.
19586 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
19587 the given vec_perm_builder.
19588 (expand_vec_perm_var): Update vec_perm_builder constructor.
19589 (expand_mult_highpart): Use vec_perm_builder instead of
19590 auto_vec_perm_indices.
19591 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
19592 vec_perm_indices instead of auto_vec_perm_indices. Use a single
19593 or double series encoding as appropriate.
19594 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
19595 vec_perm_indices instead of auto_vec_perm_indices.
19596 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
19597 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
19598 (vect_permute_store_chain): Likewise.
19599 (vect_grouped_load_supported): Likewise.
19600 (vect_permute_load_chain): Likewise.
19601 (vect_shift_permute_load_chain): Likewise.
19602 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
19603 (vect_transform_slp_perm_load): Likewise.
19604 (vect_schedule_slp_instance): Likewise.
19605 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
19606 (vectorizable_mask_load_store): Likewise.
19607 (vectorizable_bswap): Likewise.
19608 (vectorizable_store): Likewise.
19609 (vectorizable_load): Likewise.
19610 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
19611 vec_perm_indices instead of auto_vec_perm_indices. Use
19612 tree_to_vec_perm_builder to read the vector from a tree.
19613 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
19614 vec_perm_builder instead of a vec_perm_indices.
19615 (have_whole_vector_shift): Use vec_perm_builder and
19616 vec_perm_indices instead of auto_vec_perm_indices. Leave the
19617 truncation to calc_vec_perm_mask_for_shift.
19618 (vect_create_epilog_for_reduction): Likewise.
19619 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
19620 from auto_vec_perm_indices to vec_perm_indices.
19621 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
19622 instead of changing individual elements.
19623 (aarch64_vectorize_vec_perm_const): Use new_vector to install
19624 the vector in d.perm.
19625 * config/arm/arm.c (expand_vec_perm_d::perm): Change
19626 from auto_vec_perm_indices to vec_perm_indices.
19627 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
19628 instead of changing individual elements.
19629 (arm_vectorize_vec_perm_const): Use new_vector to install
19630 the vector in d.perm.
19631 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
19632 Update vec_perm_builder constructor.
19633 (rs6000_expand_interleave): Likewise.
19634 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
19635 (rs6000_expand_interleave): Likewise.
19637 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19639 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
19640 to qimode could truncate the indices.
19641 * optabs.c (expand_vec_perm_var): Likewise.
19643 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19645 * Makefile.in (OBJS): Add vec-perm-indices.o.
19646 * vec-perm-indices.h: New file.
19647 * vec-perm-indices.c: Likewise.
19648 * target.h (vec_perm_indices): Replace with a forward class
19650 (auto_vec_perm_indices): Move to vec-perm-indices.h.
19651 * optabs.h: Include vec-perm-indices.h.
19652 (expand_vec_perm): Delete.
19653 (selector_fits_mode_p, expand_vec_perm_var): Declare.
19654 (expand_vec_perm_const): Declare.
19655 * target.def (vec_perm_const_ok): Replace with...
19656 (vec_perm_const): ...this new hook.
19657 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
19658 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
19659 * doc/tm.texi: Regenerate.
19660 * optabs.def (vec_perm_const): Delete.
19661 * doc/md.texi (vec_perm_const): Likewise.
19662 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
19663 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
19664 expand_vec_perm for constant permutation vectors. Assert that
19665 the mode of variable permutation vectors is the integer equivalent
19666 of the mode that is being permuted.
19667 * optabs-query.h (selector_fits_mode_p): Declare.
19668 * optabs-query.c: Include vec-perm-indices.h.
19669 (selector_fits_mode_p): New function.
19670 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
19671 is defined, instead of checking whether the vec_perm_const_optab
19672 exists. Use targetm.vectorize.vec_perm_const instead of
19673 targetm.vectorize.vec_perm_const_ok. Check whether the indices
19674 fit in the vector mode before using a variable permute.
19675 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
19676 vec_perm_indices instead of an rtx.
19677 (expand_vec_perm): Replace with...
19678 (expand_vec_perm_const): ...this new function. Take the selector
19679 as a vec_perm_indices rather than an rtx. Also take the mode of
19680 the selector. Update call to shift_amt_for_vec_perm_mask.
19681 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
19682 Use vec_perm_indices::new_expanded_vector to expand the original
19683 selector into bytes. Check whether the indices fit in the vector
19684 mode before using a variable permute.
19685 (expand_vec_perm_var): Make global.
19686 (expand_mult_highpart): Use expand_vec_perm_const.
19687 * fold-const.c: Includes vec-perm-indices.h.
19688 * tree-ssa-forwprop.c: Likewise.
19689 * tree-vect-data-refs.c: Likewise.
19690 * tree-vect-generic.c: Likewise.
19691 * tree-vect-loop.c: Likewise.
19692 * tree-vect-slp.c: Likewise.
19693 * tree-vect-stmts.c: Likewise.
19694 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
19696 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
19697 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
19698 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
19699 (aarch64_vectorize_vec_perm_const): ...this new function.
19700 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
19701 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
19702 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
19703 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
19704 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
19705 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
19706 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
19708 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
19709 check for NEON modes.
19710 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
19711 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
19712 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
19713 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
19715 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
19716 the old VEC_PERM_CONST conditions.
19717 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
19718 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
19719 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
19720 (ia64_vectorize_vec_perm_const_ok): Merge into...
19721 (ia64_vectorize_vec_perm_const): ...this new function.
19722 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
19723 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
19724 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
19725 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
19726 * config/mips/mips.c (mips_expand_vec_perm_const)
19727 (mips_vectorize_vec_perm_const_ok): Merge into...
19728 (mips_vectorize_vec_perm_const): ...this new function.
19729 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
19730 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
19731 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
19732 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
19733 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
19734 (rs6000_expand_vec_perm_const): Delete.
19735 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
19737 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
19738 (altivec_expand_vec_perm_const_le): Take each operand individually.
19739 Operate on constant selectors rather than rtxes.
19740 (altivec_expand_vec_perm_const): Likewise. Update call to
19741 altivec_expand_vec_perm_const_le.
19742 (rs6000_expand_vec_perm_const): Delete.
19743 (rs6000_vectorize_vec_perm_const_ok): Delete.
19744 (rs6000_vectorize_vec_perm_const): New function.
19745 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
19746 an element count and rtx array.
19747 (rs6000_expand_extract_even): Update call accordingly.
19748 (rs6000_expand_interleave): Likewise.
19749 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
19750 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
19751 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
19752 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
19753 (rs6000_expand_vec_perm_const): Delete.
19754 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
19755 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
19756 (altivec_expand_vec_perm_const_le): Take each operand individually.
19757 Operate on constant selectors rather than rtxes.
19758 (altivec_expand_vec_perm_const): Likewise. Update call to
19759 altivec_expand_vec_perm_const_le.
19760 (rs6000_expand_vec_perm_const): Delete.
19761 (rs6000_vectorize_vec_perm_const_ok): Delete.
19762 (rs6000_vectorize_vec_perm_const): New function. Remove stray
19763 reference to the SPE evmerge intructions.
19764 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
19765 an element count and rtx array.
19766 (rs6000_expand_extract_even): Update call accordingly.
19767 (rs6000_expand_interleave): Likewise.
19768 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
19769 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
19771 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
19773 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19775 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
19776 vector mode and that that mode matches the mode of the data
19778 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
19779 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
19780 directly using expand_vec_perm_1 when forcing selectors into
19782 (expand_vec_perm_var): New function, split out from expand_vec_perm.
19784 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19786 * optabs-query.h (can_vec_perm_p): Delete.
19787 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
19788 * optabs-query.c (can_vec_perm_p): Split into...
19789 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
19790 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
19791 particular selector is valid.
19792 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
19793 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
19794 (vect_grouped_load_supported): Likewise.
19795 (vect_shift_permute_load_chain): Likewise.
19796 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
19797 (vect_transform_slp_perm_load): Likewise.
19798 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
19799 (vectorizable_bswap): Likewise.
19800 (vect_gen_perm_mask_checked): Likewise.
19801 * fold-const.c (fold_ternary_loc): Likewise. Don't take
19802 implementations of variable permutation vectors into account
19803 when deciding which selector to use.
19804 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
19805 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
19806 with a false third argument.
19807 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
19808 to test whether the constant selector is valid and can_vec_perm_var_p
19809 to test whether a variable selector is valid.
19811 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19813 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
19814 * optabs-query.c (can_vec_perm_p): Likewise.
19815 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
19816 instead of vec_perm_indices.
19817 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
19818 (vect_gen_perm_mask_checked): Likewise,
19819 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
19820 (vect_gen_perm_mask_checked): Likewise,
19822 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
19824 * optabs-query.h (qimode_for_vec_perm): Declare.
19825 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
19826 (qimode_for_vec_perm): ...this new function.
19827 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
19829 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19831 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
19832 does not have a conditional at the top.
19834 2018-01-02 Richard Biener <rguenther@suse.de>
19836 * ipa-inline.c (big_speedup_p): Fix expression.
19838 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
19841 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
19844 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
19848 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
19849 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
19850 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
19851 cond_taken_branch_cost 3->4.
19853 2018-01-01 Jakub Jelinek <jakub@redhat.com>
19855 PR tree-optimization/83581
19856 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
19857 TODO_cleanup_cfg if any changes have been made.
19859 PR middle-end/83608
19860 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
19861 convert_modes if target mode has the right side, but different mode
19864 PR middle-end/83609
19865 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
19866 last argument when extracting from CONCAT. If either from_real or
19867 from_imag is NULL, use expansion through memory. If result is not
19868 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
19869 the parts directly to inner mode, if even that fails, use expansion
19872 PR middle-end/83623
19873 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
19874 check for bswap in mode rather than HImode and use that in expand_unop
19877 Copyright (C) 2018 Free Software Foundation, Inc.
19879 Copying and distribution of this file, with or without modification,
19880 are permitted in any medium without royalty provided the copyright
19881 notice and this notice are preserved.