1 2018-11-27 Alan Modra <amodra@gmail.com>
3 * config.gcc (powerpc*-*-linux*): Add linux.h to tm_file.
4 * config/rs6000/linux.h (TARGET_OS_CPP_BUILTINS): Use
5 GNU_USER_TARGET_OS_CPP_BUILTINS.
6 (RS6000_ABI_NAME): Define.
7 * config/rs6000/linux64.h (TARGET_OS_CPP_BUILTINS): Use
8 GNU_USER_TARGET_OS_CPP_BUILTINS.
9 (MUSL_DYNAMIC_LINKER32): Undef before defining.
10 (UCLIBC_DYNAMIC_LINKER32, UCLIBC_DYNAMIC_LINKER64): Don't define.
11 (CHOOSE_DYNAMIC_LINKER): Don't define.
12 (GNU_USER_DYNAMIC_LINKER32, GNU_USER_DYNAMIC_LINKER64): Don't define.
13 * config/rs6000/sysv4.h (MUSL_DYNAMIC_LINKER): Undef before defining.
14 (CHOOSE_DYNAMIC_LINKER, GNU_USER_DYNAMIC_LINKER): Only define when
16 (CPP_OS_LINUX_SPEC): Remove defines and asserts handled by
17 TARGET_OS_CPP_BUILTINS.
19 2018-11-26 Jozef Lawrynowicz <jozef.l@mittosystems.com>
21 * doc/sourcebuild.texi: Document check_effective_target_newlib_nano_io.
23 2018-11-26 Jakub Jelinek <jakub@redhat.com>
26 * config/i386/i386.c (def_builtin2): If tcode == VOID_FTYPE_UINT64
27 and !TARGET_64BIT, return NULL_TREE.
30 * dwarf2out.c (secname_for_decl): For functions with
31 DECL_SECTION_NAME if in_cold_section_p, try to return
32 current_function_section's name if it is a named section.
34 2018-11-26 Maya Rashish <coypu@sdf.org>
37 * config.host (*-*-netbsd*): Use x-netbsd and host-netbsd.o.
38 * config/x-netbsd: New file.
39 * config/host-netbsd: New file.
41 2018-11-26 Jozef Lawrynowicz <jozef.l@mittosystems.com>
43 * doc/sourcebuild.texi: Document check_effective_target_size20plus.
44 Clarify documentation for check_effective_target_size32plus.
46 2018-11-26 Aaron Sawdey <acsawdey@linux.ibm.com>
48 * config/rs6000/rs6000-string.c (expand_cmp_vec_sequence): Rename
49 and modify expand_strncmp_vec_sequence.
50 (emit_final_compare_vec): Rename and modify emit_final_str_compare_vec.
51 (generate_6432_conversion): New function.
52 (expand_block_compare): Add support for vsx.
53 (expand_block_compare_gpr): New function.
54 * config/rs6000/rs6000.opt (rs6000_block_compare_inline_limit): Increase
55 default limit to 63 because of more compact vsx code.
57 2018-11-26 Uros Bizjak <ubizjak@gmail.com>
60 * config/i386/i386.c (dbx_register_map): Use IGNORED_DWARF_REGNUM
61 for registers for which no debug information can be generated.
62 Use INVALID_REGNUM for non-existent registers.
63 (dbx64_register_map): Ditto.
64 (svr4_dbx_register_map): Ditto.
66 2018-11-26 Sandra Loosemore <sandra@codesourcery.com>
69 * doc/invoke.texi (Optimize Options): Document that
70 -freorder-functions also uses "hot" and "cold" attributes.
72 2018-11-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
74 * config/aarch64/aarch64.c (aarch64_mangle_type): Fix typo in comment.
76 2018-11-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
78 * config/arm/types.md (mul64): Rename to...
79 (widen_mul64): ... This.
80 * config/arm/arm-generic.md: Rename mul64 to widen_mul64.
81 * config/arm/cortex-a15.md: Likewise.
82 * config/arm/cortex-a5.md: Likewise.
83 * config/arm/cortex-a53.md: Likewise.
84 * config/arm/cortex-a57.md: Likewise.
85 * config/arm/cortex-a7.md: Likewise.
86 * config/arm/cortex-m4.md: Likewise.
87 * config/arm/exynos-m1.md: Likewise.
88 * config/arm/marvell-pj4.md: Likewise.
89 * config/arm/xgene1.md: Likewise.
91 2018-11-26 Richard Biener <rguenther@suse.de>
93 PR tree-optimization/88182
94 * tree-vect-loop.c (vectorizable_reduction): Pick up single
95 correct reduc_def_info.
96 * tree-vect-slp.c (vect_analyze_slp_instance): Set
97 STMT_VINFO_REDUC_DEF of the first stmt.
99 2018-11-26 Andreas Krebbel <krebbel@linux.ibm.com>
101 * doc/invoke.texi: Document z14/arch12 -march option.
103 2018-11-25 Sandra Loosemore <sandra@codesourcery.com>
106 * doc/invoke.texi (Option Summary): Add -Wmissing-noreturn.
107 (Warning Options): Likewise.
109 2018-11-25 Sandra Loosemore <sandra@codesourcery.com>
112 * doc/extend.texi (Common Function Attributes): Clarify that
113 functions with "const" attribute can read const global variables.
115 2018-11-26 Alan Modra <amodra@gmail.com>
117 * config/rs6000/rs6000.h (OBJECT_PEF, TARGET_MACOS): Delete.
119 2018-11-25 Sandra Loosemore <sandra@codesourcery.com>
122 * doc/extend.texi (Common Variable Attributes): Use preferred
123 placement of type attributes in examples, plus whitespace fixes.
124 (Type Attributes): Clarify why placement of attributes
125 immediately after struct/union/enum keyword is preferred.
126 (Common Type Attributes): Use preferred placement of type
127 attributes in examples, plus more whitespace fixes.
129 2018-11-25 Paul Koning <ni1d@arrl.net>
131 * config/pdp11/pdp11.h (TARGET_HAS_NO_HW_DIVIDE): Define.
133 2018-11-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
135 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
136 attributes): Document d_runtime.
138 2018-11-25 Vladimir Makarov <vmakarov@redhat.com>
141 * ira-costs.c (record_operand_costs): Initiate register move cost
144 2018-11-23 Jeff Law <law@redhat.com>
146 PR rtl-optimization/87468
147 * tree-ssa-threadupdate.c (create_block_for_threading): Clear
148 EDGE_IGNORE on all outgoing edges of the duplicate block.
150 2018-11-23 Vladimir Makarov <vmakarov@redhat.com>
153 * ira-costs.c (record_operand_costs): Use bigger hard reg class if
154 its mode does not fit to the original class.
156 2018-11-23 Martin Sebor <msebor@redhat.com>
158 PR tree-optimization/87756
159 * expr.c (string_constant): Synthesize a string literal from
160 the address of a constant character.
161 * tree.c (build_string_literal): Add an argument.
162 * tree.h (build_string_literal): Same.
164 2018-11-23 Christoph Muellner <christoph.muellner@theobroma-systems.com>
166 * config/aarch64/aarch64-cores.def: Define emag.
167 * config/aarch64/aarch64-tune.md: Regenerated with emag.
168 * config/aarch64/aarch64.c (emag_tunings): New struct.
169 * doc/invoke.texi: Document mtune value.
171 2018-11-23 Christoph Muellner <christoph.muellner@theobroma-systems.com>
172 Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
174 * config/aarch64/aarch64.c (xgene1_tunings): Optimize Xgene1 tunings
176 * config/aarch64/aarch64.c (xgene1_tunings): Add Xgene1 specific
178 * config/aarch64/aarch64.c (xgene1_addrcost_table): Correct the post-modify
180 * config/arm/aarch-cost-tables.h (xgene1_extra_costs): Update the cost table
183 2018-11-23 Richard Biener <rguenther@suse.de>
185 PR tree-optimization/88149
186 * tree-vect-slp.c (vect_slp_analyze_node_operations): Detect
187 the case where there are two different def types for the
188 same operand at different operand position in the same stmt.
190 2018-23-11 Mihail Ionescu <mihail.ionescu@arm.com>
192 * config/arm/arm.c (arm_expand_compare_and_swap): Simplify and call
193 gen_atomic_compare_swap_1.
194 (arm_evpc_neon_vuzp): Likewise gen_neon_vuzp_internal.
195 (arm_evpc_neon_vtrn): Likewise gen_neon_vtrn_internal.
196 (arm_evpc_neon_vext): Likewise gen_neon_vext_internal.
197 (arm_evpc_neon_vzip): Likewise gen_neon_vzip_internal.
198 (arm_evpc_neon_vrev): Replace the function pointer and simplify the mode
200 * config/arm/arm.md (neon_vext<mode>),
201 (neon_vrev64<mode>, neon_vrev32<mode>),
202 (neon_vrev16<mode>, neon_vtrn<mode>_internal),
203 (neon_vzip<mode>_internal, neon_vuzp<mode>_internal): Add an '@'character
204 before the pattern name.
205 * config/arm/sync.md:
206 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1),
207 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): Likewise.
209 2018-11-23 Jakub Jelinek <jakub@redhat.com>
211 PR tree-optimization/86614
212 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return early
213 if TREE_NO_WARNING is set on ref.ref.
215 2018-11-22 Sandra Loosemore <sandra@codesourcery.com>
216 Alan Coopersmith <alan.coopersmith@oracle.com>
219 * doc/extend.texi (Designated Inits): Clarify handling of multiple
220 initializers for unions.
222 2018-11-22 Jan Hubicka <jh@suse.cz>
225 * ipa-devirt.c (type_variants_equivalent_p): Drop warn and warned
226 parameters; do not warn here.
227 (odr_subtypes_equivalent_p): Likewise.
228 (warn_odr): Fix typo.
229 (warn_types_mismatch): Do not output confused warnings on integer types.
230 (odr_types_equivalent_p): Update.
232 2018-11-22 Jakub Jelinek <jakub@redhat.com>
234 * config/i386/i386.c (ix86_option_override_internal,
235 ix86_can_inline_p, classify_argument, construct_container,
236 ix86_expand_prologue, ix86_expand_split_stack_prologue,
237 ix86_expand_carry_flag_compare, expand_set_or_movmem_via_loop,
238 expand_setmem_epilogue_via_loop, promote_duplicated_reg,
239 ix86_expand_set_or_movmem, ix86_init_builtins_va_builtins_abi):
242 * config/i386/i386.c (ix86_option_override_internal): For
243 stack_protector_guard related options, use opts_set->x_ instead
244 of global_options_set. and prefix options with opts->x_ . Move
245 defaults for offset and reg into else block.
247 2018-11-22 Uros Bizjak <ubizjak@gmail.com>
249 * config/i386/i386.c (ix86_check_avx_upper_register):
250 Return true for all SSE registers with mode bitsize > 128.
252 2018-11-22 Vladimir Makarov <vmakarov@redhat.com>
254 PR rtl-optimization/87718
255 * ira-costs.c: Remove trailing white-spaces.
256 (record_operand_costs): Add a special treatment for moves
257 involving a hard register.
259 2018-11-22 Uros Bizjak <ubizjak@gmail.com>
261 * config/i386/i386.c (ix86_avx_emit_vzeroupper): Remove.
262 (ix86_emit_mode_set) <case AVX_U128>: Emit vzeroupper here.
264 2018-11-22 Martin Liska <mliska@suse.cz>
266 * common/config/i386/i386-common.c (processor_names): Add
267 static assert and add missing "znver2".
268 (ix86_get_valid_option_values): Add checking assert for null
269 values and add "native" value if feasible.
270 * config/i386/i386.h: Do not declare size of processor_names.
271 * common/config/i386/i386-common.c:
272 * config/i386/i386.c: Add static assert for size
273 of processor_cost_table.
275 2018-11-22 Thomas Preud'homme <thomas.preudhomme@linaro.org>
277 * target-insns.def (stack_protect_combined_set): Define new standard
279 (stack_protect_combined_test): Likewise.
280 * cfgexpand.c (stack_protect_prologue): Try new
281 stack_protect_combined_set pattern first.
282 * function.c (stack_protect_epilogue): Try new
283 stack_protect_combined_test pattern first.
284 * config/arm/arm.c (require_pic_register): Add pic_reg and compute_now
285 parameters to control which register to use as PIC register and force
286 reloading PIC register respectively. Insert in the stream of insns if
288 (legitimize_pic_address): Expose above new parameters in prototype and
289 adapt recursive calls accordingly. Use pic_reg if non null instead of
291 (arm_load_pic_register): Add pic_reg parameter and use it if non null.
292 (arm_legitimize_address): Adapt to new legitimize_pic_address
294 (thumb_legitimize_address): Likewise.
295 (arm_emit_call_insn): Adapt to require_pic_register prototype change.
296 (arm_expand_prologue): Adapt to arm_load_pic_register prototype change.
297 (thumb1_expand_prologue): Likewise.
298 * config/arm/arm-protos.h (legitimize_pic_address): Adapt to prototype
300 (arm_load_pic_register): Likewise.
301 * config/arm/predicated.md (guard_addr_operand): New predicate.
302 (guard_operand): New predicate.
303 * config/arm/arm.md (movsi expander): Adapt to legitimize_pic_address
305 (builtin_setjmp_receiver expander): Adapt to thumb1_expand_prologue
307 (stack_protect_combined_set): New expander..
308 (stack_protect_combined_set_insn): New insn_and_split pattern.
309 (stack_protect_set_insn): New insn pattern.
310 (stack_protect_combined_test): New expander.
311 (stack_protect_combined_test_insn): New insn_and_split pattern.
312 (arm_stack_protect_test_insn): New insn pattern.
313 * config/arm/thumb1.md (thumb1_stack_protect_test_insn): New insn pattern.
314 * config/arm/unspecs.md (UNSPEC_SP_SET): New unspec.
315 (UNSPEC_SP_TEST): Likewise.
316 * doc/md.texi (stack_protect_combined_set): Document new standard
318 (stack_protect_set): Clarify that the operand for guard's address is
320 (stack_protect_combined_test): Document new standard pattern name.
321 (stack_protect_test): Clarify that the operand for guard's address is
324 2018-11-22 Richard Biener <rguenther@suse.de>
326 PR tree-optimization/88148
327 * tree-ssa-loop-niter.c (simplify_replace_tree): Get optional
328 valueization callback parameter and handle it.
329 * tree-ssa-loop-niter.h (simplify_replace_tree): Export.
330 * tree-ssa-sccvn.c (process_bb): Eliminate in loop niter trees.
332 2018-11-22 Richard Biener <rguenther@suse.de>
336 * lto-streamer-out.c (lto_is_streamable): Allow CALL_EXPRs
337 which can appear in size expressions.
338 * tree-streamer-in.c (unpack_ts_base_value_fields): Stream
339 CALL_EXPR_BY_DESCRIPTOR.
340 (streamer_read_tree_bitfields): Stream CALL_EXPR_IFN.
341 * tree-streamer-out.c (pack_ts_base_value_fields): Stream
342 CALL_EXPR_BY_DESCRIPTOR.
343 (streamer_write_tree_bitfields): Stream CALL_EXPR_IFN.
347 * tree.c (free_lang_data_in_one_sizepos): Free non-gimple-val
350 2018-11-22 Richard Biener <rguenther@suse.de>
352 PR tree-optimization/88069
353 * tree-ssa-sccvn.c (visit_phi): Tweak previous fix to not
354 apply to default defs.
356 2018-11-22 Jakub Jelinek <jakub@redhat.com>
360 * config/i386/i386.c (ix86_option_override_internal): Default
361 ix86_stack_protector_guard to SSP_TLS only if TARGET_THREAD_SSP_OFFSET
363 * config/i386/i386.md (stack_protect_set, stack_protect_set_<mode>,
364 stack_protect_test, stack_protect_test_<mode>): Use empty condition
365 instead of TARGET_SSP_TLS_GUARD.
367 2018-11-22 Martin Liska <mliska@suse.cz>
370 * config/i386/i386.c (ix86_option_override_internal): Error
371 about usage -mabi=ms and -fsanitize=thread.
373 2018-11-21 Uros Bizjak <ubizjak@gmail.com>
376 2013-10-26 Vladimir Makarov <vmakarov@redhat.com>
379 2013-10-25 Vladimir Makarov <vmakarov@redhat.com>
381 * lra-spills.c (lra_final_code_change): Remove useless move insns.
383 2018-11-21 Uros Bizjak <ubizjak@gmail.com>
386 * function.c (expand_function_end): Do not emit extra blockage insn.
388 2018-11-21 Lokesh Janghel <lokeshjanghel91@gmail.com>
391 * config/i386/i386.c (function_value_ms_64): Return AX_REG instead
392 of FIRST_SSE_REG for 4 or 8 byte modes.
394 2018-11-21 Jan Hubicka <jh@suse.cz>
397 * tree.c (fld_decl_context): Break out from ...
398 (free_lang_data_in_decl): ... here; free TREE_PUBLIC, TREE_PRIVATE
399 DECL_ARTIFICIAL of TYPE_DECL; do not free TREE_TYPE of TYPE_DECL.
400 (fld_incomplete_type_of): Build copy of TYP_DECL.
401 * ipa-devirt.c (free_enum_values): Rename to ...
402 (free_odr_warning_data): ... this one; free also duplicated TYPE_DECLs
403 and TREE_TYPEs of TYPE_DECLs.
404 (get_odr_type): Initialize odr_vtable_hash if needed.
406 2018-11-21 Alexandre Oliva <oliva@adacore.com>
408 * final.c (compute_discriminator): Declare. Renamed from...
409 (maybe_set_discriminator): ... this. Set and return a local.
410 (override_discriminator): New.
411 (final_scan_insn_1): Set it.
412 (notice_source_line): Adjust. Always set discriminator.
414 2018-11-21 Jakub Jelinek <jakub@redhat.com>
417 * config/aarch64/atomics.md (@aarch64_compare_and_swap<mode>): Use
418 rIJ constraint for aarch64_plus_operand rather than rn.
420 2018-11-21 Renlin Li <renlin.li@arm.com>
423 * explow.h (get_dynamic_stack_size): Declare it as external.
424 * explow.c (record_new_stack_level): Remove function static attribute.
425 * function.c (assign_stack_local_1): Dynamically align the stack slot
426 addr for parameter copy on the stack.
428 2018-11-21 Richard Biener <rguenther@suse.de>
431 * bitmap.c (bitmap_last_set_bit): Refactor to avoid warning.
432 * Makefile.in (bitmap.o-warn): Remove again.
434 2018-11-20 Jeff Law <law@redhat.com>
436 PR tree-optimization/88069
437 * tree-ssa-dom.c (record_equivalences_from_phis): Propagate away
438 degenerate virtual PHIs.
440 2018-11-21 H.J. Lu <hongjiu.lu@intel.com>
443 * config/i386/sse.md (sse4_1_<code>v8qiv8hi2<mask_name>): Replace
444 nonimmediate_operand with register_operand.
445 (avx2_<code>v8qiv8si2<mask_name>): Likewise.
446 (sse4_1_<code>v4qiv4si2<mask_name>): Likewise.
447 (sse4_1_<code>v4hiv4si2<mask_name>): Likewise.
448 (sse4_1_<code>v2qiv2di2<mask_name>): Likewise.
449 (avx512f_<code>v8qiv8di2<mask_name>): Likewise.
450 (avx2_<code>v4qiv4di2<mask_name>): Likewise.
451 (avx2_<code>v4hiv4di2<mask_name>): Likewise.
452 (sse4_1_<code>v2hiv2di2<mask_name>): Likewise.
453 (sse4_1_<code>v2siv2di2<mask_name>): Likewise.
454 (*sse4_1_<code>v8qiv8hi2<mask_name>_1): New pattern.
455 (*sse4_1_<code>v8qiv8hi2<mask_name>_2): Likewise.
456 (*avx2_<code>v8qiv8si2<mask_name>_1): Likewise.
457 (*avx2_<code>v8qiv8si2<mask_name>_2): Likewise.
458 (*sse4_1_<code>v4qiv4si2<mask_name>_1): Likewise.
459 (*sse4_1_<code>v4qiv4si2<mask_name>_2): Likewise.
460 (*sse4_1_<code>v4hiv4si2<mask_name>_1): Likewise.
461 (*sse4_1_<code>v4hiv4si2<mask_name>_2): Likewise.
462 (*avx512f_<code>v8qiv8di2<mask_name>_1): Likewise.
463 (*avx512f_<code>v8qiv8di2<mask_name>_2): Likewise.
464 (*avx2_<code>v4qiv4di2<mask_name>_1): Likewise.
465 (*avx2_<code>v4qiv4di2<mask_name>_2): Likewise.
466 (*avx2_<code>v4hiv4di2<mask_name>_1): Likewise.
467 (*avx2_<code>v4hiv4di2<mask_name>_2): Likewise.
468 (*sse4_1_<code>v2hiv2di2<mask_name>_1): Likewise.
469 (*sse4_1_<code>v2hiv2di2<mask_name>_2): Likewise.
470 (*sse4_1_<code>v2siv2di2<mask_name>_1): Likewise.
471 (*sse4_1_<code>v2siv2di2<mask_name>_2): Likewise.
473 2018-11-21 H.J. Lu <hongjiu.lu@intel.com>
475 * read-rtl.c (apply_subst_iterator): Handle define_split and
476 define_insn_and_split.
478 2018-11-21 Jakub Jelinek <jakub@redhat.com>
480 PR rtl-optimization/87817
481 * config/i386/i386.md (bmi2_bzhi_<mode>3, *bmi2_bzhi_<mode>3,
482 *bmi2_bzhi_<mode>3_1, *bmi2_bzhi_<mode>3_1_ccz): Use IF_THEN_ELSE
483 in the pattern to avoid triggering UB when operands[2] is zero.
484 (tbm_bextri_<mode>): New expander. Renamed the old define_insn to ...
485 (*tbm_bextri_<mode>): ... this.
487 2018-11-21 Tom de Vries <tdevries@suse.de>
490 * params.def (HOT_BB_COUNT_FRACTION): Terminate help message with
493 2018-11-21 Andreas Krebbel <krebbel@linux.ibm.com>
495 * configure.ac: Add check for Binutils to determine whether vector
496 load/store alignments hints are being supported.
497 * config.in: Regenerate.
498 * configure: Regenerate.
499 * config/s390/s390.c (print_operand): Support new output
501 * config/s390/s390.md ("movti"): Append alignment hint output
502 using the new output modifier 'A'.
503 * config/s390/vector.md ("mov<mode>", "*vec_tf_to_v1tf")
504 ("*vec_ti_to_v1ti"): Likewise.
506 2018-11-20 Martin Sebor <msebor@redhat.com>
508 * doc/extend.texi (Other Builtins): Add __builtin_has_attribute.
510 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
513 * ipa-devirt.c (odr_types_equivalent_p): Use operand_equal_p to
516 2018-11-20 Jakub Jelinek <jakub@redhat.com>
518 PR tree-optimization/87895
519 * omp-simd-clone.c (ipa_simd_modify_function_body): When removing
520 or replacing GIMPLE_RETURN, set EDGE_FALLTHRU on the edge to EXIT.
521 (simd_clone_adjust): Don't set EDGE_FALLTHRU here. In a loop that
522 redirects edges to EXIT to edges to incr_bb, iterate while EXIT
523 has any preds and always use EDGE_PRED (, 0).
525 2018-11-20 Uros Bizjak <ubizjak@gmail.com>
528 * mode-switching.c (create_pre_exit): After reload, always split the
529 fallthrough edge to the exit block.
531 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
533 * ipa-devirt.c (add_type_duplicate): Do not ICE on incomplete enums.
534 * tree.c (build_array_type_1): Forward declare.
535 (fld_type_variant_equal_p): Add INNER_TYPE parameter.
536 (fld_type_variant): Likewise.
537 (fld_simplified_types): New hash.
538 (fld_process_array_type): New function.
539 (fld_incomplete_type_of): Handle array and enumeration types.
540 (fld_simplified_type): Handle simplification of arrays.
541 (free_lang_data): Allocate and free simplified types hash.
543 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
546 * ipa-devirt.c (odr_subtypes_equivalent_p): Report ODR violation
547 when sybtype already violates ODR.
548 (get_odr_type): Do not ICE when insert is false and type duplicate
549 is not registered yet.
550 (register_odr_type): Be sure to register subtypes first.
552 2018-11-20 Andreas Krebbel <krebbel@linux.ibm.com>
554 * config/s390/s390.md ("clztidi2"): Swap the RTX's written to the
555 DImode parts of the target operand.
557 2018-11-20 Nathan Sidwell <nathan@acm.org>
560 * Makefile.in (bitmap.o-warn): Use -Wno-error=array-bounds.
562 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
565 * tree.c (free_lang_data_in_cgraph): Add argument fld; break out
567 (free_lang_data) ... here; update call of free_lang_data_in_cgraph.
569 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
572 * ipa-fnsummary.c (pass_ipa_fnsummary): Do not remove functions
573 * ipa.c (possible_inline_candidate_p): Break out from ..
574 (process_references): ... here ; drop before_inlining_p;
575 cleanup handling of alises.
576 (walk_polymorphic_call_targets): Likewise.
577 (symbol_table::remove_unreachable_nodes): Likewise.
578 * passes.c (pass_data_ipa_remove_symbols): New structure.
579 (pass_ipa_remove_symbols): New pass.
580 (make_pass_ipa_remove_symbols): New function.
581 * tree-pass.h (make_pass_ipa_remove_symbols): Declare.
582 * passes.def (pass_ipa_remove_symbols): Schedule after early passes.
584 2018-11-20 Richard Biener <rguenther@suse.de>
586 * tree-vect-stmts.c (vectorizable_condition): Do not get
587 at else_clause vect def for EXTRACT_LAST_REDUCTION. Remove
588 pointless vect_is_simple_use calls.
590 2018-11-20 Richard Biener <rguenther@suse.de>
592 PR tree-optimization/88074
593 * tree-ssa-dom.c (pass_dominator::execute): Do not walk
596 2018-11-20 Richard Biener <rguenther@suse.de>
598 PR tree-optimization/88069
599 * tree-ssa-sccvn.c (visit_phi): Do not value-number to unvisited
600 virtual PHI arguments.
602 2018-11-20 Ilya Leoshkevich <iii@linux.ibm.com>
605 * config/s390/s390.md: Skip LT(G) peephole when literal pool is
607 * rtl.h (contains_constant_pool_address_p): New function.
608 * rtlanal.c (contains_constant_pool_address_p): Likewise.
610 2018-11-20 Richard Biener <rguenther@suse.de>
613 * alias.c (component_uses_parent_alias_set_from): Remove
614 alias-set zero and TYPE_TYPELESS_STORAGE case both already
615 handled in other ways.
617 2018-11-20 Richard Biener <rguenther@suse.de>
619 PR tree-optimization/88087
620 * tree-ssa-pre.c (create_expression_by_pieces): Re-materialize
622 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Remember
625 2018-11-20 Richard Biener <rguenther@suse.de>
628 * tree-data-ref.c (lambda_matrix_right_hermite): Use abs_hwi.
630 2018-11-19 Jan Hubicka <hubicka@ucw.cz>
633 * ipa-devirt.c (free_enum_values): Do not ICE on ODR vilations.
635 2018-11-19 Sandra Loosemore <sandra@codesourcery.com>
638 * doc/invoke.texi (Link Options): Mention shared libraries
639 in documentation for the -l option. Simplify discussion and
640 point to the system linker documentation for details.
642 2018-11-19 Peter Bergner <bergner@linux.ibm.com>
644 PR rtl-optimization/88033
645 * ira-lives.c (non_conflicting_reg_copy_p): Skip copies from a register
646 to itself. Use HARD_REGISTER_NUM_P.
648 2018-11-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
650 * tree-vect-loop.c (vect_transform_loop): Disable further unrolling
651 of the loop if vf is non-constant.
653 2018-11-19 David Malcolm <dmalcolm@redhat.com>
655 PR tree-optimization/87025
656 * dumpfile.c (dump_context::begin_scope): Call end_any_optinfo
657 immediately after creating the scope optinfo.
658 (selftest::test_pr87025): New function.
659 (selftest::dumpfile_c_tests): Call it.
660 * optinfo-emit-json.cc (optrecord_json_writer::pop_scope): Assert
661 that we're not popping the top-level records array.
662 * optinfo.cc (optinfo::handle_dump_file_kind): Assert that we're
663 not changing the kind of a "scope" optinfo.
665 2018-11-19 David Malcolm <dmalcolm@redhat.com>
667 PR tree-optimization/87025
668 * dump-context.h: Include "optinfo.h".
669 (class optrecord_json_writer): New forward decl.
670 (dump_context::forcibly_enable_optinfo_p): Delete.
671 (dump_context::optinfo_enabled_p): New member function.
672 (dump_context::optimization_records_enabled_p): New member
674 (dump_context::set_json_writer): New member function.
675 (dump_context::emit_optinfo): New member function.
676 (dump_context::m_forcibly_enable_optinfo): Delete.
677 (dump_context::m_json_writer): New member data.
678 * dumpfile.c (dump_context::set_json_writer): New member function.
679 (dump_context::finish_any_json_writer): New member function.
680 (dump_context::end_scope): Replace call to
681 optimization_records_maybe_pop_dump_scope with call to
682 m_json_writer->pop_scope.
683 (dump_context::optinfo_enabled_p): New member function.
684 (dump_context::end_any_optinfo): Replace call to optinfo::emit with call
685 to dump_context::emit_optinfo.
686 (dump_context::emit_optinfo): New member function.
687 (temp_dump_context::temp_dump_context): Replace
688 m_forcibly_enable_optinfo with call to set_json_writer.
689 (temp_dump_context::~temp_dump_context): Clean up any json writer.
690 * optinfo-emit-json.cc (class optrecord_json_writer): Move to
692 (the_json_writer): Delete.
693 (optimization_records_start): Delete.
694 (optimization_records_finish): Delete.
695 (optimization_records_enabled_p): Delete, in favor of
696 dump_context::optimization_records_enabled_p.
697 (optimization_records_maybe_record_optinfo): Delete.
698 (optimization_records_maybe_pop_dump_scope): Delete.
699 * optinfo-emit-json.h: Include "json.h". Delete forward
701 (optimization_records_start): Delete.
702 (optimization_records_finish): Delete.
703 (optimization_records_enabled_p): Delete.
704 (optimization_records_maybe_record_optinfo): Delete.
705 (optimization_records_maybe_pop_dump_scope): Delete.
706 (class optrecord_json_writer): Move here from
707 optinfo-emit-json.cc.
708 * optinfo.cc (optinfo::emit_for_opt_problem): Replace call
709 to optinfo::emit with call to dump_context::emit_optinfo.
710 (optinfo::emit): Delete, in favor of dump_context::emit_optinfo.
711 (optinfo_enabled_p): Delete, in favor of
712 dump_context::optinfo_enabled_p.
713 (optinfo_wants_inlining_info_p): Update for conversion o
714 optimization_records_enabled_p to a member function of
716 * optinfo.h (optinfo_enabled_p): Delete, in favor of
717 dump_context::optinfo_enabled_p.
718 (optinfo::emit): Delete, in favor of dump_context::emit_optinfo.
719 * toplev.c: Include "dump-context.h".
720 (compile_file): Replace call to optimization_records_finish with
721 dump_context::finish_any_json_writer.
722 (do_compile): Replace call to optimization_records_start with
723 conditionally creating a optrecord_json_writer for the
726 2018-11-19 Jakub Jelinek <jakub@redhat.com>
728 PR tree-optimization/88071
729 * tree-vect-loop.c (vectorize_fold_left_reduction): Pass true instead
730 of false as last argument to gsi_remove.
731 * tree-vect-stmts.c (vect_finish_replace_stmt): Pass true instead of
732 false as last argument to gsi_replace.
735 * omp-expand.c: Don't include debug.h.
736 (adjust_context_and_scope): Add REGION argument. Find DECL_CONTEXT
737 from innermost outer parallel, task, teams or target that has a
738 child_fn set, or, if there is no such outer region, use
739 current_function_decl. Do the DECL_CONTEXT adjustment regardless of
740 whether a suitable BLOCK is found or not.
741 (expand_parallel_call, expand_teams_call): Don't call
742 adjust_context_and_scope here.
743 (grid_expand_target_grid_body): Revert 2017-01-25 changes.
744 (expand_omp_taskreg, expand_omp_target): Likewise. Call
745 adjust_context_and_scope.
746 * dwarf2out.c (dwarf2out_early_global_decl): For
747 decl_function_context recurse instead of calling dwarf2out_decl.
749 2018-11-20 Eric Botcazou <ebotcazou@adacore.com>
751 PR rtl-optimization/85925
752 * rtl.h (word_register_operation_p): New predicate.
753 * combine.c (record_dead_and_set_regs_1): Only apply specific handling
754 for WORD_REGISTER_OPERATIONS targets to word_register_operation_p RTX.
755 * rtlanal.c (nonzero_bits1): Likewise. Adjust couple of comments.
756 (num_sign_bit_copies1): Likewise.
758 2018-11-19 Richard Biener <rguenther@suse.de>
761 * tree.c (free_lang_data_in_one_sizepos): Free non-gimple-val
764 2018-11-19 Eric Botcazou <ebotcazou@adacore.com>
766 * compare-elim.c (struct comparison): Add not_in_a field.
767 (is_not): New static function.
768 (strip_not): Likewise.
769 (conforming_compare): Handle a NOT in the first operand.
770 (can_eliminate_compare): Likewise.
771 (find_comparison_dom_walker::before_dom_children): Likewise.
772 (try_eliminate_compare): Likewise.
773 * config/visium/visium.md (negsi2_insn_set_carry): Turn into...
774 (neg<mode>2_insn_set_carry): ...this and add missing NEG operation.
776 2018-11-19 Jonathan Wakely <jwakely@redhat.com>
778 * doc/extend.texi (Common Type Attributes): Fix typos.
780 2018-11-19 Martin Liska <mliska@suse.cz>
782 PR gcov-profile/88045
783 * coverage.c (coverage_begin_function): Add assert.
785 2018-11-18 Sandra Loosemore <sandra@codesourcery.com>
788 * doc/extend.texi (Common Function Attributes): Document that
789 no_instrument_function applies to -p and -pg, too.
790 * doc/invoke.texi (Instrumentation Options): Add cross-references
791 to docs for -p, -pg, and -finstrument-functions.
793 2018-11-18 Alan Modra <amodra@gmail.com>
795 * config/gnu-user.h (GNU_USER_TARGET_CRTI): Define.
796 (GNU_USER_TARGET_STARTFILE_SPEC): Use it here.
797 (GNU_USER_TARGET_CRTN): Define.
798 (GNU_USER_TARGET_ENDFILE_SPEC): Use it here.
799 (GNU_USER_TARGET_LINK_GCC_C_SEQUENCE_SPEC): Define.
801 2018-11-18 Alan Modra <amodra@gmail.com>
803 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Delete
804 !HAVE_LD_PIE variant.
805 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
807 2018-11-17 Nick Clifton <nickc@redhat.com>
808 Sandra Loosemore <sandra@codesourcery.com>
811 * doc/invoke.texi (Overall Options): Document that --help and
812 --help=<value> options cannot be combined.
814 2018-11-17 Jim Wilson <jimw@sifive.com>
816 * config/riscv/riscv.c (epilogue_cfa_sp_offset): New.
817 (riscv_restore_reg): If restoring HARD_FRAME_POINTER_REGNUM, and
818 epilogue_cfa_sp_offset set, then add REG_CFA_DEF_CFA regnote.
819 (riscv_expand_epilogue): Initialize epilogue_cfa_sp_offset. Set it
820 to step2 if frame_pointer_needed and step1 is 0.
822 2018-11-17 Sandra Loosemore <sandra@codesourcery.com>
825 * doc/cpp.texi (System Headers): Add note about implicit
826 extern "C" block on targets that define SYSTEM_IMPLICIT_EXTERN_C.
828 2018-11-17 Sandra Loosemore <sandra@codesourcery.com>
831 * doc/invoke.texi (C++ Dialect Options): Clarify usage of
832 -fno-implicit-templates.
834 2018-11-17 Jakub Jelinek <jakub@redhat.com>
836 PR tree-optimization/87546
837 * tree-vect-patterns.c (vect_look_through_possible_promotion): Add
838 min_precision variable, initially set it to orig_precision, only does
839 something if op_type's precision is <= min_precision and update
840 min_precision whenever calling set_op.
842 2018-11-16 Jan Hubicka <hubicka@ucw.cz>
845 * ipa-devirt.c (warn_odr): Look for main variant to get TYPE_DECL.
847 2018-11-16 Sandra Loosemore <sandra@codesourcery.com>
849 * doc/invoke.texi (Option Summary): Fix whitespace and line
850 breaks in @gccoptlist environments.
851 (Warning Options): Likewise.
852 (Optimize Options): Likewise.
853 (PowerPC SPE Options): Likewise.
854 (RS/6000 and PowerPC Options): Likewise.
856 2018-11-16 Jeff Law <law@redhat.com>
858 * config/mn10300/mn10300.md (adddi3_degenerate): Remove bogus
861 2018-11-16 Sandra Loosemore <sandra@codesourcery.com>
864 * doc/invoke.texi (Optimize Options): Update options enabled by
865 fprofile-generate, -fprofile-use, and -fauto-profile.
867 2018-11-16 Jozef Lawrynowicz <jozef.l@mittosystems.com>
870 * target-def.h: Initialize TARGET_ASM_{,UN}ALIGNED_P{S,D,T}I_OP.
871 Add them to the TARGET_ASM_{,UN}ALIGNED_INT_OP structs.
872 * target.def: Enumerate TARGET_ASM_{,UN}ALIGNED_P{S,D,T}I_OP in
874 * target.h: Add psi, pdi, pti to struct asm_int_op definition.
875 * targhooks.c (default_print_patchable_function_entry): Assert
876 asm_int_op does not return a NULL string.
877 * varasm.c (integer_asm_op): Return the op for a partial int type
878 when the requested size does not correspond to an integer type.
879 * config/msp430/msp430.c: Initialize TARGET_ASM_{,UN}ALIGNED_PSI_OP.
880 * doc/tm.texi: Regenerate.
882 2018-11-12 Jason Merrill <jason@redhat.com>
884 * gimplify.c (gimplify_case_label_expr): Handle hot/cold attributes.
886 2018-11-16 Michael Meissner <meissner@linux.ibm.com>
888 * config/rs6000/constraints.md (wF constraint): Remove power9
889 fusion documentation. Just document wF's use for power8 fusion.
890 * config/rs6000/predicates.md (p9_fusion_reg_operand): Delete.
891 (fusion_gpr_addis): Delete power9 fusion support. Change power8
892 fusion support to require the upper 12 bits to be all 0's or all
894 (fusion_gpr_mem_load): Add comment.
895 (fusion_addis_mem_combo_load): Remove power9 fusion support.
896 (fusion_addis_mem_combo_store): Delete.
897 (fusion_offsettable_mem_operand): Delete.
898 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Do not set
900 (ISA_3_0_MASKS_SERVER): Delete power9 fusion.
901 (POWERPC_MASKS): Delete power9 fusion.
902 * config/rs6000/rs6000-protos.h (emit_fusion_load_store): Delete.
903 (fusion_p9_p): Delete.
904 (expand_fusion_p9_load): Delete.
905 (expand_fusion_p9_store): Delete.
906 (emit_fusion_p9_load): Delete.
907 (emit_fusion_p9_store): Delete.
908 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Delete power9
910 (rs6000_option_override_internal): Set power8 fusion based on
911 whether we are tuning for power8. Delete power9 fusion support.
912 (rs6000_opt_masks): Delete -mpower9-fusion switch.
913 (emit_fusion_load): Rename emit_fusion_load_store to
914 emit_fusion_load, and drop fusion store support. Update callers.
915 (emit_fusion_load_store): Likewise.
916 (emit_fusion_gpr_load): Likewise.
917 (fusion_p9_p): Delete.
918 (expand_fusion_p9_load): Delete.
919 (expand_fusion_p9_store): Delete.
920 (emit_fusion_p9_load): Delete.
921 (emit_fusion_p9_store): Delete.
922 * config/rs6000/rs6000.md (UNSPEC_FUSION_P9): Delete.
923 (GPR_FUSION): Delete.
924 (FPR_FUSION): Delete.
925 (power9 fusion peephole2s): Delete.
926 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Delete.
927 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Delete.
928 (fusion_vsx_<P:mode>_<FPR_FUSION:mode>_load): Delete.
929 (fusion_vsx_<P:mode>_<FPR_FUSION:mode>_store): Delete.
930 (fusion_p9_<mode>_constant): Delete.
931 * config/rs6000/rs6000.opt (-mpower9-fusion): Delete undocumented
932 power9 fusion switch.
933 * doc/md.texi (PowerPC constraints): Update wF constraint
934 documentation for power8 fusion only.
936 2018-11-16 Jakub Jelinek <jakub@redhat.com>
938 PR rtl-optimization/87475
939 * cfgrtl.c (patch_jump_insn): Allow redirection failure for
940 CROSSING_JUMP_P insns.
941 (cfg_layout_redirect_edge_and_branch): Don't ICE if ret is NULL.
943 2018-11-16 Uros Bizjak <ubizjak@gmail.com>
946 * config/i386/i386.md (floatunsdidf2): Allow only 64bit AVX512F targets.
947 * config/i386/sse.md (UNSPEC_MOVDI_TO_SSE): New UNSPEC.
948 (movdi_to_sse): Rewrite using UNSPEC_MOVDI_TO_SSE unspec.
950 2018-11-16 Jakub Jelinek <jakub@redhat.com>
953 * optabs.c (expand_binop): For op0_mode use GET_MODE (op0), unless it
954 is VOIDmode, in which case use int_mode. Similarly for op1_mode.
956 2018-11-16 Eric Botcazou <ebotcazou@adacore.com>
958 * tree-switch-conversion.h (switch_decision_tree::emit_case_nodes): Add
959 location_t parameter.
960 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
961 (switch_decision_tree::do_jump_if_equal): Likewise.
962 * tree-switch-conversion.c (switch_decision_tree::emit): Pass location
963 of switch statement to emit_case_nodes.
964 (switch_decision_tree::emit_cmp_and_jump_insns): Add LOC parameter and
965 set it on the newly built GIMPLE comparison statement.
966 (switch_decision_tree::do_jump_if_equal): Likewise.
967 (switch_decision_tree::emit_case_nodes): Add LOC parameter and pass it
968 in calls to do_jump_if_equal as well as recursive calls.
970 2018-11-16 Andi Kleen <ak@linux.intel.com>
972 * config/i386/i386.md: Allow memory operands to ptwrite.
974 2018-11-16 Michael Matz <matz@suse.de>
976 * system.h (PRsa): New macro.
977 (SIZE_AMOUNT): Cast number to uint64_t.
978 * alloc-pool.h (pool_usage::dump): Don't use %zu but PRsa.
979 (pool_usage::dump_footer): Likewise and also use PRIu64.
980 * bitmap.h (bitmap_usage::dump): Likewise.
981 * ggc-common.c (ggc_usage::dump): Likewise.
982 * ggc-page.c (ggc_print_statistics): Likewise.
983 * input.c (dump_line_table_statistics): Likewise.
984 * mem-stats.h (mem_usage::dump): Likewise.
985 (mem_usage::dump_footer): Likewise.
986 * rtl.c (dump_rtx_statistics): Likewise.
987 * tree-cfg.c (dump_cfg_stats): Likewise.
988 * tree-dfa.c (dump_dfa_stats): Likewise.
989 * tree-phinodes.c (phinodes_print_statistics): Likewise.
990 * tree-ssanames (ssanames_print_statistics): Likewise.
991 * vec.c (vec_usage::dump): Likewise.
992 (vec_usage::dump_footer): Likewise.
994 2018-11-16 Richard Biener <rguenther@suse.de>
996 PR tree-optimization/88011
997 * tree-vrp.c (extract_range_from_binary_expr): Fix error in
998 replacing set_value_range_to_undefined and
999 set_value_range_to_varying with method calls.
1001 2018-11-16 Ilya Leoshkevich <iii@linux.ibm.com>
1003 * config/s390/s390.md
1004 (*r<noxa>sbg_<mode>_srl_bitmask): Do not delegate arithmetic to
1006 (*r<noxa>sbg_<mode>_sll): Likewise.
1007 (*r<noxa>sbg_<mode>_srl): Likewise.
1008 (*r<noxa>sbg_sidi_srl): New pattern.
1010 2018-11-16 Jerome Lambourg <lambourg@adacore.com>
1012 * config/vxworks.h (TARGET_VXWORKS_HAVE_CTORS_DTORS): New macro.
1013 Default to TARGET_VXWORKS_RTP.
1014 (SUPPORTS_INIT_PRIORITY): Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
1015 of TARGET_VXWORKS_RTP.
1016 * config/vxworksae.h: Also define TARGET_VXWORKS_HAVE_CTORS_DTORS.
1017 * config/vxworks.c: Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
1018 of TARGET_VXWORKS_RTP to set targetm.have_ctors_dtors.
1020 2018-11-15 Sandra Loosemore <sandra@codesourcery.com>
1023 * doc/extend.texi (Common Type Attributes): Make it explicit
1024 that attribute "packed" can apply to C++ classes.
1026 2018-11-15 Martin Sebor <msebor@redhat.com>
1030 * tree.c (type_argument_type): New function.
1031 * tree.h (type_argument_type): Declare it.
1032 * gcc/doc/extend.texi (alloc_align): Update and clarify.
1033 (alloc_size, nonnull, sentinel): Same.
1035 2018-11-15 Andrew Stubbs <ams@codesourcery.com>
1036 Kwok Cheung Yeung <kcy@codesourcery.com>
1038 * tree-vect-stmts.c (vectorizable_store): Don't ICE when
1039 int_mode_for_size fails.
1040 (vectorizable_load): Likewise.
1042 2018-11-15 David Malcolm <dmalcolm@redhat.com>
1044 * doc/ux.texi (Group logically-related diagnostics): Move
1045 discussion of auto_diagnostic_group into this new subsection.
1046 Give an example of where this grouping is used.
1048 2018-11-15 David Malcolm <dmalcolm@redhat.com>
1051 * Makefile.in (OBJS): Move json.o to...
1052 (OBJS-libcommon): ...here and add diagnostic-format-json.o.
1053 * common.opt (fdiagnostics-format=): New option.
1054 (diagnostics_output_format): New enum.
1055 * diagnostic-format-json.cc: New file.
1056 * diagnostic.c (default_diagnostic_final_cb): New function, taken
1057 from start of diagnostic_finish.
1058 (diagnostic_initialize): Initialize final_cb to
1059 default_diagnostic_final_cb.
1060 (diagnostic_finish): Move "being treated as errors" messages to
1061 default_diagnostic_final_cb. Call any final_cb.
1062 (default_diagnostic_finalizer): Add diagnostic_t param.
1063 (diagnostic_report_diagnostic): Pass "orig_diag_kind" to
1064 diagnostic_finalizer callback.
1065 * diagnostic.h (enum diagnostics_output_format): New enum.
1066 (diagnostic_finalizer_fn): Reimplement, adding diagnostic_t param.
1067 (struct diagnostic_context): Add "final_cb".
1068 (default_diagnostic_finalizer): Add diagnostic_t param.
1069 (diagnostic_output_format_init): New decl.
1070 * doc/invoke.texi (-fdiagnostics-format): New option.
1071 * dwarf2out.c (gen_producer_string): Ignore
1072 OPT_fdiagnostics_format_.
1073 * gcc.c (driver_handle_option): Handle OPT_fdiagnostics_format_.
1074 * lto-wrapper.c (append_diag_options): Ignore it.
1075 * opts.c (common_handle_option): Handle it.
1077 2018-11-15 David Malcolm <dmalcolm@redhat.com>
1079 PR tree-optimization/88015
1080 * graphite-isl-ast-to-gimple.c
1081 (translate_isl_ast_to_gimple::scop_to_isl_ast): Add missing check
1083 * graphite-sese-to-poly.c (build_poly_scop): Likewise.
1085 2018-11-15 Richard Biener <rguenther@suse.de>
1088 * gimple.c (gimple_call_flags): Union flags from decl, type
1090 * trans-mem.c (is_tm_pure_call): Simplify.
1092 2018-11-15 Richard Biener <rguenther@suse.de>
1094 PR tree-optimization/88031
1095 * tree-vect-loop.c (vectorizable_reduction): Move check
1096 for multiple types earlier so we get the expected dump.
1097 Simplify calls to vectorizable_condition.
1098 * tree-vect-stmts.h (vectorizable_condition): Update prototype.
1099 * tree-vect-stmts.c (vectorizable_condition): Instead of
1100 reduc_def and reduc_index take just a flag. Simplify
1101 code-generation now that we can rely on the defs being set up.
1102 (vectorizable_comparison): Remove unused argument.
1104 2018-11-15 Richard Biener <rguenther@suse.de>
1106 PR tree-optimization/88030
1107 * tree-complex.c (need_eh_cleanup): New global.
1108 (update_complex_assignment): Mark blocks that need EH update.
1109 (expand_complex_comparison): Likewise.
1110 (tree_lower_complex): Allocate and deallocate need_eh_cleanup,
1111 perform EH cleanup and schedule CFG cleanup if that did anything.
1113 2018-11-15 Jakub Jelinek <jakub@redhat.com>
1115 PR rtl-optimization/88018
1116 * cfgrtl.c (fixup_abnormal_edges): Guard moving insns to fallthru edge
1117 on the presence of fallthru edge, rather than if it is a USE or not.
1119 2018-11-15 Richard Biener <rguenther@suse.de>
1122 * tree-data-ref.c (analyze_miv_subscript): Guard calls to
1123 analyze_subscript_affine_affine properly.
1125 2018-11-15 Xianmiao Qu <xianmiao_qu@c-sky.com>
1127 * config/csky/csky-linux-elf.h (LINUX_DYNAMIC_LINKER): Remove.
1128 (GLIBC_DYNAMIC_LINKER): Define.
1129 (LINUX_TARGET_LINK_SPEC): Update the dynamic linker's name.
1131 2018-11-15 Bin Cheng <bin.cheng@linux.alibaba.com>
1133 PR tree-optimization/84648
1134 * tree-ssa-loop-niter.c (adjust_cond_for_loop_until_wrap): New.
1135 (number_of_iterations_cond): Adjust exit cond for loop-until-wrap case
1136 by calling adjust_cond_for_loop_until_wrap.
1138 2018-11-15 Sandra Loosemore <sandra@codesourcery.com>
1141 * doc/extend.texi (Common Function Attributes): Clarify linker
1142 restrictions on "aligned" attribute.
1143 (Common Variable Attributes): Likewise. Mention that linker
1144 restrictions don't apply to stack-allocated variables.
1146 2018-11-15 Joern Rennecke <joern.rennecke@riscy-ip.com>
1148 * gensupport.c (add_predicate_code): Properly handle ZERO_EXTRACT
1151 2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
1155 * doc/invoke.texi (Optimize Options): Remove bad example about
1156 interaction between -flto and -O. Replace it with a note that
1157 you need to compile with -O and not just link. Copy-edit -flto
1158 discussion to reduce verbiage and improve flow.
1160 2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
1163 * doc/invoke.texi (Optimize Options): Clarify that -O0 and -Og
1164 also suppress many optimizations. Alphabetize option lists for
1165 -O1, -O2, and -Os. Add list of options disabled with -Og, and
1166 correct documentation for those options to say that.
1167 * opts.c (default_options_table): Sort table by level and option
1168 name, to make it easier to correlate to the manual.
1170 2018-11-14 Uros Bizjak <ubizjak@gmail.com>
1172 * config/i386/i386.c (ix86_print_operand_address_as): Simplify
1173 printing of the names of segment registers.
1175 2018-11-14 Nathan Sidwell <nathan@acm.org>
1179 * dwarf2out.c (dwarf2out_finish): Apply resolve_addr to comdat
1182 2018-11-14 David Malcolm <dmalcolm@redhat.com>
1184 * Makefile.in (CFLAGS-optinfo-emit-json.o): Add $(ZLIBINC).
1186 2018-11-14 Jakub Jelinek <jakub@redhat.com>
1188 P1236R1 - Signed integers are two's complement
1189 * doc/invoke.texi (Wshift-overflow): Adjust documentation for c++2a.
1192 * hash-map.h (hash_map::iterator::reference_pair): New class.
1193 (hash_map::iterator::operator*): Return it rather than std::pair.
1195 2018-11-14 Jeff Law <law@redhat.com>
1197 * optabs.c (expand_binop): Pass INT_MODE to operand_subword_force
1198 iff the operand is a constant.
1200 2018-11-14 Aldy Hernandez <aldyh@redhat.com>
1202 * gimple-ssa-evrp-analyze.c
1203 (evrp_range_analyzer::record_ranges_from_incoming_edge): Rename
1204 ignore_equivs_equal_p to equal_p.
1205 * ipa-cp.c (meet_with_1): Use equal_p instead of
1206 ignore_equivs_equal_p.
1207 * ipa-prop.c (ipa_vr_ggc_hash_traits::equal): Same.
1208 * tree-vrp.c (value_range::ignore_equivs_equal_p): Remove.
1209 (value_range::operator==): Remove.
1210 (value_range::operator!=): Remove.
1211 (vrp_prop::visit_stmt): Use equal_p.
1212 * tree-vrp.h (value_range): Remove operator==, operator!=,
1213 ignore_equivs_equal_p.
1214 * vr-values.c (update_value_range): Use equal_p.
1216 2018-11-14 Michael Matz <matz@suse.de>
1219 * gimplify.c (collect_fallthrough_labels): Add new argument,
1220 return location via that, don't modify statements.
1221 (warn_implicit_fallthrough_r): Adjust call, don't use
1222 statement location directly.
1224 2018-11-14 Richard Biener <rguenther@suse.de>
1227 * tree-data-ref.c (split_constant_offset): Add wrapper
1228 allocating a cache hash-map.
1229 (split_constant_offset_1): Cache results of expanding
1230 expressions from SSA def stmts.
1232 2018-11-14 Richard Biener <rguenther@suse.de>
1235 * tree-data-ref.c (lambda_matrix_row_add): Change const1 argument
1237 (lambda_vector_mult_const): Likewise.
1238 (lambda_matrix_right_hermite): Use lambda_int temporaries.
1240 2018-11-14 Wilco Dijkstra <wdijkstr@arm.com>
1241 Jackson Woodruff <jackson.woodruff@arm.com>
1243 PR 71026/tree-optimization
1244 * match.pd: Simplify floating point comparisons.
1246 2018-11-14 Jakub Jelinek <jakub@redhat.com>
1248 PR rtl-optimization/87817
1249 * config/i386/i386.c (ix86_fold_builtin): For _bzhi_u{32,64} if
1250 last argument has low 8 bits clear, fold to 0.
1252 2018-11-14 Iain Sandoe <iain@sandoe.co.uk>
1254 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p): Cast
1255 MAX_OFILE_ALIGNMENT as needed.
1257 2018-11-14 Richard Biener <rguenther@suse.de>
1259 PR tree-optimization/88019
1260 * graphite-isl-ast-to-gimple.c (ternary_op_to_tree): Rewrite
1261 COND_EXPR arguments to non-trapping overflow.
1263 2018-11-14 Claudiu Zissulescu <claziss@synopsys.com>
1265 * config/arc/arc.h (reg_class): Reorder registers classes, remove
1266 unused register classes.
1267 (REG_CLASS_NAMES): Likewise.
1268 (REG_CLASS_CONTENTS): Likewise.
1269 (FIXED_REGISTERS): Make lp_count fixed.
1270 (BASE_REG_CLASS): Remove ACC16_BASE_REGS reference.
1271 (PROGRAM_COUNTER_REGNO): Remove.
1272 * config/arc/arc.c (arc_conditional_register_usage): Remove unused
1273 register classes, use constants for register numbers, remove
1274 reg_class_contents references.
1275 (arc_process_double_reg_moves): Add asserts.
1276 (arc_secondary_reload): Remove LPCOUNT_REG reference, use
1277 lra_in_progress predicate.
1278 (arc_init_reg_tables): Remove unused register classes.
1279 (arc_register_move_cost): Likewise.
1280 (arc_preferred_reload_class): Likewise.
1281 (hwloop_optimize): Update rtx patterns involving lp_count register.
1282 (arc_return_address_register): Rename ILINK1, INLINK2 regnums macros.
1283 * config/arc/constraints.md ("c"): Choose between GENERAL_REGS and
1284 CHEAP_CORE_REGS. Former one will be used for LRA.
1285 ("Rac"): Choose between GENERAL_REGS and ALL_CORE_REGS. Former
1286 one will be used for LRA.
1287 ("w"): Choose between GENERAL_REGS and WRITABLE_CORE_REGS. Former
1288 one will be used for LRA.
1289 ("W"): Choose between GENERAL_REGS and MPY_WRITABLE_CORE_REGS.
1290 Former one will be used for LRA.
1291 ("f"): Delete constraint.
1294 ("l"): Change it from register constraint to constraint.
1295 * config/arc/arc.md (movqi_insn): Remove unsed lp_count constraints.
1296 (movhi_insn): Likewise.
1297 (movsi_insn): Update pattern.
1300 (stack_tie): Remove 'b' constraint letter.
1302 (R9_REG, R15_REG, R16_REG, R25_REG): Likewise.
1303 (R32_REG, R40_REG, R41_REG, R42_REG, R43_REG, R44_REG): Likewise.
1304 (R57_REG, R59_REG, PCL_REG): Likewise.
1305 (ILINK1_REGNUM): Renamed to ILINK1_REG.
1306 (ILINK2_REGNUM): Renamed to ILINK2_REG.
1308 (SP_REGS): Likewise.
1309 (Rcw): Remove unused reg classes.
1310 * config/arc/predicates.md (dest_reg_operand): Just default on
1311 register_operand predicate.
1312 (mpy_dest_reg_operand): Likewise.
1313 (move_dest_operand): Use macros instead of constants.
1314 * config/arc/arc.opt (mlra): Switch to lra as default.
1316 2018-11-14 Richard Biener <rguenther@suse.de>
1318 PR tree-optimization/87974
1319 * tree-vect-loop.c (vectorizable_reduction): When computing
1320 the vectorized reduction PHI vector type ignore constant
1323 2018-11-14 Jakub Jelinek <jakub@redhat.com>
1325 PR tree-optimization/87977
1326 * tree-ssa-math-opts.c (optimize_recip_sqrt): Don't reuse division
1327 stmt, build a new one and replace the old one with it. Formatting fix.
1328 Call release_ssa_name (x) if !has_other_use and !delete_div.
1329 (pass_cse_reciprocals::execute): Before calling optimize_recip_sqrt
1330 verify lhs of stmt is still def.
1332 2018-11-13 Peter Bergner <bergner@linux.ibm.com>
1334 PR rtl-optimization/87507
1335 * lower-subreg.c (operand_for_swap_move_operator): New function.
1336 (simple_move): Strip simple operators.
1337 (find_pseudo_copy): Likewise.
1338 (resolve_operand_for_swap_move_operator): New function.
1339 (resolve_simple_move): Strip simple operators and swap operands.
1341 2018-11-13 Jakub Jelinek <jakub@redhat.com>
1343 PR tree-optimization/87898
1344 * omp-simd-clone.c (ipa_simd_modify_stmt_ops): Formatting fix.
1345 (ipa_simd_modify_function_body): Remove debug stmts where the first
1346 argument was changed into a non-decl.
1348 2018-11-13 Peter Bergner <bergner@linux.ibm.com>
1350 PR rtl-optimization/87899
1351 * lra-lives.c (start_living): Update white space in comment.
1352 (enum point_type): New.
1353 (sparseset_contains_pseudos_p): New function.
1354 (update_pseudo_point): Likewise.
1355 (make_hard_regno_live): Use HARD_REGISTER_NUM_P macro.
1356 (make_hard_regno_dead): Likewise. Remove ignore_reg_for_conflicts
1357 handling. Move early exit after adding conflicts.
1358 (mark_pseudo_live): Use HARD_REGISTER_NUM_P macro. Add early exit
1359 if regno is already live. Remove all handling of program points.
1360 (mark_pseudo_dead): Use HARD_REGISTER_NUM_P macro. Add early exit
1361 after adding conflicts. Remove all handling of program points and
1362 ignore_reg_for_conflicts.
1363 (mark_regno_live): Use HARD_REGISTER_NUM_P macro. Remove return value
1364 and do not guard call to mark_pseudo_live.
1365 (mark_regno_dead): Use HARD_REGISTER_NUM_P macro. Remove return value
1366 and do not guard call to mark_pseudo_dead.
1367 (check_pseudos_live_through_calls): Use HARD_REGISTER_NUM_P macro.
1368 (process_bb_lives): Use HARD_REGISTER_NUM_P and HARD_REGISTER_P macros.
1369 Use new function update_pseudo_point. Handle register copies by
1370 removing the source register from the live set. Handle INOUT operands.
1371 Update to the next program point using the unused_set, dead_set and
1373 (lra_create_live_ranges_1): Use HARD_REGISTER_NUM_P macro.
1375 2018-11-13 David Malcolm <dmalcolm@redhat.com>
1377 * builtins.c: Replace "source_location" with "location_t".
1378 * diagnostic-show-locus.c: Likewise.
1379 * diagnostic.c: Likewise.
1380 * dumpfile.c: Likewise.
1381 * gcc-rich-location.h: Likewise.
1382 * genmatch.c: Likewise.
1383 * gimple.h: Likewise.
1384 * gimplify.c: Likewise.
1385 * input.c: Likewise.
1386 * input.h: Likewise. Eliminate the typedef.
1387 * omp-expand.c: Likewise.
1388 * selftest.h: Likewise.
1389 * substring-locations.h (get_source_location_for_substring):
1391 (get_location_within_string): ...this.
1392 * tree-cfg.c: Replace "source_location" with "location_t".
1393 * tree-cfgcleanup.c: Likewise.
1394 * tree-diagnostic.c: Likewise.
1395 * tree-into-ssa.c: Likewise.
1396 * tree-outof-ssa.c: Likewise.
1397 * tree-parloops.c: Likewise.
1398 * tree-phinodes.c: Likewise.
1399 * tree-phinodes.h: Likewise.
1400 * tree-ssa-loop-ivopts.c: Likewise.
1401 * tree-ssa-loop-manip.c: Likewise.
1402 * tree-ssa-phiopt.c: Likewise.
1403 * tree-ssa-phiprop.c: Likewise.
1404 * tree-ssa-threadupdate.c: Likewise.
1405 * tree-ssa.c: Likewise.
1406 * tree-ssa.h: Likewise.
1407 * tree-vect-loop-manip.c: Likewise.
1409 2018-11-13 Michael Ploujnikov <michael.ploujnikov@oracle.com>
1411 * doc/extend.texi: Fix typo in the weakref description.
1413 2018-11-13 Richard Biener <rguenther@suse.de>
1415 PR tree-optimization/86991
1416 * tree-vect-loop.c (vect_is_slp_reduction): Delay reduction
1417 group building until we have successfully detected the SLP
1419 (vect_is_simple_reduction): Remove fixup code here.
1421 2018-11-13 David Malcolm <dmalcolm@redhat.com>
1423 * dumpfile.c (VERIFY_DUMP_ENABLED_P): New macro.
1424 (dump_gimple_stmt): Use it.
1425 (dump_gimple_stmt_loc): Likewise.
1426 (dump_gimple_expr): Likewise.
1427 (dump_gimple_expr_loc): Likewise.
1428 (dump_generic_expr): Likewise.
1429 (dump_generic_expr_loc): Likewise.
1430 (dump_printf): Likewise.
1431 (dump_printf_loc): Likewise.
1432 (dump_dec): Likewise.
1433 (dump_dec): Likewise.
1434 (dump_hex): Likewise.
1435 (dump_symtab_node): Likewise.
1436 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
1437 Guard dump call with dump_enabled_p.
1438 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl): Likewise.
1439 * graphite-optimize-isl.c (optimize_isl): Likewise.
1440 * graphite.c (graphite_transform_loops): Likewise.
1441 * tree-loop-distribution.c (pass_loop_distribution::execute): Likewise.
1442 * tree-parloops.c (parallelize_loops): Likewise.
1443 * tree-ssa-loop-niter.c (number_of_iterations_exit): Likewise.
1444 * tree-vect-data-refs.c (vect_analyze_group_access_1): Likewise.
1445 (vect_prune_runtime_alias_test_list): Likewise.
1446 * tree-vect-loop.c (vect_update_vf_for_slp): Likewise.
1447 (vect_estimate_min_profitable_iters): Likewise.
1448 * tree-vect-slp.c (vect_record_max_nunits): Likewise.
1449 (vect_build_slp_tree_2): Likewise.
1450 (vect_supported_load_permutation_p): Likewise.
1451 (vect_slp_analyze_operations): Likewise.
1452 (vect_slp_analyze_bb_1): Likewise.
1453 (vect_slp_bb): Likewise.
1454 * tree-vect-stmts.c (vect_analyze_stmt): Likewise.
1455 * tree-vectorizer.c (try_vectorize_loop_1): Likewise.
1456 (pass_slp_vectorize::execute): Likewise.
1457 (increase_alignment): Likewise.
1459 2018-11-13 David Malcolm <dmalcolm@redhat.com>
1462 * ipa-inline.c (report_inline_failed_reason): Guard calls to
1463 cl_target_option_print_diff and cl_optimization_print_diff with
1466 2018-11-13 David Malcolm <dmalcolm@redhat.com>
1468 * doc/invoke.texi (-fsave-optimization-record): Note that the
1469 output is compressed.
1470 * optinfo-emit-json.cc: Include <zlib.h>.
1471 (optrecord_json_writer::write): Compress the output.
1473 2018-11-13 Aldy Hernandez <aldyh@redhat.com>
1475 * tree-vrp.c (value_range_base::dump): Dump type.
1476 Do not use INF nomenclature for 1-bit types.
1477 (dump_value_range): Group all variants to common dumping code.
1478 (debug): New overloaded functions for value_ranges.
1479 (value_range_base::dump): Remove no argument version.
1480 (value_range::dump): Same.
1482 2018-11-13 Richard Biener <rguenther@suse.de>
1484 PR tree-optimization/87931
1485 * tree-vect-loop.c (vect_is_simple_reduction): Restrict
1486 nested cycles we support to latch computations vectorizable_reduction
1489 2018-11-13 Martin Liska <mliska@suse.cz>
1491 PR tree-optimization/87885
1492 * cfghooks.c (account_profile_record): Rename
1494 (profile_record_check_consistency): ... this.
1495 Calculate missing num_mismatched_freq_in.
1496 (profile_record_account_profile): New function
1497 that calculates time and size of a function.
1498 * cfghooks.h (struct profile_record): Remove
1500 (struct cfg_hooks): Remove after_pass flag.
1501 (account_profile_record): Rename to ...
1502 (profile_record_check_consistency): ... this.
1503 (profile_record_account_profile): New.
1504 * cfgrtl.c (rtl_account_profile_record): Remove
1506 * passes.c (check_profile_consistency): Do only
1508 (account_profile): Calculate size and time of
1510 (pass_manager::dump_profile_report): Reformat
1512 (execute_one_ipa_transform_pass): Call
1513 consistency check before clean upand call account_profile
1514 after a clean up is done.
1515 (execute_one_pass): Call check_profile_consistency and
1516 account_profile instead of using after_pass flag..
1517 * tree-cfg.c (gimple_account_profile_record): Likewise.
1519 2018-11-13 Martin Liska <mliska@suse.cz>
1522 * config/i386/i386.c (ix86_option_override_internal): Error
1523 about usage -mabi=ms and -fsanitize={,kernel-}address.
1525 2018-11-13 Andre Vieira <andre.simoesdiasvieira@arm.com>
1527 * config/aarch64/aarch64.c
1528 (aarch64_vectorize_preferred_vector_alignment): Change return type to
1530 (aarch64_simd_vector_alignment_reachable): Adapt to preferred vector
1531 alignment being a poly int.
1532 * doc/tm.texi (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Change
1533 return type to poly_uint64.
1534 * target.def (default_preferred_vector_alignment): Likewise.
1535 * targhooks.c (default_preferred_vector_alignment): Likewise.
1536 * targhooks.h (default_preferred_vector_alignment): Likewise.
1537 * tree-vect-data-refs.c (vect_calculate_target_alignment): Likewise.
1538 (vect_compute_data_ref_alignment): Adapt to vector alignment being a
1540 (vect_update_misalignment_for_peel): Likewise.
1541 (vect_enhance_data_refs_alignment): Likewise.
1542 (vect_find_same_alignment_drs): Likewise.
1543 (vect_duplicate_ssa_name_ptr_info): Likewise.
1544 (vect_setup_realignment): Likewise.
1545 (vect_can_force_dr_alignment_p): Change alignment parameter type to
1547 * tree-vect-loop-manip.c (get_misalign_in_elems): Learn to construct a
1548 mask with a compile time variable vector alignment.
1549 (vect_gen_prolog_loop_niters): Adapt to vector alignment being a poly
1551 (vect_do_peeling): Exit early if vector alignment is not constant.
1552 * tree-vect-stmts.c (ensure_base_align): Adapt to vector alignment being
1554 (vectorizable_store): Likewise.
1555 (vectorizable_load): Likweise.
1556 * tree-vectorizer.h (struct dr_vec_info): Make target_alignment field a
1558 (vect_known_alignment_in_bytes): Adapt to vector alignment being a
1560 (vect_can_force_dr_alignment_p): Change alignment parameter type to
1563 2018-11-13 Richard Biener <rguenther@suse.de>
1565 PR tree-optimization/87962
1566 * tree-vect-loop.c (vect_is_simple_reduction): More reliably
1567 detect outer reduction for disqualifying in-loop uses.
1569 2018-11-13 Richard Biener <rguenther@suse.de>
1571 PR tree-optimization/87967
1572 * tree-vect-loop.c (vect_transform_loop): Also copy PHIs
1573 for constants for the scalar loop.
1575 2018-11-13 Alan Modra <amodra@gmail.com>
1577 * config/rs6000/linux64.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Exclude
1578 integer constants when -mcmodel=medium.
1580 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
1582 * config/arc/arc.h (FUNCTION_PROFILER): Redefine to empty.
1583 * config/arc/elf.h (PROFILE_HOOK): Define.
1584 * config/arc/linux.h (PROFILE_HOOK): Likewise.
1586 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
1588 * config/arc/arc.c (hwloop_optimize): Bailout when detecting a
1589 jump table data in the text section.
1591 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
1593 * config/arc/arc.c (arc_eh_return_address_location): Repurpose it
1594 to fit the eh_return pattern.
1595 * config/arc/arc.md (eh_return): Define.
1596 (VUNSPEC_ARC_EH_RETURN): Likewise.
1597 * config/arc/arc-protos.h (arc_eh_return_address_location): Match
1599 * config/arc/arc.h (EH_RETURN_HANDLER_RTX): Remove it.
1602 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
1604 * common/config/arc/arc-common.c (arc_option_optimization_table):
1605 Millicode optimization is default on for size optimizations.
1606 * config/arc/arc-protos.h (arc_check_multi): New function.
1607 * config/arc/arc.c (RTX_OK_FOR_OFFSET_P): Rearange.
1608 (ENTER_LEAVE_START_REG): Define.
1609 (ENTER_LEAVE_END_REG): Likewise.
1610 (arc_override_options): Disable millicode when long calls option
1612 (arc_frame_info): Change it from int to bool.
1613 (arc_compute_frame_size): Clean up.
1614 (arc_save_restore): Remove.
1615 (frame_save_reg): New function.
1616 (frame_restore_reg): Likewise.
1617 (arc_enter_leave_p): Likewise.
1618 (arc_save_callee_saves): Likewise.
1619 (arc_restore_callee_saves): Likewise.
1620 (arc_save_callee_enter): Likewise.
1621 (arc_restore_callee_leave): Likewise.
1622 (arc_save_callee_milli): Likewise.
1623 (arc_restore_callee_milli): Likewise.
1624 (arc_expand_prologue): Reimplement to emit enter/leave
1626 (arc_expand_epilogue): Likewise.
1627 (arc_check_multi): New function.
1628 * config/arc/arc.md (push_multi_fp): New pattern.
1629 (push_multi_fp_blink): Likewise.
1630 (pop_multi_fp): Likewise.
1631 (pop_multi_fp_blink): Likewise.
1632 (pop_multi_fp_ret): Likewise.
1633 (pop_multi_fp_blink_ret): Likewise.
1634 * config/arc/arc.opt (mmillicode): Update option.
1635 (mcode-density-frame): New option.
1636 * config/arc/predicates.md (push_multi_operand): New predicate.
1637 (pop_multi_operand): Likewise.
1638 * doc/invoke.texi (ARC): Update ARC options information.
1641 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
1643 * config/arc/arc-protos.h (gen_operands_ldd_std): Add.
1644 * config/arc/arc.c (operands_ok_ldd_std): New function.
1645 (mem_ok_for_ldd_std): Likewise.
1646 (gen_operands_ldd_std): Likewise.
1647 * config/arc/arc.md: Add peephole2 rules for std/ldd.
1649 2018-11-13 Eric Botcazou <ebotcazou@adacore.com>
1651 * toplev.c (output_stack_usage): Turn test on flag_stack_usage into
1652 test on stack_usage_file.
1653 (lang_dependent_init): Do not open the .su file if generating LTO.
1655 2018-11-13 Jakub Jelinek <jakub@redhat.com>
1657 PR rtl-optimization/87918
1658 * simplify-rtx.c (simplify_merge_mask): For COMPARISON_P, use
1659 simplify_gen_relational rather than simplify_gen_binary.
1661 2018-11-13 Richard Biener <rguenther@suse.de>
1663 * tree-ssanames.h (set_range_info): Use value_range_base.
1664 (get_range_info): Likewise.
1665 * tree-ssanames.c (set_range_info): Likewise.
1666 (get_range_info): Likewise.
1667 * tree-vrp.c (value_range_base::union_helper): Split
1668 out common parts of value_range[_base]::union_.
1669 (value_range_base::union_): Update.
1670 (value_range::union_): Likewise.
1671 (determine_value_range_1): Use value_range_base.
1672 (determine_value_range): Likewise.
1673 * tree-vrp.h (value_range_base::union_helper): Move ...
1674 (value_range::union_helper): ... from here.
1676 2018-11-13 Alan Modra <amodra@gmail.com>
1678 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Negate
1680 (rs6000_secondary_reload_gpr): Don't call find_replacement.
1682 2018-11-13 Sandra Loosemore <sandra@codesourcery.com>
1685 * doc/invoke.texi (Optimize Options): Clarify that the
1686 l1-cache-line-size, l1-cache-size, and l2-cache-size parameters
1687 apply to data cache size.
1689 2018-11-13 Alan Modra <amodra@gmail.com>
1691 * config/rs6000/predicates.md (logical_const_operand),
1692 (logical_operand): Correct comment.
1693 * config/rs6000/rs6000.c (rs6000_emit_prologue): Comment fix.
1695 2018-11-13 Alan Modra <amodra@gmail.com>
1697 * config/rs6000/rs6000.md (addsi3_high): Prefix with '*'.
1699 2018-11-13 Alan Modra <amodra@gmail.com>
1701 * gcc/doc/md.texi (Machine Constraints): Correct rs6000 h constraint
1703 * config/rs6000/rs6000.md (movsi_internal1): Delete MT%0 case
1704 covered by alternative.
1705 (movcc_internal1): Ignore h for register preference.
1706 (mov<mode>_hardfloat64): Likewise.
1707 (mov<mode>_softfloat): Ignore c, l, h for register preference.
1709 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
1711 PR preprocessor/47823
1712 * doc/cpp.texi (Alternatives to Wrapper #ifndef): Move #pragma once
1715 * doc/extend.texi (Pragmas): Note additional pragmas documented
1718 2018-11-13 Kugan Vivekanandarajah <kuganv@linaro.org>
1722 * tree-scalar-evolution.c (expression_expensive_p): Make
1723 BUILTIN POPCOUNT as expensive when backend does not define it.
1725 2018-11-12 Fredrik Noring <noring@nocrew.org>
1727 * config.gcc: Update with-llsc defaults for MIPS r5900.
1729 2018-11-12 Martin Liska <mliska@suse.cz>
1731 PR gcov-profile/87442
1732 * common.opt: Add -fprofile-filter-files and -fprofile-exclude-files
1734 * doc/invoke.texi: Document them.
1735 * tree-profile.c (parse_profile_filter): New.
1736 (parse_profile_file_filtering): Likewise.
1737 (release_profile_file_filtering): Likewise.
1738 (include_source_file_for_profile): Likewise.
1739 (tree_profiling): Filter source files based on the
1740 newly added options.
1742 2018-11-12 Martin Liska <mliska@suse.cz>
1745 * doc/extend.texi: Add missing values for __builtin_cpu_is and
1746 __builtin_cpu_supports for x86 target.
1748 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
1751 * doc/rtl.texi (Constants): Clarify that mode of "high" doesn't
1753 (Arithmetic): Likewise for "lo_sum".
1755 2018-11-12 Renlin Li <renlin.li@arm.com>
1758 * dse.c (get_stored_val): Add check for compile-time constantness
1761 2018-11-12 Sudakshina Das <sudi.das@arm.com>
1763 * config/arm/arm-cpus.in (armv8_5, sb, predres): New features.
1764 (ARMv8_5a): New fgroup.
1765 (armv8.5-a): New arch.
1766 (armv8-a, armv8.1-a, armv8.2-a, armv8.3-a, armv8.4-a): New options
1768 * config/arm/arm-tables.opt: Regenerate.
1769 * config/arm/t-aprofile: Add matching rules for -march=armv8.5-a
1770 * config/arm/t-arm-elf (all_v8_archs): Add armv8.5-a.
1771 * config/arm/t-multilib (v8_5_a_simd_variants): New variable.
1772 Add matching rules for -march=armv8.5-a and extensions.
1773 * doc/invoke.texi (ARM options): Document -march=armv8.5-a.
1774 Add sb and predres to all armv8-a except armv8.5-a.
1776 2018-11-12 Richard Biener <rguenther@suse.de>
1778 * tree-vrp.h (value_range[_base]::set): Make public. Provide
1779 overload for single value.
1780 (value_range[_base]::set_nonnull): New.
1781 (value_range[_base]::set_null): Likewise.
1782 (value_range): Document bitmap copying behavior, mark
1783 copy constructor and assignment operator deleted.
1784 (value_range::move): New.
1785 (value_range::set_and_canonicalize): Default bitmap to zero.
1786 (set_value_range_to_nonnull): Remove.
1787 (set_value_range_to_null): Likewise.
1788 (set_value_range): Likewise.
1789 (set_value_range_to_value): Likewise.
1790 (extract_range_from_unary_expr): Work on value_range_base.
1791 (extract_range_from_binary_expr_1): Likewise. Rename to...
1792 (extract_range_from_binary_expr): ... this.
1793 * tree-vrp.c (value_range::update): Clear equiv bitmap
1795 (value_range::move): New, move equiv bitmap.
1796 (value_range_base::set_undefined): Avoid assignment.
1797 (value_range::set_undefined): Likewise.
1798 (value_range_base::set_varying): Likewise.
1799 (value_range::set_varying): Likewise.
1800 (set_value_range): Remove.
1801 (value_range_base::set): New overload for value.
1802 (value_range::set): Likewise.
1803 (set_value_range_to_nonnull): Remove.
1804 (value_range_base::set_nonnull): New.
1805 (value_range::set_nonnull): Likewise.
1806 (set_value_range_to_null): Remove.
1807 (value_range_base::set_null): New.
1808 (value_range::set_null): Likewise.
1809 (range_is_null): Work on value_range_base.
1810 (range_is_nonnull): Likewise.
1811 (ranges_from_anti_range): Likewise.
1812 (extract_range_into_wide_ints): Likewise.
1813 (extract_range_from_multiplicative_op): Likewise.
1814 (extract_range_from_binary_expr): Likewise. Update for API changes.
1815 (extract_range_from_unary_expr): Likewise. Remove OBJ_TYPE_REF
1817 (value_range::intersect_helper): Avoid copy and assignment.
1818 (value_range::union_helper): Likewise.
1819 (determine_value_range_1): Adjust.
1820 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
1821 Avoid assignment by using move.
1822 (evrp_range_analyzer::record_ranges_from_stmt): Avoid assignment.
1823 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
1825 * tree-ssanames.c (get_range_info): Likewise.
1826 * vr-values.h (vr_values::get_vr_for_comparison): Adjust API.
1827 * vr-values.c (vr_values::get_value_range): Adjust.
1828 (vr_values::update_value_range): Likewise.
1829 (symbolic_range_based_on_p): Work on value_range_base.
1830 (vr_values::extract_range_from_binary_expr): Use value_range_base.
1831 (vr_values::extract_range_from_unary_expr): Likewise.
1832 (vr_values::extract_range_from_cond_expr): Avoid assignment.
1833 (vr_values::extract_range_from_comparison): Adjust.
1834 (vr_values::check_for_binary_op_overflow): Use value_range_base.
1835 (vr_values::extract_range_basic): Adjust.
1836 (vr_values::adjust_range_with_scev): Likewise.
1837 (vr_values::vrp_visit_assignment_or_call): Likewise.
1838 (vr_values::get_vr_for_comparison): Change API to avoid
1839 assignment and copy construction.
1840 (vr_values::compare_name_with_value): Adjust accordingly.
1841 (vr_values::compare_names): Likewise.
1842 (vr_values::extract_range_from_phi_node): Avoid assignment and
1843 bogus in-place modify of equiv bitmap.
1844 (vr_values::simplify_bit_ops_using_ranges): Use value_range_base.
1845 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Adjust
1846 for extract_range_from_unary_expr API change.
1847 * ipa-cp.c (ipa_vr_operation_and_type_effects): Likewise.
1849 2018-11-12 Eric Botcazou <ebotcazou@adacore.com>
1851 * config/mcore/mcore.h (WORD_REGISTER_OPERATIONS): Remove duplicate.
1853 2018-11-12 Richard Biener <rguenther@suse.de>
1855 * tree-vrp.h (value_range_base::symbolic_p,
1856 value_range_base::constant_p, value_range_base::zero_p,
1857 value_range_base::singleton_p): Move from value_range.
1858 (value_range::dump): Add.
1859 * gimple-ssa-evrp-analyze.c
1860 (evrp_range_analyzer::record_ranges_from_phis): Use set_varying.
1861 * ipa-cp.c (ipcp_vr_lattice::print): Use dump_value_range.
1862 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
1864 * tree-vrp.c (value_range::symbolic_p): Move to value_range_base.
1865 (value_range::constant_p): Likewise.
1866 (value_range::singleton_p): Likewise.
1867 (value_range_base::dump): Add.
1868 (set_value_range_to_undefined): Remove.
1869 (set_value_range_to_varying): Likewise.
1870 (range_int_cst_p): Take value_range_base argument.
1871 (range_int_cst_singleton_p): Likewise.
1872 (value_range_constant_singleton): Likewise.
1873 (vrp_set_zero_nonzero_bits): Likewise.
1874 (extract_range_from_multiplicative_op): Use set_varying.
1875 (extract_range_from_binary_expr_1): Likewise. Use set_undefined.
1876 (extract_range_from_unary_expr): Likewise.
1877 (dump_value_range_base): Change to overload of dump_value_range.
1878 (vrp_prop::vrp_initialize): Use set_varying and set_undefined.
1879 (vrp_prop::visit_stmt): Likewise.
1880 (value_range::intersect_helper): Likewise.
1881 (value_range::union_helper): Likewise.
1882 (determine_value_range_1): Likewise.
1884 2018-11-12 Richard Biener <rguenther@suse.de>
1886 * tree-vrp.c (set_value_range_to_nonnull): Clear equiv.
1887 (set_value_range_to_null): Likewise.
1888 * vr-values.c (vr_values::extract_range_from_comparison):
1889 Clear equiv for constant singleton ranges.
1891 2018-11-12 Wei Xiao <wei3.xiao@intel.com>
1893 * config/i386/sse.md: Combine VFIXUPIMM* patterns
1894 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
1895 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
1896 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Remove.
1897 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
1898 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
1899 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Remove.
1901 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
1904 * doc/extend.texi (Common Type Attributes): For the align type
1905 attribute, copy language about decreasing alignment from the
1906 corresponding variable attribute.
1908 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
1910 * config/csky/csky.md (*fpuv2_nmulsf3_1, *fpuv2_nmuldf3_1): Handle
1913 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
1916 * doc/invoke.texi (C++ Dialect Options): Add warning about mixing
1917 -frtti and -fno-rtti code.
1919 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
1922 * doc/extend.texi (Other Builtins): Document probability associated
1923 with __builtin_expect.
1925 2018-11-11 Uros Bizjak <ubizjak@gmail.com>
1928 * config/i386/i386.h (STACK_BOUNDARY): Use TARGET_64BIT_MS_ABI
1929 instead of (TARGET_64BIT && ix86_abi == MS_ABI).
1930 * config/i386/darwin.h (STACK_BOUNDARY): Ditto.
1931 * config/i386/cygming.h (STACK_BOUNDARY): Remove.
1933 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
1935 * config/csky/csky-linux-elf.h (CC1_SPEC): Support -profile.
1937 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
1939 * config/csky/csky.h (ASM_PREFERRED_EH_DATA_FORMAT): Define.
1941 2018-11-11 Richard Biener <rguenther@suse.de>
1943 * tree-vrp.h (class value_range_base): New base class for
1944 value_range containing all but the m_equiv member.
1945 (dump_value_range_base): Add.
1946 (range_includes_zero_p): Work on value_range_base.
1947 * tree-vrp.c (value_range_base::set): Split out base handling
1949 (value_range::set): this.
1950 (value_range::set_equiv): New.
1951 (value_range_base::value_range_base): New constructors.
1952 (value_range_base::check): Split out base handling from...
1953 (value_range::check): this.
1954 (value_range::equal_p): Refactor in terms of
1955 ignore_equivs_equal_p which is now member of the base.
1956 (value_range_base::set_undefined): New.
1957 (value_range_base::set_varying): Likewise.
1958 (value_range_base::dump):Split out base handling from...
1959 (value_range::dump): this.
1960 (value_range_base::set_and_canonicalize): Split out base handling
1962 (value_range::set_and_canonicalize): this.
1963 (value_range_base::union_): New.
1964 * ipa-prop.h (struct ipa_jump_func): Use value_range_base *
1966 * ipa-cp.c (class ipcp_vr_lattice): Use value_range_base
1967 instead of value_range everywhere.
1968 (ipcp_vr_lattice::print): Use dump_value_range_base.
1969 (ipcp_vr_lattice::meet_with): Adjust.
1970 (ipcp_vr_lattice::meet_with_1): Likewise.
1971 (ipa_vr_operation_and_type_effects): Likewise.
1972 (propagate_vr_across_jump_function): Likewise.
1973 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Likewise.
1974 (ipa_get_value_range): Likewise.
1975 (ipa_set_jfunc_vr): Likewise.
1976 (ipa_compute_jump_functions_for_edge): Likewise.
1978 2018-11-10 Sandra Loosemore <sandra@codesourcery.com>
1981 * doc/invoke.texi (Optimize Options): Add @opindex entries
1982 for the positive forms of -fno-xxx and -mno-xxx options
1983 that were lacking them.
1985 2018-11-10 Segher Boessenkool <segher@kernel.crashing.org>
1987 * combine.c (make_more_copies): Only make an intermediate copy if the
1988 dest of a move is a pseudo.
1990 2018-11-09 Maya Rashish <coypu@sdf.org>
1993 * config/netbsd-elf.h (NETBSD_STARTFILE_SPEC): Use crtbeginS.o for PIE.
1994 (NETBSD_ENDFILE_SPEC): Use crtendS.o for PIE.
1996 2018-11-09 Sandra Loosemore <sandra@codesourcery.com>
2000 * doc/invoke.texi (Optimize Options): Clarify default behavior
2001 for -fno-toplevel-reorder, -fno-defer-pop, and -fno-branch-count-reg.
2003 2018-11-09 Bernd Edlinger <bernd.edlinger@hotmail.de>
2005 PR tree-optimization/87940
2006 * expr.c (string_constant): Don't strip NOPS in subexpressions.
2007 Fold PLUS_EXPR correctly.
2009 2018-11-09 Ilya Leoshkevich <iii@linux.ibm.com>
2012 * config/s390/s390.c (s390_safe_relative_long_p): New function.
2013 (annotate_constant_pool_refs): Skip insns which support
2014 relative addressing.
2015 (annotate_constant_pool_refs_1): New helper function.
2016 (find_constant_pool_ref): Skip insns which support relative
2018 (find_constant_pool_ref_1): New helper function.
2019 (replace_constant_pool_ref): Skip insns which support
2020 relative addressing.
2021 (replace_constant_pool_ref_1): New helper function.
2022 (s390_mainpool_start): Adapt to the new signature.
2023 (s390_mainpool_finish): Likewise.
2024 (s390_chunkify_start): Likewise.
2025 (s390_chunkify_finish): Likewise.
2026 (pass_s390_early_mach::execute): Likewise.
2027 (s390_prologue_plus_offset): Likewise.
2028 (s390_emit_prologue): Likewise.
2029 (s390_emit_epilogue): Likewise.
2031 2018-11-09 Jakub Jelinek <jakub@redhat.com>
2033 * gimplify.c (gimplify_scan_omp_clauses): Call sorry_at for valid
2034 but unsupported lastprivate with conditional modifier.
2036 2018-11-09 Jeff Law <law@redhat.com>
2038 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): Handle
2039 unused argument better. Add gcc_unreachable to silence warning.
2041 2018-11-09 Martin Sebor <msebor@redhat.com>
2044 * attribs.c (has_attribute): New helper function.
2045 (decls_mismatched_attributes, maybe_diag_alias_attributes): Same.
2046 * attribs.h (decls_mismatched_attributes): Declare.
2047 * cgraphunit.c (handle_alias_pairs): Call maybe_diag_alias_attributes.
2048 (maybe_diag_incompatible_alias): Use OPT_Wattribute_alias_.
2049 * common.opt (-Wattribute-alias): Take an argument.
2050 (-Wno-attribute-alias): New option.
2051 * doc/extend.texi (Common Function Attributes): Document copy.
2052 (Common Variable Attributes): Same.
2053 * doc/invoke.texi (-Wmissing-attributes): Document enhancement.
2054 (-Wattribute-alias): Document new option argument.
2056 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
2058 * config/arm/parsecpu.awk (/alias/): Tighten invisible alias
2059 matching criteria. Remove unused array initializer.
2061 2018-11-09 Bill Schmidt <wschmidt@linux.ibm.com>
2062 Jinsong Ji <jji@us.ibm.com>
2064 * config/rs6000/xmmintrin.h (_mm_cvtss_si32): Fix incorrect
2065 constraints by introducing a new temporary.
2066 (_mm_cvtss_si64): Likewise.
2068 2018-11-09 Martin Liska <mliska@suse.cz>
2070 * common.opt: Add -fipa-stack-alignment flag.
2071 * doc/invoke.texi: Document it.
2072 * final.c (rest_of_clean_state): Guard stack
2073 shrinking with flag.
2075 2018-11-09 Martin Liska <mliska@suse.cz>
2077 * cgraph.h (ipa_discover_readonly_nonaddressable_vars): Rename
2079 (ipa_discover_variable_flags): ... this.
2080 * common.opt: Come up with new flag -fipa-reference-addressable.
2081 * doc/invoke.texi: Document it.
2082 * ipa-reference.c (propagate): Call the renamed fn.
2083 * ipa-visibility.c (whole_program_function_and_variable_visibility):
2085 * ipa.c (ipa_discover_readonly_nonaddressable_vars): Renamed to
2087 (ipa_discover_variable_flags): ... this. Discover
2088 non-addressable variables only with the newly added flag.
2089 * opts.c: Enable the newly added flag with -O1 and higher
2092 2018-11-09 David Malcolm <dmalcolm@redhat.com>
2094 * json.cc (selftest::test_writing_literals): Fix comment.
2096 2018-11-09 Martin Liska <mliska@suse.cz>
2098 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not copy
2099 string to a stack buffer.
2100 (aarch64_parse_cpu): Likewise.
2101 (aarch64_parse_tune): Likewise.
2103 2018-11-09 Richard Biener <rguenther@suse.de>
2105 PR tree-optimization/87953
2106 * tree-vect-loop.c (vectorizable_reduction): For analysis
2107 always pass ops[0] to vectorizable_condition.
2109 2018-11-09 Stafford Horne <shorne@gmail.com>
2110 Richard Henderson <rth@twiddle.net>
2111 Joel Sherrill <joel@rtems.org>
2113 * common/config/or1k/or1k-common.c: New file.
2114 * config/or1k/*: New.
2115 * config.gcc (or1k*-*-*): New.
2116 * configure.ac (or1k*-*-*): New test for openrisc tls.
2117 * configure: Regenerated.
2118 * doc/install.texi: Document OpenRISC triplets.
2119 * doc/invoke.texi: Document OpenRISC arguments.
2120 * doc/md.texi: Document OpenRISC.
2122 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
2124 * config/arm/arm-cpus.in (arm7tdmi): Add an alias for arm7tdmi-s.
2125 (arm7tdmi-s): Delete CPU.
2126 (arm710t): Add aliases for arm720t and arm740t.
2127 (arm720t, arm740t): Delete CPUs.
2128 (arm920t): Add aliases for arm920, arm922t and arm940t.
2129 (arm920, arm922t, arm940t): Delete CPUs.
2130 (arm10tdmi): Add alias for arm1020t.
2131 (arm1020t): Delete CPU.
2132 (arm9e): Add aliases for arm946e-s, arm966e-s and arm968e-s.
2133 (arm946e-s, arm966e-s, arm968e-s): Delete CPUs.
2134 (arm10e): Add aliases for arm1020e and arm1022e.
2135 (arm1020e, arm1022e): Delete CPU.
2136 * config/arm/arm.md (generic_sched): Remove entries that are now
2138 (generic_vfp): Likewise.
2139 * config/arm/arm1020e.md: Simplify tuning selection based on alias
2141 * config/arm/arm-tune.md: Regenerated.
2142 * config/arm/arm-tables.opt: Regenerated.
2144 2018-11-09 Richard Biener <rguenther@suse.de>
2146 PR tree-optimization/87621
2147 * tree-vect-loop.c (vectorizable_reduction): Handle reduction
2148 op with only phi inputs.
2149 * tree-ssa-loop-ch.c: Include tree-ssa-sccvn.h.
2150 (ch_base::copy_headers): Run CSE on copied loop headers.
2151 (pass_ch_vect::process_loop_p): Simplify.
2153 2018-11-09 Alexandre Oliva <oliva@adacore.com>
2155 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Adjust
2156 for non-w64 x86_64 biarch.
2158 2018-11-09 Alexandre Oliva <aoliva@redhat.com>
2160 PR rtl-optimization/86438
2161 * compare-elim.c (try_eliminate_compare): Use SET_SRC instead
2162 of in_b for the compare if in_b is SET_DEST.
2165 * config/i386/i386.c (ix86_const_not_ok_for_debug_p): Reject
2166 non-toplevel UNSPEC.
2168 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
2170 * tree-vrp.c (value_range::check): Do not access internals
2172 (value_range::singleton_p): Same.
2173 (value_range::type): Same.
2174 (vrp_finalize): Use value_range API.
2176 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
2178 * tree-vrp.c (may_contain_p): Do not access m_min/m_max directly.
2180 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
2182 * vr-values.c (vr_values::get_value_range): Use value_range API
2183 instead of piecing together ranges.
2184 (vr_values::update_value_range): Same.
2186 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
2188 * gimple-fold.c (size_must_be_zero_p): Use value_range API instead
2189 of performing ad-hoc calculations.
2190 * tree-ssanames.c (set_range_info): New overloaded function
2191 accepting value_range &.
2192 (get_range_info): Same.
2193 * tree-ssanames.h (set_range_info_raw): Remove.
2194 (set_range_info): New prototype.
2195 (get_range_info): Same.
2196 * tree-vrp.h (value_range::null_p): Rename to zero_p.
2197 * tree-vrp.c (value_range::null_p): Same.
2199 2018-11-09 Jan Hubicka <jh@suse.cz>
2201 * tree.c (fld_type_variant_equal_p): Test user align flag.
2202 (flt_type_variant): Copy user align flag.
2203 (fld_incomplete_type_of): Clear it.
2205 2018-11-09 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2207 * config/arm/neon.md (div<mode>3): New pattern.
2209 2018-11-08 Andi Kleen <ak@linux.intel.com>
2211 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PTWRITE_SET): New.
2212 (OPTION_MASK_ISA_PTWRITE_UNSET): New.
2213 (ix86_handle_option): Handle OPT_mptwrite.
2214 * config/i386/cpuid.h (bit_PTWRITE): Add.
2215 * config/i386/driver-i386.c (host_detect_local_cpu): Detect ptwrite.
2216 * config/i386/i386-builtin.def (BDESC): Add ptwrite32/64.
2217 * config/i386/i386-c.c (ix86_target_macros_internal): Define __PTWRITE__.
2218 * config/i386/i386.c (ix86_target_string): Handle ptwrite.
2219 (ix86_option_override_internal): Handle PTA_PTWRITE.
2220 (ix86_valid_target_attribute_inner_p): Define ptwrite.
2221 (def_builtin2): Force UINT64 to be 64bit only.
2222 * config/i386/i386.h (TARGET_PTWRITE): Add.
2223 (TARGET_PTWRITE_P): Add.
2225 * config/i386/i386.md: Define ptwrite.
2226 * config/i386/i386.opt: Add -mptwrite.
2227 * config/i386/immintrin.h (_ptwrite64): Add.
2229 * doc/extend.texi: Document __builtin_ia32_ptwrite*.
2230 * doc/invoke.texi: Document -mptwrite.
2232 2018-11-08 Peter Bergner <bergner@linux.ibm.com>
2234 PR rtl-optimization/87600
2235 * cfgexpand.c (expand_asm_stmt): Catch illegal asm constraint usage.
2236 * lra-constraints.c (process_alt_operands): Skip illegal hard
2237 register usage. Prefer reloading non hard register operands.
2239 2018-11-08 Sandra Loosemore <sandra@codesourcery.com>
2242 * doc/invoke.texi (Optimize Options): Clarify default behavior
2243 for -fno-sched-interblock and -fno-sched-spec.
2245 2018-11-08 Roman Geissler <roman.geissler@amadeus.com>
2247 * collect2.c (linker_select): Add USE_LLD_LD.
2248 (ld_suffixes): Add ld.lld.
2249 (main): Handle -fuse-ld=lld.
2250 * common.opt (-fuse-ld=lld): New option.
2251 * doc/invoke.texi (-fuse-ld=lld): Document.
2252 * opts.c (common_handle_option): Handle OPT_fuse_ld_lld.
2254 2018-11-08 Paul Koning <ni1d@arrl.net>
2256 * config/pdp11/constraints.md: Add "Z" series constraints for use
2257 with pre-dec and post-inc addressing.
2258 * config/pdp11/pdp11-protos.m (expand_block_move): Delete.
2259 (pdp11_expand_operands): Add int argument (word count).
2260 (pdp11_sp_frame_offset): Delete.
2261 (pdp11_cmp_length): New function.
2262 (pushpop_regeq): New function.
2263 * config/pdp11/pdp11.c (TARGET_STACK_PROTECT_RUNTIME_ENABLED_P):
2265 (pdp11_expand_prologue, pdp11_expand_epilogue): Rewrite for new
2267 (pdp11_initial_elimination_offset): Ditto.
2268 (pdp11_expand_operands): Add word count argument. Bugfixes.
2269 (output_move_multiple): Change how pointer adjustment is done.
2270 (pdp11_gen_int_label): Correct format.
2271 (output_ascii): Ditto.
2272 (pdp11_asm_output_var): Add code for DEC assembler case.
2273 (pdp11_asm_print_operand): Bugfix for CONST_DOUBLE holding integer
2275 (legitimate_const_double_p): Ditto.
2276 (pdp11_register_move_cost): Adjust for new register classes.
2277 (pdp11_regno_reg_class): Ditto.
2278 (expand_block_move): Delete.
2279 (pushpop_regeq): New function.
2280 (pdp11_legitimate_address_p): Bugfix in check for constant
2282 (pdp11_sp_frame_offset): Delete.
2283 (pdp11_reg_save_size): New helper function for new frame layout.
2284 (output_addr_const_pdp11): Remove CONST_DOUBLE case.
2285 (pdp11_expand_shift): Bugfix in check for constant shift count.
2286 (pdp11_shift_length): Ditto.
2287 (pdp11_assemble_shift): Copy input to pdp11_expand_operands.
2288 (pdp11_cmp_length): New function.
2289 * config/pdp11/pdp11.h (TARGET_CPU_CPP_BUILTINS): Add macros for
2290 some compile options.
2291 (FIXED_REGISTERS): Remove HARD_FRAME_POINTER_REGNUM.
2292 (CALL_USED_REGISTERS): Ditto.
2293 (ELIMINABLE_REGS): Ditto.
2294 (REGISTER_NAMES): Ditto.
2295 (reg_class): Add classes NOTR0_REG through NOTSP_REG for use by Z
2297 (REG_CLASS_NAMES): Ditto.
2298 (REG_CLASS_CONTENTS): Ditto. Also remove
2299 HARD_FRAME_POINTER_REGNUM.
2300 (CPU_REG_CLASS): New macro.
2301 (CLASS_MAX_NREGS): Adjust for new register classes.
2302 (FUNCTION_PROFILER): Make no-op.
2303 (may_call_alloca): Remove unused declaration.
2304 (ASM_OUTPUT_ALIGN): Add workaround for PR87795.
2305 (ASM_OUTPUT_SKIP): Fix format.
2306 * config/pdp11/pdp11.md (unspecv): Add UNSPECV_MOVMEM.
2307 (HARD_FRAME_POINTER_REGNUM): Remove.
2309 (*rts): Rename. Remove epilogue related checks.
2310 (cmpsi, cmpdi): New insn.
2311 (cbranch<mode>4): Change to apply to SI and DI modes as well.
2312 (mov<mode>): Change constraints to enforce that push/pop
2313 destination cannot use the same register as source.
2314 (*mov<mode><cc_cc>): Ditto.
2315 (movmemhi, movmemhi1, movmemhi_nocc): Change to expand block move
2316 at assembly output rather than as RTL expander.
2317 (zero_extendqihi2): Bugfix in check for same registers.
2318 (adddi3_nocc): Bugfix in check for constant operand.
2319 (addsi3_nocc): Ditto.
2320 (subdi3_nocc): Ditto.
2321 (subsi3_nocc): Ditto.
2322 (negdi2_nocc): Copy input to pdp11_expand_operands.
2323 (negsi2_nocc): Ditto.
2324 (bswap2_nocc): Ditto.
2325 * config/pdp11/pdp11.opt (mlra): Fix documentation.
2326 * config/pdp11/t-pdp11: Use -Os.
2328 2018-11-08 Richard Earnshaw <rearnsha@arm.com>
2330 * config/arm/parsecpu.awk (/alias/): New parsing rule.
2331 (/begin cpu/): Check that the cpu name hasn't been previously defined.
2332 (gen_comm_data): Print out CPU alias tables.
2333 (check_cpu): Match aliases when checking the CPU name.
2334 * config/arm/arm-protos.h (cpu_alias): New structure.
2335 (cpu_option): Add entry for aliases.
2336 * config/arm/arm-cpus.in (strongarm): Add aliases for strongarm110
2337 strongarm1100 and strongarm1110.
2338 (strongarm110, strongarm1100, strongarm1110): Delete CPU entries.
2339 (config/arm/arm-generic.md): Remove redundant references to
2340 strongarm110, strongarm1100 and strongarm1110.
2341 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
2342 Scan aliases for additional hints.
2343 (arm_parse_cpu_option_name): Also match a cpu name against the list
2345 * config/arm/arm-tables.opt: Regenerated.
2346 * config/arm/arm-tune.md: Regenerated.
2348 2018-11-08 Jakub Jelinek <jakub@redhat.com>
2350 * builtin-types.def (BT_FN_VOID_BOOL, BT_FN_VOID_SIZE_SIZE_PTR,
2351 BT_FN_UINT_UINT_PTR_PTR, BT_FN_UINT_OMPFN_PTR_UINT_UINT,
2352 BT_FN_BOOL_UINT_LONGPTR_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
2353 BT_FN_BOOL_UINT_ULLPTR_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR,
2354 BT_FN_BOOL_LONG_LONG_LONG_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
2355 BT_FN_BOOL_BOOL_ULL_ULL_ULL_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR): New.
2356 * gengtype.c (open_base_files): Add omp-general.h.
2357 * gimple.c (gimple_build_omp_critical):
2358 (gimple_build_omp_taskgroup): Add CLAUSES argument. Call
2359 gimple_omp_taskgroup_set_clauses.
2360 (gimple_build_omp_atomic_load): Add mo argument, call
2361 gimple_omp_atomic_set_memory_order.
2362 (gimple_build_omp_atomic_store): Likewise.
2363 (gimple_copy): Adjust handling of GIMPLE_OMP_TASKGROUP.
2364 * gimple.def (GIMPLE_OMP_TASKGROUP): Use GSS_OMP_SINGLE_LAYOUT
2366 (GIMPLE_OMP_TEAMS): Use GSS_OMP_PARALLEL_LAYOUT instead
2367 of GSS_OMP_SINGLE_LAYOUT, adjust comments.
2368 * gimple.h (enum gf_mask): Add GF_OMP_TEAMS_HOST, GF_OMP_TASK_TASKWAIT
2369 and GF_OMP_ATOMIC_MEMORY_ORDER. Remove GF_OMP_ATOMIC_SEQ_CST, use
2370 different value for GF_OMP_ATOMIC_NEED_VALUE.
2371 (struct gimple_statement_omp_taskreg): Add GIMPLE_OMP_TEAMS to
2373 (struct gimple_statement_omp_single_layout): And remove here.
2374 (struct gomp_teams): Inherit from gimple_statement_omp_taskreg rather
2375 than gimple_statement_omp_single_layout.
2376 (is_a_helper <gimple_statement_omp_taskreg *>::test): Allow
2378 (is_a_helper <const gimple_statement_omp_taskreg *>::test): Likewise.
2379 (gimple_omp_subcode): Formatting fix.
2380 (gimple_omp_teams_child_fn, gimple_omp_teams_child_fn_ptr,
2381 gimple_omp_teams_set_child_fn, gimple_omp_teams_data_arg,
2382 gimple_omp_teams_data_arg_ptr, gimple_omp_teams_set_data_arg,
2383 gimple_omp_teams_host, gimple_omp_teams_set_host,
2384 gimple_omp_task_taskwait_p, gimple_omp_task_set_taskwait_p,
2385 gimple_omp_taskgroup_clauses, gimple_omp_taskgroup_clauses_ptr,
2386 gimple_omp_taskgroup_set_clauses): New inline functions.
2387 (gimple_build_omp_atomic_load): Add enum omp_memory_order argument.
2388 (gimple_build_omp_atomic_store): Likewise.
2389 (gimple_omp_atomic_seq_cst_p): Remove.
2390 (gimple_omp_atomic_memory_order): New function.
2391 (gimple_omp_atomic_set_seq_cst): Remove.
2392 (gimple_omp_atomic_set_memory_order): New function.
2393 (gimple_build_omp_taskgroup): Add clauses argument.
2394 * gimple-pretty-print.c (dump_gimple_omp_taskgroup): New function.
2395 (dump_gimple_omp_task): Print taskwait with depend clauses.
2396 (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Use
2397 dump_omp_atomic_memory_order.
2398 (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP.
2399 * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP_ALLOC_ONLY,
2400 GOVD_MAP_FROM_ONLY and GOVD_NONTEMPORAL.
2401 (enum omp_region_type): Reserve bits 1 and 2 for auxiliary flags,
2402 renumber values of most of ORT_* enumerators, add ORT_HOST_TEAMS,
2403 ORT_COMBINED_HOST_TEAMS, ORT_TASKGROUP, ORT_TASKLOOP and
2404 ORT_UNTIED_TASKLOOP enumerators.
2405 (enum gimplify_defaultmap_kind): New.
2406 (struct gimplify_omp_ctx): Remove target_map_scalars_firstprivate and
2407 target_map_pointers_as_0len_arrays members, add defaultmap.
2408 (new_omp_context): Initialize defaultmap member.
2409 (gimple_add_tmp_var): Handle ORT_TASKGROUP like ORT_WORKSHARE.
2410 (maybe_fold_stmt): Don't fold even in host teams regions.
2411 (omp_firstprivatize_variable): Handle ORT_TASKGROUP like
2412 ORT_WORKSHARE. Test ctx->defaultmap[GDMK_SCALAR] instead of
2413 ctx->omp_firstprivatize_variable.
2414 (omp_add_variable): Don't add private/firstprivate for VLAs in
2416 (omp_default_clause): Print "taskloop" rather than "task" if
2418 (omp_notice_variable): Handle ORT_TASKGROUP like ORT_WORKSHARE.
2419 Handle new defaultmap clause kinds.
2420 (omp_is_private): Handle ORT_TASKGROUP like ORT_WORKSHARE. Allow simd
2421 iterator to be lastprivate or private. Fix up diagnostics if linear
2422 is used on collapse>1 simd iterator.
2423 (omp_check_private): Handle ORT_TASKGROUP like ORT_WORKSHARE.
2424 (gimplify_omp_depend): New function.
2425 (gimplify_scan_omp_clauses): Add shared clause on parallel for
2426 combined parallel master taskloop{, simd} if taskloop has
2427 firstprivate, lastprivate or reduction clause. Handle
2428 OMP_CLAUSE_REDUCTION_TASK diagnostics. Adjust tests for
2429 ORT_COMBINED_TEAMS. Gimplify depend clauses with iterators. Handle
2430 cancel and simd OMP_CLAUSE_IF_MODIFIERs. Handle
2431 OMP_CLAUSE_NONTEMPORAL. Handle new defaultmap clause kinds. Handle
2432 OMP_CLAUSE_{TASK,IN}_REDUCTION. Diagnose invalid conditional
2434 (gimplify_adjust_omp_clauses_1): Ignore GOVD_NONTEMPORAL. Handle
2435 GOVD_MAP_ALLOC_ONLY and GOVD_MAP_FROM_ONLY.
2436 (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_NONTEMPORAL. Handle
2437 OMP_CLAUSE_{TASK,IN}_REDUCTION.
2438 (gimplify_omp_task): Handle taskwait with depend clauses.
2439 (gimplify_omp_for): Add shared clause on parallel for combined
2440 parallel master taskloop{, simd} if taskloop has firstprivate,
2441 lastprivate or reduction clause. Use ORT_TASKLOOP or
2442 ORT_UNTIED_TASKLOOP instead of ORT_TASK or ORT_UNTIED_TASK. Adjust
2443 tests for ORT_COMBINED_TEAMS. Handle C++ range for loops with
2444 NULL TREE_PURPOSE in OMP_FOR_ORIG_DECLS. Firstprivatize
2445 __for_end and __for_range temporaries on OMP_PARALLEL for
2446 distribute parallel for{, simd}. Move OMP_CLAUSE_REDUCTION
2447 and OMP_CLAUSE_IN_REDUCTION from taskloop to the task construct
2448 sandwiched in between two taskloops.
2449 (computable_teams_clause): Test ctx->defaultmap[GDMK_SCALAR]
2450 instead of ctx->omp_firstprivatize_variable.
2451 (gimplify_omp_workshare): Set ort to ORT_HOST_TEAMS or
2452 ORT_COMBINED_HOST_TEAMS if not inside of target construct. If
2453 host teams, use gimplify_and_return_first etc. for body like
2454 for target or target data constructs, and at the end call
2455 gimple_omp_teams_set_host on the GIMPLE_OMP_TEAMS object.
2456 (gimplify_omp_atomic): Use OMP_ATOMIC_MEMORY_ORDER instead
2457 of OMP_ATOMIC_SEQ_CST, pass it as new argument to
2458 gimple_build_omp_atomic_load and gimple_build_omp_atomic_store, remove
2459 gimple_omp_atomic_set_seq_cst calls.
2460 (gimplify_expr) <case OMP_TASKGROUP>: Move handling into a separate
2461 case, handle taskgroup clauses.
2462 * lto-streamer-out.c (hash_tree): Handle
2463 OMP_CLAUSE_{TASK,IN}_REDUCTION.
2464 * Makefile.in (GTFILES): Add omp-general.h.
2465 * omp-builtins.def (BUILT_IN_GOMP_TASKWAIT_DEPEND,
2466 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_START,
2467 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_START,
2468 BUILT_IN_GOMP_LOOP_START, BUILT_IN_GOMP_LOOP_ORDERED_START,
2469 BUILT_IN_GOMP_LOOP_DOACROSS_START,
2470 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_NEXT,
2471 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
2472 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_START,
2473 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_START,
2474 BUILT_IN_GOMP_LOOP_ULL_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_START,
2475 BUILT_IN_GOMP_LOOP_ULL_DOACROSS_START,
2476 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_NEXT,
2477 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
2478 BUILT_IN_GOMP_PARALLEL_LOOP_NONMONOTONIC_RUNTIME,
2479 BUILT_IN_GOMP_PARALLEL_LOOP_MAYBE_NONMONOTONIC_RUNTIME,
2480 BUILT_IN_GOMP_PARALLEL_REDUCTIONS, BUILT_IN_GOMP_SECTIONS2_START,
2481 BUILT_IN_GOMP_TEAMS_REG, BUILT_IN_GOMP_TASKGROUP_REDUCTION_REGISTER,
2482 BUILT_IN_GOMP_TASKGROUP_REDUCTION_UNREGISTER,
2483 BUILT_IN_GOMP_TASK_REDUCTION_REMAP,
2484 BUILT_IN_GOMP_WORKSHARE_TASK_REDUCTION_UNREGISTER): New builtins.
2485 * omp-expand.c (workshare_safe_to_combine_p): Return false for
2486 non-worksharing loops.
2487 (omp_adjust_chunk_size): Don't adjust anything if chunk_size is zero.
2488 (determine_parallel_type): Don't combine parallel with worksharing
2489 which has _reductemp_ clause.
2490 (expand_parallel_call): Emit the GOMP_*nonmonotonic_runtime* or
2491 GOMP_*maybe_nonmonotonic_runtime* builtins instead of GOMP_*runtime*
2492 if there is nonmonotonic modifier or if there is no modifier and no
2493 ordered clause. For dynamic and guided schedule without monotonic
2494 and nonmonotonic modifier, default to nonmonotonic.
2495 (expand_omp_for): Likewise. Adjust expand_omp_for_generic caller, use
2496 GOMP_loop{,_ull}{,_ordered,_doacross}_start builtins if there are
2498 (expand_task_call): Add GOMP_TASK_FLAG_REDUCTION flag to flags if
2499 there are any reduction clauses.
2500 (expand_taskwait_call): New function.
2501 (expand_teams_call): New function.
2502 (expand_omp_taskreg): Allow GIMPLE_OMP_TEAMS and call
2503 expand_teams_call for it. Formatting fix. Handle taskwait with
2505 (expand_omp_for_generic): Add SCHED_ARG argument. Handle expansion
2506 of worksharing loops with task reductions.
2507 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk): Handle
2508 expansion of worksharing loops with task reductions.
2509 (expand_omp_sections): Handle expansion of sections with task
2511 (expand_omp_synch): For host teams call expand_omp_taskreg.
2512 (omp_memory_order_to_memmodel): New function.
2513 (expand_omp_atomic_load, expand_omp_atomic_store,
2514 expand_omp_atomic_fetch_op): Use it and gimple_omp_atomic_memory_order
2515 instead of gimple_omp_atomic_seq_cst_p.
2516 (build_omp_regions_1, omp_make_gimple_edges): Treat taskwait with
2517 depend clauses as a standalone directive.
2518 * omp-general.c (enum omp_requires): New variable.
2519 (omp_extract_for_data): Initialize have_reductemp member. Allow
2520 NE_EXPR even in OpenMP loops, transform them into LT_EXPR or
2521 GT_EXPR loops depending on incr sign. Formatting fixes.
2522 * omp-general.h (struct omp_for_data): Add have_reductemp member.
2523 (enum omp_requires): New enum.
2524 (omp_requires_mask): Declare.
2525 * omp-grid.c (grid_eliminate_combined_simd_part): Formatting fix.
2527 * omp-low.c (struct omp_context): Add task_reductions and
2528 task_reduction_map fields.
2529 (is_host_teams_ctx): New function.
2530 (is_taskreg_ctx): Return true also if is_host_teams_ctx.
2531 (use_pointer_for_field): Use is_global_var instead of
2532 TREE_STATIC || DECL_EXTERNAL, and apply only if not privatized
2534 (build_outer_var_ref): Ignore taskgroup outer contexts.
2535 (delete_omp_context): Release task_reductions and task_reduction_map.
2536 (scan_sharing_clauses): Don't add any fields for reduction clause on
2537 taskloop. Handle OMP_CLAUSE__REDUCTEMP_. Handle
2538 OMP_CLAUSE_{IN,TASK}_REDUCTION and OMP_CLAUSE_REDUCTION with task
2539 modifier. Don't ignore shared clauses in is_host_teams_ctx contexts.
2540 Handle OMP_CLAUSE_NONTEMPORAL.
2541 (add_taskreg_looptemp_clauses): Add OMP_CLAUSE__REDUCTEMP_ clause if
2543 (scan_omp_parallel): Add _reductemp_ clause if there are any reduction
2544 clauses with task modifier.
2545 (scan_omp_task): Handle taskwait with depend clauses.
2546 (finish_taskreg_scan): Move field corresponding to _reductemp_ clause
2547 first. Move also OMP_CLAUSE__REDUCTEMP_ clause in front if present.
2548 Handle GIMPLE_OMP_TEAMS like GIMPLE_OMP_PARALLEL.
2549 (scan_omp_for): Fix comment formatting.
2550 (scan_omp_teams): Handle host teams constructs.
2551 (check_omp_nesting_restrictions): Allow teams with no outer
2552 OpenMP context. Adjust diagnostics for teams strictly nested into
2553 some explicit OpenMP construct other than target. Allow OpenMP atomics
2554 inside of simd regions.
2555 (scan_omp_1_stmt): Call scan_sharing_clauses for taskgroups.
2556 (scan_omp_1_stmt) <case GIMPLE_OMP_TEAMS>: Temporarily bump
2557 taskreg_nesting_level while scanning host teams construct.
2558 (task_reduction_read): New function.
2559 (lower_rec_input_clauses): Handle OMP_CLAUSE_REDUCTION on taskloop
2560 construct. Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE__REDUCTEMP_
2561 clauses. Handle OMP_CLAUSE_REDUCTION with task modifier. Remove
2562 second argument create_tmp_var if it is NULL. Don't ignore shared
2563 clauses in is_host_teams_ctx contexts. Handle
2564 OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE on OMP_CLAUSE_FIRSTPRIVATE
2566 (lower_reduction_clauses): Ignore reduction clauses with task
2567 modifier. Remove second argument create_tmp_var if it is NULL.
2568 Initialize OMP_ATOMIC_MEMORY_ORDER to relaxed.
2569 (lower_send_clauses): Ignore reduction clauses with task modifier.
2570 Handle OMP_CLAUSE__REDUCTEMP_. Don't send anything for
2571 OMP_CLAUSE_REDUCTION on taskloop. Handle OMP_CLAUSE_IN_REDUCTION.
2572 (maybe_add_implicit_barrier_cancel): Add OMP_RETURN argument, don't
2573 rely that it is the last stmt in body so far. Ignore outer taskgroup
2575 (omp_task_reductions_find_first, omp_task_reduction_iterate,
2576 lower_omp_task_reductions): New functions.
2577 (lower_omp_sections): Handle reduction clauses with taskgroup
2578 modifiers. Adjust maybe_add_implicit_barrier_cancel caller.
2579 (lower_omp_single): Adjust maybe_add_implicit_barrier_cancel caller.
2580 (lower_omp_for): Likewise. Handle reduction clauses with taskgroup
2582 (lower_omp_taskgroup): Handle taskgroup reductions.
2583 (create_task_copyfn): Copy over OMP_CLAUSE__REDUCTEMP_ pointer.
2584 Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE_REDUCTION clauses.
2585 (lower_depend_clauses): If there are any
2586 OMP_CLAUSE_DEPEND_DEPOBJ or OMP_CLAUSE_DEPEND_MUTEXINOUTSET
2587 depend clauses, use a new array format. If OMP_CLAUSE_DEPEND_LAST is
2588 seen, assume lowering is done already and return early. Set kind
2589 on artificial depend clause to OMP_CLAUSE_DEPEND_LAST.
2590 (lower_omp_taskreg): Handle reduction clauses with task modifier on
2591 parallel construct. Handle reduction clause on taskloop construct.
2592 Handle taskwait with depend clauses.
2593 (lower_omp_1): Use lower_omp_taskreg instead of lower_omp_teams
2594 for host teams constructs.
2595 * tree.c (omp_clause_num_ops): Add in_reduction, task_reduction,
2596 nontemporal and _reductemp_ clause entries.
2597 (omp_clause_code_name): Likewise.
2598 (walk_tree_1): Handle OMP_CLAUSE_{IN,TASK}_REDUCTION,
2599 OMP_CLAUSE_NONTEMPORAL and OMP_CLAUSE__REDUCTEMP_.
2600 * tree-core.h (enum omp_clause_code): Add
2601 OMP_CLAUSE_{{IN,TASK}_REDUCTION,NONTEMPORAL,_REDUCTEMP_}.
2602 (enum omp_clause_defaultmap_kind, enum omp_memory_order): New.
2603 (struct tree_base): Add omp_atomic_memory_order field into union.
2604 Remove OMP_ATOMIC_SEQ_CST comment.
2605 (enum omp_clause_depend_kind): Add OMP_CLAUSE_DEPEND_MUTEXINOUTSET
2606 and OMP_CLAUSE_DEPEND_DEPOBJ.
2607 (struct tree_omp_clause): Add subcode.defaultmap_kind.
2608 * tree.def (OMP_TASKGROUP): Add another operand, move next to other
2609 OpenMP constructs with body and clauses operands.
2610 * tree.h (OMP_BODY): Use OMP_MASTER instead of OMP_TASKGROUP.
2611 (OMP_CLAUSES): Use OMP_TASKGROUP instead of OMP_SINGLE.
2612 (OMP_TASKGROUP_CLAUSES): Define.
2613 (OMP_CLAUSE_DECL): Use OMP_CLAUSE__REDUCTEMP_ instead of
2614 OMP_CLAUSE__LOOPTEMP_.
2615 (OMP_ATOMIC_SEQ_CST): Remove.
2616 (OMP_ATOMIC_MEMORY_ORDER, OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE,
2617 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL): Define.
2618 (OMP_CLAUSE_REDUCTION_CODE, OMP_CLAUSE_REDUCTION_INIT,
2619 OMP_CLAUSE_REDUCTION_MERGE, OMP_CLAUSE_REDUCTION_PLACEHOLDER,
2620 OMP_CLAUSE_REDUCTION_DECL_PLACEHOLDER,
2621 OMP_CLAUSE_REDUCTION_OMP_ORIG_REF): Handle
2622 OMP_CLAUSE_{,IN_,TASK_}REDUCTION.
2623 (OMP_CLAUSE_REDUCTION_TASK, OMP_CLAUSE_REDUCTION_INSCAN,
2624 OMP_CLAUSE_DEFAULTMAP_KIND, OMP_CLAUSE_DEFAULTMAP_CATEGORY,
2625 OMP_CLAUSE_DEFAULTMAP_BEHAVIOR, OMP_CLAUSE_DEFAULTMAP_SET_KIND):
2627 * tree-inline.c (remap_gimple_stmt): Remap taskgroup clauses.
2628 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
2629 OMP_CLAUSE__REDUCTEMP_, OMP_CLAUSE_NONTEMPORAL.
2630 (convert_local_omp_clauses): Likewise. Remove useless test.
2631 * tree-parloops.c (create_call_for_reduction_1): Pass
2632 OMP_MEMORY_ORDER_RELAXED as new argument to
2633 dump_gimple_omp_atomic_load and dump_gimple_omp_atomic_store.
2634 * tree-pretty-print.c (dump_omp_iterators): New function.
2635 (dump_omp_clause): Handle OMP_CLAUSE__REDUCTEMP_,
2636 OMP_CLAUSE_NONTEMPORAL, OMP_CLAUSE_{TASK,IN}_REDUCTION. Print
2637 reduction modifiers. Handle OMP_CLAUSE_DEPEND_DEPOBJ and
2638 OMP_CLAUSE_DEPEND_MUTEXINOUTSET. Print iterators in depend clauses.
2639 Print __internal__ for OMP_CLAUSE_DEPEND_LAST. Handle cancel and
2640 simd OMP_CLAUSE_IF_MODIFIERs. Handle new kinds of
2641 OMP_CLAUSE_DEFAULTMAP. Print conditional: for
2642 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL.
2643 (dump_omp_atomic_memory_order): New function.
2644 (dump_generic_node): Use it. Print taskgroup clauses. Print
2645 taskwait with depend clauses.
2646 * tree-pretty-print.h (dump_omp_atomic_memory_order): Declare.
2647 * tree-streamer-in.c (unpack_ts_omp_clause_value_fields):
2648 Handle OMP_CLAUSE_{TASK,IN}_REDUCTION.
2649 * tree-streamer-out.c (pack_ts_omp_clause_value_fields,
2650 write_ts_omp_clause_tree_pointers): Likewise.
2652 2018-11-08 David Malcolm <dmalcolm@redhat.com>
2655 * doc/invoke.texi (-fdump-ipa-): Document the "-optimized",
2656 "-missed", "-note", and "-all" sub-options.
2657 * ipa-inline.c (caller_growth_limits): Port from fprintf to dump
2659 (can_early_inline_edge_p): Likewise.
2660 (want_early_inline_function_p): Likewise.
2661 (want_inline_self_recursive_call_p): Likewise.
2662 (recursive_inlining): Likewise.
2663 (inline_small_functions): Likewise.
2664 (flatten_function): Likewise.
2665 (ipa_inline): Likewise.
2666 (inline_always_inline_functions): Likewise.
2667 (early_inline_small_functions): Likewise.
2668 (early_inliner): Likewise.
2669 * tree-inline.c (expand_call_inline): Likewise.
2671 2018-11-08 David Malcolm <dmalcolm@redhat.com>
2673 * pretty-print.c (pp_format): Handle %f.
2674 (selftest::test_pp_format): Add test of %f.
2675 * pretty-print.h (pp_double): New macro.
2677 2018-11-08 David Malcolm <dmalcolm@redhat.com>
2679 * dump-context.h (ASSERT_IS_CGRAPH_NODE): New macro.
2680 * dumpfile.c (make_item_for_dump_cgraph_node): Move to before...
2681 (dump_pretty_printer::decode_format): Implement "%C" for
2683 (selftest::test_capture_of_dump_calls): Rename "where" to
2684 "stmt_loc". Convert test_decl to a function decl and set its
2685 location. Add a symbol_table_test RAII instance and a
2686 cgraph_node, using it to test "%C" and dump_symtab_node.
2688 2018-11-08 Eric Botcazou <ebotcazou@adacore.com>
2691 * cgraphclones.c (duplicate_thunk_for_node): Also set DECL_IGNORED_P.
2693 2018-11-08 David Malcolm <dmalcolm@redhat.com>
2695 * cgraph.c: Include "selftest.h".
2696 (saved_symtab): New variable.
2697 (selftest::symbol_table_test::symbol_table_test): New ctor.
2698 (selftest::symbol_table_test::~symbol_table_test): New dtor.
2699 (selftest::test_symbol_table_test): New test.
2700 (selftest::cgraph_c_tests): New.
2701 * cgraph.h (saved_symtab): New decl.
2702 (selftest::symbol_table_test): New class.
2703 * selftest-run-tests.c (selftest::run_tests): Call
2704 selftest::cgraph_c_tests.
2705 * selftest.h (selftest::cgraph_c_tests): New decl.
2707 2018-11-08 Richard Biener <rguenther@suse.de>
2709 * tree-data-ref.h (lambda_int): New typedef.
2710 (lambda_vector_gcd): Adjust.
2711 (lambda_vector_new): Likewise.
2712 (lambda_matrix_new): Likewise.
2713 * tree-data-ref.c (print_lambda_vector): Adjust.
2715 2018-11-08 Richard Biener <rguenther@suse.de>
2717 PR tree-optimization/87929
2718 * tree-complex.c (expand_complex_comparison): Clean EH.
2720 2018-11-08 Martin Liska <mliska@suse.cz>
2722 * doc/extend.texi: Reword.
2723 * predict.c (expr_expected_value_1): Likewise.
2725 2018-11-08 Richard Biener <rguenther@suse.de>
2727 PR tree-optimization/87913
2728 * tree-ssa-phiopt.c (minmax_replacement): Turn EQ/NE compares
2729 of extreme values to ordered comparisons.
2731 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
2734 * doc/invoke.texi (Code Gen Options): Clarify -fno-common behavior.
2736 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
2738 * doc/invoke.texi: Remove leading dash from @opindex entries
2739 throughout the file.
2741 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
2744 * doc/invoke.texi (Option Summary): Add -e and --entry.
2745 (Link Options): Likewise.
2747 2018-11-07 Nathan Sidwell <nathan@acm.org>
2750 * Makefile.in (bitmap.o-warn): Add -Wno-error to unbreak
2751 --disable-checking bootstrap.
2753 2018-11-07 Hafiz Abid Qadeer <abidh@codesourcery.com>
2755 * configure: Regenerated.
2757 2018-11-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
2760 * stor-layout.c (compute_record_mode): Set TYPE_MODE of UNION_TYPE
2761 to the mode of the widest field iff the widest field has mode class
2762 MODE_INT, or MODE_PARTIAL_INT and the union would be passed by
2765 2018-11-07 Nikolai Merinov <n.merinov@inango-systems.com>
2767 * common.opt: Add -Wattribute-warning.
2768 * doc/invoke.texi: Add documentation for -Wno-attribute-warning.
2769 * expr.c (expand_expr_real_1): Add new attribute to warning_at
2770 call to allow user configure behavior of "warning" attribute.
2772 2018-11-07 Segher Boessenkool <segher@kernel.crashing.org>
2774 * target.def: Put @: after every vs., e.g., and i.e. where it is
2775 followed by whitespace.
2776 * doc/extend.texi: Ditto.
2777 * doc/fragments.texi: Ditto.
2778 * doc/gimple.texi: Ditto.
2779 * doc/implement-c.texi: Ditto.
2780 * doc/install.texi: Ditto.
2781 * doc/invoke.texi: Ditto.
2782 * doc/md.texi: Ditto.
2783 * doc/plugins.texi: Ditto.
2784 * doc/rtl.texi: Ditto.
2785 * doc/sourcebuild.texi: Ditto.
2786 * doc/tm.texi.in: Ditto.
2787 * doc/ux.texi: Ditto.
2788 * doc/tm.texi: Regenerate.
2790 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2792 * config/arm/arm-cpus.in (ares): New entry.
2793 * config/arm/arm-tables.opt: Regenerate.
2794 * config/arm/arm-tune.md: Likewise.
2795 * doc/invoke.texi (ARM Options): Document ares.
2797 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2799 * config/aarch64/aarch64-cores.def (ares): Define.
2800 * config/aarch64/aarch64-tune.md: Regenerate.
2801 * doc/invoke.texi (AArch64 Options): Document ares value for mtune.
2803 2018-11-07 Jan Hubicka <jh@suse.cz>
2805 * tree.c (fld_incomplete_type_of): Clear TREE_ADDRESSABLE.
2806 (free_lang_data_in_decl): Set TREE_ADDRESSABLE for public vars and
2807 functions; clear TYPE_DECL_SUPPRESS_DEBUG and DECL_MODE for
2810 2018-11-07 Richard Biener <rguenther@suse.de>
2812 PR tree-optimization/87914
2813 * tree-vect-loop.c (vect_is_simple_reduction): Improve detection
2815 (vectorizable_reduction): Handle shifts and rotates by dispatching
2816 to vectorizable_shift.
2817 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Handle
2818 in-loop uses of vect_nested_cycle defs. Merge cycle and internal
2820 (vectorizable_shift): Export and handle being called as
2822 (vect_analyze_stmt): Call vectorizable_shift after
2823 vectorizable_reduction.
2824 * tree-vectorizer.h (vectorizable_shift): Declare.
2826 2018-11-07 Jan Hubicka <jh@suse.cz>
2828 * ipa-devirt.c (odr_types_equivalent_p): Expect constants
2829 than const decls in TREE_VALUE of enum.
2830 (dump_type_inheritance_graph): Improve duplicate dumping.
2831 (free_enum_values): New.
2832 (build_type_inheritance_graph): Use it.
2833 * tree.c (free_lang_data_in_type): Free TYPE_VALUES of enums
2834 which are not main variants or not ODR types.
2835 (verify_type_variant): Expect variants to have no TYPE_VALUES.
2837 2018-11-07 Richard Biener <rguenther@suse.de>
2839 * ipa-inline.c (want_inline_small_function_p): Compute
2840 big_speedup_p lazily and last.
2842 2018-11-07 Jan Hubicka <jh@suse.cz>
2844 * tree.c (fld_type_variant_equal_p): Skip TYPE_ALIGN check when
2845 building incomplete variant of complete type.
2846 (fld_type_variant): Do not copy TYPE_ALIGN when building incomplete
2847 variant of complete type.
2849 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2851 * config/mips/mips.c: Fix typo in documentation of
2852 mips_loongson_ext2_prefetch_cookie.
2853 (mips_option_override): fix brain twister logical.
2854 * config/mips/mips.h: Fix typo in documentation of
2855 ISA_HAS_CTZ_CTO and define pattern.
2856 * config/mips/mips.md (prefetch): Hoist EXT2 above
2858 (prefetch_indexed): Hoist EXT2 above the EXT block.
2860 2018-11-07 Jan Hubicka <jh@suse.cz>
2862 * tree.c (free_lang_data_in_type): Add fld parameter; simplify
2863 return and parameter types of function and method types.
2864 (free_lang_data_in_cgraph): Update.
2866 2018-11-07 Martin Liska <mliska@suse.cz>
2868 PR rtl-optimization/87868
2869 * postreload-gcse.c (eliminate_partially_redundant_load): Set
2870 threshold to max_count if we would overflow.
2871 * profile-count.h: Make max_count a public constant.
2873 2018-11-07 Martin Liska <mliska@suse.cz>
2875 * mem-stats.h: Fix GNU coding style.
2877 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2879 * config/mips/gs264e.md: New.
2880 * config/mips/mips-cpus.def: Define gs264e.
2881 * config/mips/mips-tables.opt: Regenerate.
2882 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
2884 (mips_issue_rate): Add support for gs264e.
2885 (mips_multipass_dfa_lookahead): Likewise.
2886 * config/mips/mips.h: Define TARGET_GS264E and TUNE_GS264E.
2887 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs264e.
2888 (MIPS_ASE_MSA_SPEC): New.
2889 (BASE_DRIVER_SELF_SPECS): march=gs264e implies -mmsa.
2890 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS264E.
2891 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS264E.
2892 * config/mips/mips.md: Include gs264e.md.
2893 (processor): Add gs264e.
2894 * config/mips/mips.opt (MSA): Use Mask instead of Var.
2895 * doc/invoke.texi: Add gs264e to supported architectures.
2897 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2899 * config/mips/gs464e.md: New.
2900 * config/mips/mips-cpus.def: Define gs464e.
2901 * config/mips/mips-tables.opt: Regenerate.
2902 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
2904 (mips_issue_rate): Add support for gs464e.
2905 (mips_multipass_dfa_lookahead): Likewise.
2906 (mips_option_override): Enable MMI, EXT and EXT2 for gs464e.
2907 * config/mips/mips.h: Define TARGET_GS464E and TUNE_GS464E.
2908 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464e.
2909 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS464E.
2910 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS464E.
2911 * config/mips/mips.md: Include gs464e.md.
2912 (processor): Add gs464e.
2913 * doc/invoke.texi: Add gs464e to supported architectures.
2915 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2917 * config/mips/loongson3a.md: Rename to ...
2918 * config/mips/gs464.md: ... here.
2919 * config/mips/mips-cpus.def: Define gs464; Add loongson3a
2920 as an alias of gs464 processor.
2921 * config/mips/mips-tables.opt: Regenerate.
2922 * config/mips/mips.c (mips_issue_rate): Use PROCESSOR_GS464
2923 instead of PROCESSOR_LOONGSON_3A.
2924 (mips_multipass_dfa_lookahead): Use TUNE_GS464 instead of
2926 (mips_option_override): Enable MMI and EXT for gs464.
2927 * config/mips/mips.h: Rename TARGET_LOONGSON_3A to TARGET_GS464;
2928 Rename TUNE_LOONGSON_3A to TUNE_GS464.
2929 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464.
2930 (ISA_HAS_ODD_SPREG, ISA_AVOID_DIV_HILO, ISA_HAS_FUSED_MADD4,
2931 ISA_HAS_UNFUSED_MADD4): Use TARGET_GS464 instead of
2933 * config/mips/mips.md: Include gs464.md instead of loongson3a.md.
2934 (processor): Add gs464;
2935 * doc/invoke.texi: Add gs464 to supported architectures.
2937 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2939 * config/mips/mips-protos.h
2940 (mips_loongson_ext2_prefetch_cookie): New prototype.
2941 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): New.
2942 (mips_option_override): Enable TARGET_LOONGSON_EXT when
2943 TARGET_LOONGSON_EXT2 is true.
2944 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define
2945 __mips_loongson_ext2, __mips_loongson_ext_rev=2.
2946 (ISA_HAS_CTZ_CTO): New, true if TARGET_LOONGSON_EXT2.
2947 (ISA_HAS_PREFETCH): Include TARGET_LOONGSON_EXT and
2948 TARGET_LOONGSON_EXT2.
2949 (ASM_SPEC): Add mloongson-ext2 and mno-loongson-ext2.
2950 (define_insn "ctz<mode>2"): New insn pattern.
2951 (define_insn "prefetch"): Include TARGET_LOONGSON_EXT2.
2952 (define_insn "prefetch_indexed_<mode>"): Include
2953 TARGET_LOONGSON_EXT and TARGET_LOONGSON_EXT2.
2954 * config/mips/mips.opt (-mloongson-ext2): Add option.
2955 * gcc/doc/invoke.texi (-mloongson-ext2): Document.
2957 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2959 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Add
2960 __mips_loongson_ext.
2961 (MIPS_ASE_LOONGSON_EXT_SPEC): New.
2962 (BASE_DRIVER_SELF_SPECS): march=loongson3a implies
2964 (ASM_SPEC): Add mloongson-ext and mno-loongson-ext.
2965 * config/mips/mips.md (mul<mode>3, mul<mode>3_mul3_nohilo,
2966 <u>div<mode>3, <u>mod<mode>3, prefetch): Use TARGET_LOONGSON_EXT
2967 instead of TARGET_LOONGSON_3A.
2968 * config/mips/mips.opt (-mloongson-ext): Add option.
2969 * gcc/doc/invoke.texi (-mloongson-ext): Document.
2971 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2973 * config.gcc (extra_headers): Add loongson-mmiintrin.h.
2974 * config/mips/loongson.md: Move to ...
2975 * config/mips/loongson-mmi.md: here; Adjustment.
2976 * config/mips/loongson.h: Move to ...
2977 State as deprecated. Include loongson-mmiintrin.h for back
2978 compatibility and warning.
2979 * config/mips/loongson-mmiintrin.h: ... here.
2980 * config/mips/mips.c (mips_hard_regno_mode_ok_uncached,
2981 mips_vector_mode_supported_p, AVAIL_NON_MIPS16): Use
2982 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
2983 (mips_option_override): Make sure MMI use hard float;
2984 (mips_shift_truncation_mask, mips_expand_vpc_loongson_even_odd,
2985 mips_expand_vpc_loongson_pshufh, mips_expand_vpc_loongson_bcast,
2986 mips_expand_vector_init): Use TARGET_LOONGSON_MMI instead of
2987 TARGET_LOONGSON_VECTORS.
2988 * gcc/config/mips/mips.h (TARGET_LOONGSON_VECTORS): Delete.
2989 (TARGET_CPU_CPP_BUILTINS): Add __mips_loongson_mmi.
2990 (MIPS_ASE_DSP_SPEC, MIPS_ASE_LOONGSON_MMI_SPEC): New.
2991 (BASE_DRIVER_SELF_SPECS): march=loongson2e/2f/3a implies
2993 (SHIFT_COUNT_TRUNCATED): Use TARGET_LOONGSON_MMI instead of
2994 TARGET_LOONGSON_VECTORS.
2995 * gcc/config/mips/mips.md (MOVE64, MOVE128): Use
2996 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
2997 (Loongson MMI patterns): Include loongson-mmi.md instead of
2999 * gcc/config/mips/mips.opt (-mloongson-mmi): New option.
3000 * gcc/doc/invoke.texi (-mloongson-mmi): Document.
3002 2018-11-07 Richard Biener <rguenther@suse.de>
3005 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Fixup
3006 BLOCK_ABSTRACT_ORIGIN to be the ultimate origin.
3008 2018-11-07 Alexandre Oliva <aoliva@redhat.com>
3010 PR rtl-optimization/87874
3011 * lra.c (lra_substitute_pseudo): Do not create a subreg for
3014 2018-11-06 Aaron Sawdey <acsawdey@linux.ibm.com>
3016 * config/rs6000/rs6000.md (bswap<mode>2): Force address into register
3017 if not in indexed or indirect form.
3018 (bswap<mode>2_load): Change predicate to indexed_or_indirect_operand.
3019 (bswap<mode>2_store): Ditto.
3021 2018-11-06 Richard Earnshaw <rearnsha@arm.com>
3023 * config/aarch64/aarch64.md (speculation_tracker): Set the mode for
3026 2018-11-06 Richard Biener <rguenther@suse.de>
3028 PR tree-optimization/86850
3029 * vec.h (vec<T, va_heap, vl_ptr>::splice): Check src.length ()
3030 instead of src.m_vec.
3032 2018-11-06 Jan Hubicka <jh@suse.cz>
3034 * tree.c (fld_simplified_type_name): Break out form ...
3035 (free_lang_data_in_type): ... here.
3036 (fld_type_variant_equal_p): Use it.
3038 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3040 * config/default-d.c: Include memmodel.h.
3042 * config/sol2-d.c: New file.
3043 * config/t-sol2 (sol2-d.o): New rule.
3044 * config.gcc <*-*-solaris2*>: Set d_target_objs,
3045 target_has_targetdm.
3047 2018-11-06 Jan Hubicka <jh@suse.cz>
3049 * tree.c (fld_type_variant): Also copy alignment; be sure that
3050 new variant is equal.
3052 2018-11-06 Ilya Leoshkevich <iii@linux.ibm.com>
3055 * config/s390/s390.md: Add relative_long attribute.
3057 2018-11-06 Jan Hubicka <jh@suse.cz>
3059 * ipa-pure-const.c (check_decl): Do not test TYPE_NEEDS_CONSTRUCTING.
3060 * lto-streamer-out.c (hash_tree): Do not hash TYPE_NEEDS_CONSTRUCTING.
3061 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Do not
3062 stream TYPE_NEEDS_CONSTRUCTING.
3063 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
3064 * tree.c (free_lang_data_in_type): Clear TYPE_NEEDS_CONSTRUCTING.
3066 2018-11-06 Richard Biener <rguenther@suse.de>
3068 * tree-vect-slp.c (vect_slp_bb): Move opening of vect_slp_analyze_bb
3070 (vect_slp_analyze_bb_1): ... here to avoid hiding optimized locations.
3072 2018-11-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3074 * gcc/config/msp430/msp430.h (REG_CLASS_CONTENTS): Add R0 to
3075 REG_CLASS_CONTENTS[GEN_REGS].
3076 (REGNO_REG_CLASS): Return NO_REGS for R2 and R3.
3078 2018-11-06 Jan Hubicka <jh@suse.cz>
3080 * tree.c (fld_simplified_type_of): Clear TYPELESS_STORAGE flag.
3082 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3085 * config/sol2.h (ASAN_CC1_SPEC): Define.
3086 (LD_WHOLE_ARCHIVE_OPTION): Define.
3087 (LD_NO_WHOLE_ARCHIVE_OPTION): Define.
3088 (ASAN_REJECT_SPEC): Provide default.
3089 (LIBASAN_EARLY_SPEC): Define.
3090 (LIBTSAN_EARLY_SPEC): Define.
3091 (LIBLSAN_EARLY_SPEC): Define.
3092 * config/i386/sol2.h (CC1_SPEC): Redefine.
3093 (ASAN_REJECT_SPEC): Define.
3095 * config/sparc/sparc.c (sparc_asan_shadow_offset): Declare.
3096 (TARGET_ASAN_SHADOW_OFFSET): Define.
3097 (sparc_asan_shadow_offset): New function.
3098 * config/sparc/sol2.h (CC1_SPEC): Append ASAN_CC1_SPEC.
3099 (ASAN_REJECT_SPEC): Define.
3101 2018-11-06 Jan Hubicka <jh@suse.cz>
3103 * tree.c (fld_type_variant): Copy canonical type.
3104 (fld_incomplete_type_of): Check that canonical types looks sane;
3105 copy canonical type.
3106 (verify_type): Accept when incomplete type has complete canonical type.
3108 2018-11-06 Jan Hubicka <jh@suse.cz>
3110 * tree.c (free_lang_data): Reset overwite_assembler_name,
3111 print_xnode, print_decl, print_type and print_identifier of
3114 2018-11-06 Richard Biener <rguenther@suse.de>
3116 PR tree-optimization/87889
3117 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
3118 Do nothing if old and new arg are the same
3120 2018-11-06 Andreas Krebbel <krebbel@linux.ibm.com>
3123 * config/s390/s390.md ("*r<noxa>sbg_di_rotl"): Remove mode
3124 attributes for operands 3 and 4.
3126 2018-11-06 Richard Biener <rguenther@suse.de>
3129 * simplify-rtx.c (simplify_binary_operation_1): Add pattern
3130 matching bitfield insertion.
3132 2018-11-06 Alexandre Oliva <aoliva@redhat.com>
3134 * auto-inc-dec.c: Include valtrack.h. Improve comments.
3135 (reg_next_debug_use): New.
3136 (attempt_change): Propagate adjusted expression into affected
3138 (merge_in_block): Track uses in debug insns.
3139 (pass_inc_dec::execute): Allocate and release
3142 2018-11-06 Wei Xiao <wei3.xiao@intel.com>
3144 * config/i386/avx512fintrin.h: Update VFIXUPIMM* intrinsics.
3145 (_mm512_fixupimm_round_pd): Update parameters and builtin.
3146 (_mm512_maskz_fixupimm_round_pd): Ditto.
3147 (_mm512_fixupimm_round_ps): Ditto.
3148 (_mm512_maskz_fixupimm_round_ps): Ditto.
3149 (_mm_fixupimm_round_sd): Ditto.
3150 (_mm_maskz_fixupimm_round_sd): Ditto.
3151 (_mm_fixupimm_round_ss): Ditto.
3152 (_mm_maskz_fixupimm_round_ss): Ditto.
3153 (_mm512_fixupimm_pd): Ditto.
3154 (_mm512_maskz_fixupimm_pd): Ditto.
3155 (_mm512_fixupimm_ps): Ditto.
3156 (_mm512_maskz_fixupimm_ps): Ditto.
3157 (_mm_fixupimm_sd): Ditto.
3158 (_mm_maskz_fixupimm_sd): Ditto.
3159 (_mm_fixupimm_ss): Ditto.
3160 (_mm_maskz_fixupimm_ss): Ditto.
3161 (_mm512_mask_fixupimm_round_pd): Update builtin.
3162 (_mm512_mask_fixupimm_round_ps): Ditto.
3163 (_mm_mask_fixupimm_round_sd): Ditto.
3164 (_mm_mask_fixupimm_round_ss): Ditto.
3165 (_mm512_mask_fixupimm_pd): Ditto.
3166 (_mm512_mask_fixupimm_ps): Ditto.
3167 (_mm_mask_fixupimm_sd): Ditto.
3168 (_mm_mask_fixupimm_ss): Ditto.
3169 * config/i386/avx512vlintrin.h:
3170 (_mm256_fixupimm_pd): Update parameters and builtin.
3171 (_mm256_maskz_fixupimm_pd): Ditto.
3172 (_mm256_fixupimm_ps): Ditto.
3173 (_mm256_maskz_fixupimm_ps): Ditto.
3174 (_mm_fixupimm_pd): Ditto.
3175 (_mm_maskz_fixupimm_pd): Ditto.
3176 (_mm_fixupimm_ps): Ditto.
3177 (_mm_maskz_fixupimm_ps): Ditto.
3178 (_mm256_mask_fixupimm_pd): Update builtin.
3179 (_mm256_mask_fixupimm_ps): Ditto.
3180 (_mm_mask_fixupimm_pd): Ditto.
3181 (_mm_mask_fixupimm_ps): Ditto.
3182 * config/i386/i386-builtin-types.def: Add new types and remove useless ones.
3183 * config/i386/i386-builtin.def: Update builtin definitions.
3184 * config/i386/i386.c: Handle new builtin types and remove useless ones.
3185 * config/i386/sse.md: Update VFIXUPIMM* patterns.
3186 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
3187 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
3188 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Update.
3189 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
3190 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
3191 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Update.
3192 * config/i386/subst.md:
3193 (round_saeonly_sd_mask_operand4): Add new subst_attr.
3194 (round_saeonly_sd_mask_op4): Ditto.
3195 (round_saeonly_expand_operand5): Ditto.
3196 (round_saeonly_expand): Update.
3198 2018-11-05 Max Filippov <jcmvbkbc@gmail.com>
3200 * config/xtensa/uclinux.h (XTENSA_ALWAYS_PIC): Change to 0.
3202 2018-11-05 Segher Boessenkool <segher@kernel.crashing.org>
3204 PR rtl-optimization/87871
3205 * combine.c (make_more_copies): Skip if dest is frame_pointer_rtx.
3207 2018-11-05 Paul Koning <ni1d@arrl.net>
3209 * doc/sourcebuild.texi (target attributes): Document new "inf"
3210 effective target keyword.
3212 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
3214 * config/s390/s390.c (s390_register_move_cost): Increase costs for
3215 moves involving the CC reg.
3217 2018-11-05 Richard Biener <rguenther@suse.de>
3219 PR tree-optimization/87873
3220 * tree-ssa-loop-manip.h (split_loop_exit_edge): Add copy_constants_p
3222 * tree-ssa-loop-manip.c (split_loop_exit_edge): Likewise.
3223 * tree-vect-loop.c (vect_transform_loop): When splitting the
3224 loop exit also create forwarder PHIs for constants.
3225 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
3226 Handle constant to_arg, add extra checking we match up the correct
3229 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
3231 * config/s390/s390.md: QImode and HImode for load on condition.
3233 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
3235 * config/s390/predicates.md: Fix typo.
3236 * config/s390/s390.md: Allow immediates for load on condition.
3238 2018-11-05 Martin Liska <mliska@suse.cz>
3240 * alloc-pool.h (struct pool_usage): Use SIZE_AMOUNT.
3241 * bitmap.h (struct bitmap_usage): Likewise.
3242 * ggc-common.c (SCALE): Remove.
3244 (struct ggc_usage): Use SIZE_AMOUNT. And update
3246 * ggc-page.c (SCALE): Remove.
3247 (STAT_LABEL): Remove.
3248 (ggc_print_statistics): Use SIZE_AMOUNT.
3249 * gimple.h (SCALE): Remove.
3251 * input.c (ONE_K): Remove.
3254 (STAT_LABEL): Likewise.
3255 (FORMAT_AMOUNT): Likewise.
3256 (dump_line_table_statistics): Use SIZE_AMOUNT.
3257 * mem-stats.h (struct mem_usage): Likewise.
3258 * rtl.c (dump_rtx_statistics): Likewise.
3259 (rtx_alloc_counts): Change type to size_t.
3260 (rtx_alloc_sizes): Likewise.
3261 (rtx_count_cmp): New.
3262 (dump_rtx_statistics): Sort first based on counts.
3263 * tree.c (tree_nodes_cmp): New.
3264 (tree_codes_cmp): New.
3265 (dump_tree_statistics): Sort first based on counts.
3266 * system.h (ONE_K): New.
3268 (SIZE_SCALE): Likewise.
3269 (SIZE_LABEL): Likewise.
3270 (SIZE_AMOUNT): Likewise.
3271 * tree-cfg.c (dump_cfg_stats): Use SIZE_AMOUNT.
3272 * tree-dfa.c (dump_dfa_stats): Likewise.
3273 * tree-phinodes.c (phinodes_print_statistics): Likewise.
3274 * tree-ssanames.c (ssanames_print_statistics): Likewise.
3275 * tree.c (dump_tree_statistics): Likewise.
3276 * vec.c (struct vec_usage): Likewise.
3277 * trans-mem.c (tm_mangle): Enlarge buffer in order to not
3278 trigger a -Werror=format-overflow with
3279 --enable-gather-detailed-stats.
3281 2018-11-05 Martin Liska <mliska@suse.cz>
3283 * mem-stats.h (mem_alloc_description::release_instance_overhead):
3285 * vec.c (struct vec_usage): Register m_element_size.
3286 (vec_prefix::register_overhead): New arguments: elements and
3288 (vec_prefix::release_overhead): Subtract elements.
3289 * vec.h (struct vec_prefix): Change signature.
3290 (va_heap::reserve): Pass proper arguments.
3291 (va_heap::release): Likewise.
3293 2018-11-05 Martin Liska <mliska@suse.cz>
3295 * mem-stats.h (mem_alloc_description::get_list): Fix GNU coding
3299 2018-11-05 Richard Biener <rguenther@suse.de>
3301 * tree-scalar-evolution.h (final_value_replacement_loop): Update
3303 * tree-scalar-evolution.c (final_value_replacement_loop): Return
3304 whether anything was done.
3305 (scev_const_prop): Remove constant propagation part, fold
3307 * tree-ssa-loop.c (pass_scev_cprop::execute): ... here.
3308 (pass_data_scev_cprop): TODO_cleanup_cfg is now done
3311 2018-11-05 Jakub Jelinek <jakub@redhat.com>
3313 PR tree-optimization/87859
3314 * gimple-ssa-store-merging.c (struct merged_store_group): Add
3315 only_constants and first_nonmergeable_order members.
3316 (merged_store_group::merged_store_group): Initialize them.
3317 (merged_store_group::do_merge): Clear only_constants member if
3318 adding something other than INTEGER_CST store.
3319 (imm_store_chain_info::coalesce_immediate_stores): Don't merge
3320 stores with order >= first_nonmergeable_order. Use
3321 merged_store->only_constants instead of always recomputing it.
3322 Set merged_store->first_nonmergeable_order if we've skipped any
3323 stores. Attempt to merge overlapping INTEGER_CST stores that
3324 we would otherwise skip.
3327 * match.pd (X + Y < X): Don't optimize if TYPE_OVERFLOW_SANITIZED.
3329 2018-11-05 Xuepeng Guo <xuepeng.guo@intel.com>
3332 * config/i386/emmintrin.h (__v16qs): New to cope with option
3334 (_mm_cmpeq_epi8): Replace __v16qi with __v16qs.
3335 (_mm_cmplt_epi8): Likewise.
3336 (_mm_cmpgt_epi8): Likewise.
3338 2018-11-05 Richard Biener <rguenther@suse.de>
3340 PR rtl-optimization/87852
3341 * fwprop.c (use_killed_between): Only consider single-defs of the
3342 use whose definition statement dominates the use.
3344 2018-11-05 Martin Liska <mliska@suse.cz>
3347 * doc/invoke.texi: Remove options that are
3348 not disabled with -Os.
3350 2018-11-05 Martin Liska <mliska@suse.cz>
3353 * doc/extend.texi: Update constrain about the last argument
3354 of __builtin_expect_with_probability.
3356 2018-11-05 Martin Liska <mliska@suse.cz>
3359 * predict.c (expr_expected_value_1): Verify
3360 that last argument is a real constants and emit
3363 2018-11-05 Martin Liska <mliska@suse.cz>
3365 PR gcov-profile/77698
3366 * ipa-profile.c (ipa_profile): Adjust hotness threshold
3369 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3371 PR tree-optimization/86572
3372 * builtins.c (c_strlen): Handle negative offsets in a safe way.
3374 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3376 PR tree-optimization/87672
3377 * gimple-fold.c (gimple_fold_builtin_stxcpy_chk): Gimplify.
3378 * tree-ssa-strlen.c (handle_builtin_strcat): Adjust object size.
3380 2018-11-04 Uros Bizjak <ubizjak@gmail.com>
3383 * cfgexpand.c (pass_expand::execute): Move the call to
3384 finish_eh_generation in front of the call to expand_stack_alignment.
3386 2018-11-04 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
3388 * common/config/i386/i386-common.c (processor_alias_table): Add
3390 * config.gcc (i[34567]86-*-linux* | ...): Add znver2.
3391 (case ${target}): Add znver2.
3392 * config/i386/driver-i386.c: (host_detect_local_cpu): Let
3393 -march=native recognize znver2 processors.
3394 * config/i386/i386-c.c (ix86_target_macros_internal): Add znver2.
3395 * config/i386/i386.c (m_znver2): New definition.
3396 (m_ZNVER): New definition.
3397 (m_AMD_MULTIPLE): Includes m_znver2.
3398 (processor_cost_table): Add znver2 entry.
3399 (processor_target_table): Add znver2 entry.
3400 (get_builtin_code_for_version): Set priority for
3402 (processor_model): Add M_AMDFAM17H_ZNVER2.
3403 (arch_names_table): Ditto.
3404 (ix86_reassociation_width): Include znver2.
3405 * config/i386/i386.h (TARGET_znver2): New definition.
3406 (struct ix86_size_cost): Add TARGET_ZNVER2.
3407 (enum processor_type): Add PROCESSOR_ZNVER2.
3408 * config/i386/i386.md (define_attr "cpu"): Add znver2.
3409 * config/i386/x86-tune-costs.h: (processor_costs) Add znver2 costs.
3410 * config/i386/x86-tune-sched.c: (ix86_issue_rate): Add znver2.
3411 (ix86_adjust_cost): Add znver2.
3412 * config/i386/x86-tune.def: Replace m_ZNVER1 by m_ZNVER.
3413 * gcc/doc/extend.texi: Add details about znver2.
3414 * gcc/doc/invoke.texi: Add details about znver2.
3416 2018-11-03 Sandra Loosemore <sandra@codesourcery.com>
3420 * config/nios2/nios2.c (nios2_rtx_costs): Recognize <mul>sidi3
3423 2018-11-02 Aaron Sawdey <acsawdey@linux.ibm.com>
3425 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Pay
3426 attention to TARGET_AVOID_XFORM and BYTES_BIG_ENDIAN.
3428 2018-11-02 Richard Earnshaw <rearnsha@arm.com>
3430 * config/aarch64/aarch64.c ((aarch64_override_options): Disable
3431 shrink-wrapping when -mtrack-speculation.
3433 2018-11-02 Richard Biener <rguenther@suse.de>
3435 * tree-ssa-coalesce.c (struct coalesce_list): Add obstack member.
3436 (pop_cost_one_pair): Do not free pair.
3437 (pop_best_coalesce): Likewise.
3438 (create_coalesce_list): Initialize obstack.
3439 (delete_coalesce_list): Free obstack.
3440 (find_coalesce_pair): Obstack-allocate coalesce pairs.
3441 (add_cost_one_coalesce): Likewise.
3442 (struct live_track): Remove bitmap pointer indirections.
3443 (new_live_track): Adjust.
3444 (delete_live_track): Likewise.
3445 (live_track_remove_partition): Likewise.
3446 (live_track_add_partition): Likewise.
3447 (live_track_live_p): Likewise.
3448 (live_track_process_def): Likewise.
3449 (live_track_clear_base_vars): Likewise.
3451 2018-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3453 * configure.ac (gcc_cv_as_sparc_register_op): Remove.
3454 * configure: Regenerate.
3455 * config.in: Regenerate.
3456 * config/sparc/sparc.c (sparc_output_scratch_registers): Remove
3457 HAVE_AS_REGISTER_PSEUDO_OP guard.
3458 * config/sparc/sparc.h (ASM_ARCH64_SPEC): Remove
3459 !HAVE_AS_REGISTER_PSEUDO_OP support.
3460 (ASM_DECLARE_REGISTER_GLOBAL): Remove HAVE_AS_REGISTER_PSEUDO_OP
3463 2018-11-02 Richard Biener <rguenther@suse.de>
3465 * dwarf2out.c (build_abbrev_table): Guard lookup_external_ref call
3468 2018-11-02 Richard Biener <rguenther@suse.de>
3470 PR tree-optimization/87776
3471 * tree-ssa-sccvn.c (do_rpo_vn): Do not mark backedges
3472 executable when iterating but running into rpo-vn-max-loop-depth
3473 and not eliding the iteration.
3475 2018-11-30 Jan Hubicka <jh@suse.cz>
3477 * tree.c (free_lang_data_in_decl): Free TREE_TYPE and alignment of
3480 2018-11-01 Aaron Sawdey <acsawdey@linux.ibm.com>
3482 * config/rs6000/rs6000-protos.h (rs6000_address_for_fpconvert): Remove
3484 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem):
3485 Combine with rs6000_address_for_fpconvert.
3486 (rs6000_address_for_fpconvert): Combine with
3487 rs6000_force_indexed_or_indirect_mem.
3488 (rs6000_expand_vector_init): Change function call from
3489 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
3490 * config/rs6000/rs6000.md (floatsi<mode>2_lfiwax): Change call from
3491 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
3492 (floatsi<mode>2_lfiwax_mem): Ditto.
3493 (floatunssi<mode>2_lfiwzx): Ditto.
3494 (floatunssi<mode>2_lfiwzx_mem): Ditto.
3495 (float<QHI:mode><FP_ISA3:mode>2): Ditto.
3496 (floatuns<QHI:mode><FP_ISA3:mode>2): Ditto.
3497 (fix_trunc<mode>si2_stfiwx): Ditto.
3498 (fixuns_trunc<mode>si2_stfiwx): Ditto.
3499 (float_<mode>si2_hw): Ditto.
3500 (floatuns_<mode>si2_hw): Ditto.
3501 * config/rs6000/vsx.md (*vsx_extract_si): Ditto.
3502 (vsx_splat_<mode>): Ditto.
3504 2018-11-01 Joseph Myers <joseph@codesourcery.com>
3506 * configure.ac (gcc_cv_as_ia64_ltoffx_ldxmov_relocs): Add newline
3507 at end of assembler input text.
3508 * configure: Regenerate.
3510 2018-11-01 Jakub Jelinek <jakub@redhat.com>
3512 PR tree-optimization/87826
3513 * gimple-ssa-store-merging.c (do_shift_rotate): Punt if count is
3514 negative or larger or equal to type's precision.
3516 2018-10-31 Alexandre Oliva <aoliva@redhat.com>
3518 * opts.c (default_options_table): Do not enable
3519 OPT_fdelayed_branch at -Og.
3520 * doc/invoke.texi (-fdelayed-branch): Document it.
3522 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
3524 * optabs-libfuncs.c (build_libfunc_function_visibility):
3525 New, split out from...
3526 (build_libfunc_function): ... here.
3527 (init_one_libfunc_visibility): New, split out from ...
3528 (init_one_libfunc): ... here.
3530 * config/aarch64/atomics.md (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse):
3531 scratch register need not be early-clobber. Document the reason
3532 why we cannot use ST<OP>.
3534 2018-10-31 Joseph Myers <joseph@codesourcery.com>
3537 * configure.ac: Remove AC_PREREQ. Use AC_LANG_SOURCE. Use single
3538 line for second argument of AC_DEFINE_UNQUOTED.
3539 * doc/install.texi (Tools/packages necessary for modifying GCC):
3540 Update to autoconf 2.69 and automake 1.15.1.
3541 * aclocal.m4, config.in, configure: Regenerate.
3543 2018-10-31 Pat Haugen <pthaugen@us.ibm.com>
3545 * tree-ssa-loop-ivcanon.c (try_peel_loop): Fix typo and remove dead
3548 2018-10-31 Martin Liska <mliska@suse.cz>
3551 * common/config/aarch64/aarch64-common.c (aarch64_parse_extension):
3552 Add new argument invalid_extension.
3553 (aarch64_get_all_extension_candidates): New function.
3554 (aarch64_rewrite_selected_cpu): Add NULL to function call.
3555 * config/aarch64/aarch64-protos.h (aarch64_parse_extension): Add
3557 (aarch64_get_all_extension_candidates): New function.
3558 * config/aarch64/aarch64.c (aarch64_parse_arch): Add new
3559 argument invalid_extension.
3560 (aarch64_parse_cpu): Likewise.
3561 (aarch64_print_hint_for_extensions): New function.
3562 (aarch64_validate_mcpu): Provide hint about invalid extension.
3563 (aarch64_validate_march): Likewise.
3564 (aarch64_handle_attr_arch): Pass new argument.
3565 (aarch64_handle_attr_cpu): Provide hint about invalid extension.
3566 (aarch64_handle_attr_isa_flags): Likewise.
3568 2018-10-31 Richard Biener <rguenther@suse.de>
3572 * tree-outof-ssa.c (insert_backedge_copies): Restrict
3573 copy generation to useful cases. Place the copy before
3574 the definition of the backedge value when possible.
3576 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
3578 * config/arc/arc-arch.h (ARC_TUNE_ARC7XX): New tune value.
3579 * config/arc/arc.c (arc_active_insn): New function.
3580 (check_store_cacheline_hazard): Likewise.
3581 (workaround_arc_anomaly): Use check_store_cacheline_hazard.
3582 (arc_override_options): Disable delay slot scheduler for older
3584 (arc_store_addr_hazard_p): New implementation, old one renamed to
3586 (arc_store_addr_hazard_internal_p): Renamed.
3587 (arc_reorg): Don't combine into brcc instructions which are part
3588 of hardware hazard solution.
3589 * config/arc/arc.md (attr tune): Consider new arc7xx tune value.
3590 (tune_arc700): Likewise.
3591 * config/arc/arc.opt (arc7xx): New tune value.
3592 * config/arc/arc700.md: Improve A7 scheduler.
3594 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
3596 * config/arc/arc.c (arc_override_options): Remove
3597 TARGET_COMPACT_CASESI.
3598 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Update.
3599 (CASE_VECTOR_MODE): Likewise.
3600 (CASE_VECTOR_PC_RELATIVE): Likewise.
3601 (CASE_VECTOR_SHORTEN_MODE): Likewise.
3602 (CASE_VECTOR_SHORTEN_MODE1): Delete.
3603 (ADDR_VEC_ALIGN): Update.
3604 (ASM_OUTPUT_CASE_LABEL): Undefine.
3605 (ASM_OUTPUT_BEFORE_CASE_LABEL): Undefine.
3606 (TARGET_BI_BIH): Define.
3607 (DEFAULT_BRANCH_INDEX): Likewise.
3608 * config/arc/arc.md (casesi): Rework to accept BI/BIH
3609 instructions, remove compact_casesi use case.
3610 (casesi_compact_jump): Remove.
3611 (casesi_dispatch): New pattern.
3612 * config/arc/arc.opt: Add mbranch-index option. Deprecate
3613 compact_casesi option.
3614 * doc/invoke.texi: Document mbranch-index option.
3616 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
3618 * config/arc/arc.c (arc_get_tp): Remove function.
3619 (arc_emit_call_tls_get_addr): Likewise.
3620 (arc_call_tls_get_addr): New function.
3621 (arc_legitimize_tls_address): Make use of arc_call_tls_get_addr.
3622 * config/arc/arc.md (tls_load_tp_soft): Remove.
3623 (tls_gd_get_addr): Likewise.
3625 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
3627 * config/arc/arc.md (mulsi3): Remove call to mulsi_600_lib.
3628 (mulsi3_600_lib): Remove pattern.
3629 (umulsi3_highpart_600_lib_le): Likewise.
3630 (umulsi3_highpart): Remove call to umulsi3_highpart_600_lib_le.
3631 (umulsidi3): Remove call to umulsidi3_600_lib.
3632 (umulsidi3_600_lib): Remove pattern.
3633 (peephole2): Remove peephole using the above deprecated patterns.
3635 2018-10-31 Thomas Preud'homme <thomas.preudhomme@linaro.org>
3638 * config/arm/arm.c (arm_option_check_internal): Disable the combined
3639 use of -mslow-flash-data and -mword-relocations.
3640 (arm_option_override): Enable -mword-relocations if -fpic or -fPIC.
3641 * config/arm/arm.md (SYMBOL_REF MOVT splitter): Stop checking for
3643 * doc/invoke.texi (-mword-relocations): Mention conflict with
3645 (-mslow-flash-data): Reciprocally.
3647 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
3649 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Force
3650 16-byte modes held in GP registers to use an even regno.
3652 * config/aarch64/aarch64.c (aarch64_emit_bic): Remove.
3653 (aarch64_atomic_ldop_supported_p): Remove.
3654 (aarch64_gen_atomic_ldop): Remove.
3655 * config/aarch64/atomic.md (atomic_<atomic_optab><ALLI>):
3656 Fully expand LSE operations here.
3657 (atomic_fetch_<atomic_optab><ALLI>): Likewise.
3658 (atomic_<atomic_optab>_fetch<ALLI>): Likewise.
3659 (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse): Drop atomic_op iterator
3660 and use ATOMIC_LDOP instead; use register_operand for the input;
3661 drop the split and emit insns directly.
3662 (aarch64_atomic_fetch_<ATOMIC_LDOP><ALLI>_lse): Likewise.
3663 (aarch64_atomic_<atomic_op>_fetch<ALLI>_lse): Remove.
3664 (@aarch64_atomic_load<ATOMIC_LDOP><ALLI>): Remove.
3666 * config/aarch64/aarch64.c (aarch64_emit_atomic_swap): Remove.
3667 (aarch64_gen_atomic_ldop): Don't call it.
3668 * config/aarch64/atomics.md (atomic_exchange<ALLI>):
3669 Use aarch64_reg_or_zero.
3670 (aarch64_atomic_exchange<ALLI>): Likewise.
3671 (aarch64_atomic_exchange<ALLI>_lse): Remove split; remove & from
3672 operand 0; use aarch64_reg_or_zero for input; merge ...
3673 (@aarch64_atomic_swp<ALLI>): ... this and remove.
3675 * config/aarch64/aarch64.c (aarch64_gen_compare_reg_maybe_ze): New.
3676 (aarch64_split_compare_and_swap): Use it.
3677 (aarch64_expand_compare_and_swap): Likewise. Remove convert_modes;
3678 test oldval against the proper predicate.
3679 * config/aarch64/atomics.md (@atomic_compare_and_swap<ALLI>):
3680 Use nonmemory_operand for expected.
3681 (cas_short_expected_pred): New.
3682 (@aarch64_compare_and_swap<SHORT>): Use it; use "rn" not "rI" to match.
3683 (@aarch64_compare_and_swap<GPI>): Use "rn" not "rI" for expected.
3684 * config/aarch64/predicates.md (aarch64_plushi_immediate): New.
3685 (aarch64_plushi_operand): New.
3687 * config/aarch64/aarch64.c (aarch64_expand_compare_and_swap):
3688 Force oldval into the rval register for TARGET_LSE; emit the compare
3689 during initial expansion so that it may be deleted if unused.
3690 (aarch64_gen_atomic_cas): Remove.
3691 * config/aarch64/atomics.md (@aarch64_compare_and_swap<SHORT>_lse):
3692 Change =&r to +r for operand 0; use match_dup for operand 2;
3693 remove is_weak and mod_f operands as unused. Drop the split
3695 (@aarch64_atomic_cas<SHORT>): ... this pattern's output; remove.
3696 (@aarch64_compare_and_swap<GPI>_lse): Similarly.
3697 (@aarch64_atomic_cas<GPI>): Similarly.
3699 2018-10-31 Richard Biener <rguenther@suse.de>
3701 * tree-eh.c (replace_trapping_overflow): Simplify ABS_EXPR case
3704 2018-10-31 Sameera Deshpande <sameera.deshpande@linaro.org>
3706 * config/aarch64/aarch64-cores.def (saphira): Use saphira pipeline.
3707 * config/aarch64/aarch64.md: Include saphira.md
3708 * config/aarch64/saphira.md: New file for pipeline description.
3710 2018-10-30 Martin Sebor <msebor@redhat.com>
3713 * gimple-ssa-sprintf.c (format_directive): Use %G to include
3715 (sprintf_dom_walker::compute_format_length):
3716 Avoid setting POSUNDER4K here.
3717 (get_destination_size): Handle null argument values.
3718 (get_user_idx_format): New function.
3719 (sprintf_dom_walker::handle_gimple_call): Handle all printf-like
3720 functions, including user-defined with attribute format printf.
3721 Use %G to include inlining context.
3722 Set POSUNDER4K here.
3724 2018-10-30 Jan Hubicka <jh@suse.cz>
3726 * params.def (lto-partitions): Bump from 32 to 128.
3728 2018-10-30 Jan Hubicka <jh@suse.cz>
3731 (free_lang_data_d, add_tree_to_fld_list, fld_worklist_push): Move
3733 (free_lang_data_in_type): Forward declare.
3734 (fld_type_variant_equal_p): New function.
3735 (fld_type_variant): New function
3736 (fld_incomplete_types): New hash.
3737 (fld_incomplete_type_of): New function
3738 (fld_simplfied-type): New function.
3739 (free_lang_data_in_decl): Add fld parameter; simplify type of FIELD_DECL
3740 (free_lang_data): Allocate and free fld_incomplete_type; update call
3741 of free_lang_data_in_decl.
3743 2018-10-30 Eric Botcazou <ebotcazou@adacore.com>
3745 * gcov.c (output_lines): Remove duplicate line.
3747 2018-10-30 Aaron Sawdey <acsawdey@linux.ibm.com>
3749 * config/rs6000/rs6000.md (bswapdi2): Force address into register
3750 if not in indexed or indirect form.
3751 (bswapdi2_load): Change predicate to indexed_or_indirect_operand.
3752 (bswapdi2_store): Ditto.
3753 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem): New
3755 * config/rs6000/rs6000-protos.h (rs6000_force_indexed_or_indirect_mem):
3756 Prototype for helper function.
3758 2018-10-30 Martin Sebor <msebor@redhat.com>
3760 * doc/extend.texi (optimize): Clarify/expand attribute documentation.
3761 (target, pragma GCC optimize, pragma GCC target): Ditto.
3763 2018-10-30 Jonathan Wakely <jwakely@redhat.com>
3765 * doc/extend.texi: Fix prototype and description of
3766 __builtin_expect_with_probability.
3768 2018-10-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
3770 * cgraph.h (clone_function_name_1): Replaced by new
3771 clone_function_name_numbered that takes name as string; for
3772 privatize_symbol_name_1 use only.
3773 (clone_function_name): Renamed to
3774 clone_function_name_numbered to be explicit about numbering.
3775 (clone_function_name): New two-argument function that does
3776 not number its output.
3777 (clone_function_name): New three-argument function that
3778 takes a number to append to its output.
3779 * cgraphclones.c (duplicate_thunk_for_node):
3780 (clone_function_name_1): Renamed.
3781 (clone_function_name_numbered): Two new functions.
3782 (clone_function_name): Improved documentation.
3783 (cgraph_node::create_virtual_clone): Use clone_function_name_numbered.
3784 * config/rs6000/rs6000.c (make_resolver_func): Ditto.
3785 * final.c (final_scan_insn_1): Use the new clone_function_name
3787 * multiple_target.c (create_dispatcher_calls): Ditto.
3788 (create_target_clone): Ditto.
3789 * omp-expand.c (grid_expand_target_grid_body): Ditto.
3790 * omp-low.c (create_omp_child_function_name): Ditto.
3791 * omp-simd-clone.c (simd_clone_create): Ditto.
3792 * symtab.c (simd_symtab_node::noninterposable_alias): Use the
3793 new clone_function_name without numbering.
3795 2018-10-30 Richard Earnshaw <rearnsha@arm.com>
3797 * alloc-pool.h (base_pool_allocator <TBlockAllocator>::initialize):
3798 Assert that the allocation size is not zero.
3800 2018-10-30 Richard Biener <rguenther@suse.de>
3802 PR tree-optimization/87800
3803 * tree-vect-slp.c (vect_build_slp_tree_2): Reject any
3804 non-induction or reduction PHIs.
3806 2018-10-30 Sameera Deshpande <sameera.deshpande@linaro.org
3808 * config/aarch64/falkor-tag-collision-avoidance.c
3809 (execute_tag_collision_avoidance): Call df_note_add_problem.
3811 2018-10-30 Martin Liska <mliska@suse.cz>
3813 * doc/extend.texi: Fix typo in documentation
3814 of __builtin_expect_with_probability.
3816 2018-10-29 David Malcolm <dmalcolm@redhat.com>
3819 * input.c (get_substring_ranges_for_loc): Detect if
3820 linemap_resolve_location gives us a NULL map, and reject
3823 2018-10-29 Iain Buclaw <ibuclaw@gdcproject.org>
3825 * config.gcc (xstormy16-*-elf): Set tm_d_file.
3827 2018-10-29 David Malcolm <dmalcolm@redhat.com>
3828 Martin Sebor <msebor@redhat.com>
3829 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
3831 * doc/ux.texi (Quoting): New subsection, adapted from material at
3832 https://gcc.gnu.org/wiki/DiagnosticsGuidelines written by
3833 MartinSebor and ManuelLopezIbanez.
3834 (Fix-it hints): Note that fix-it hints shouldn't be marked for
3837 2018-10-29 Kugan Vivekanandarajah <kuganv@linaro.org>
3840 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Fix niter
3843 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
3845 * gcc.target/powerpc/ssse3-check.h: Remove inaccurate comment.
3847 2018-10-29 Paul Koning <ni1d@arrl.net>
3849 * doc/rtl.texi (CONST_WIDE_INT_ELT): Give correct macro name.
3851 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
3853 * gcc/config/rs6000/mmintrin.h (_mm_packs_pi16, _mm_packs_pi32,
3854 _mm_packs_pu16, _mm_unpackhi_pi8, _mm_unpacklo_pi8, _mm_add_pi8,
3855 _mm_add_pi16, _mm_add_pi32, _mm_sub_pi8, _mm_sub_pi16, _mm_sub_pi32,
3856 _mm_cmpgt_pi8, _mm_cmpeq_pi16, _mm_cmpgt_pi16, _mm_cmpeq_pi32,
3857 _mm_cmpgt_pi32, _mm_adds_pi8, _mm_adds_pi16, _mm_adds_pu8,
3858 _mm_adds_pu16, _mm_subs_pi8, _mm_subs_pi16, _mm_subs_pu8,
3859 _mm_subs_pu16, _mm_madd_pi16, _mm_mulhi_pi16, _mm_mullo_pi16,
3860 _mm_sll_pi16, _mm_sra_pi16, _mm_srl_pi16, _mm_set1_pi16, _mm_set1_pi8):
3861 Change 'vector' to '__vector'.
3862 * gcc/config/rs6000/xmmintrin.h (_mm_cvtps_pi32, _mm_cvttps_pi32,
3863 _mm_cvtps_pi16, _mm_cvtps_pi8, _mm_max_pi16, _mm_max_pu8, _mm_min_pi16,
3864 _mm_min_pu8, _mm_mulhi_pu16, _mm_shuffle_pi16, _mm_avg_pu8,
3865 _mm_avg_pu16): Likewise. And, whitespace corrections.
3867 2018-10-29 Richard Biener <rguenther@suse.de>
3869 PR tree-optimization/87785
3870 * tree-vect-slp.c (vect_gather_slp_loads): Only gather
3873 2018-10-29 Olivier Hainque <hainque@adacore.com>
3875 * VXWORKS_NET_LIBS_RTP: New macro, network part of VXWORKS_LIBS_RTP.
3877 2018-10-29 Olivier Hainque <hainque@adacore.com>
3879 * config/vxworks.h (VXWORKS_PERSONALITY): New VxWorks
3880 ports configuration macro, defaults to "gnu".
3881 (VXWORKS_OS_CPP_BUILTINS): Use it. Feed TOOL and
3882 TOOL_FAMILY instead of _VX_TOOL/_VXTOOL_FAMILY.
3884 2018-10-29 Olivier Hainque <hainque@adacore.com>
3886 * config/vxworks.h (CLEAR_INSN_CACHE): #undef before #define.
3888 2018-10-29 Richard Biener <rguenther@suse.de>
3890 PR tree-optimization/87790
3891 * tree-vect-slp.c (vect_mark_slp_stmts): Simplify.
3892 (vect_make_slp_decision): Adjust.
3893 (vect_slp_analyze_bb_1): Likewise.
3894 (vect_detect_hybrid_slp_stmts): Properly union SLP type over
3897 2018-10-29 Richard Biener <rguenther@suse.de>
3899 PR tree-optimization/87785
3900 * tree-vect-slp.c (vect_build_slp_tree_2): Remove loads argument
3902 (vect_build_slp_tree): Likewise.
3903 (vect_gather_slp_loads): New function.
3904 (vect_analyze_slp_instance): Gather loads separately from the
3907 2018-10-29 Martin Liska <mliska@suse.cz>
3909 * Makefile.in: Make dependency to json.o.
3910 * doc/gcov.texi: Document new JSON format, remove
3911 old intermediate format documentation.
3912 * gcov.c (struct function_info): Come up with m_name and
3914 (function_info::function_info): Initialize it.
3915 (function_info::~function_info): Release it.
3916 (main): Rename flag_intermediate_format to flag_json_format.
3917 (print_usage): Describe --json-format.
3918 (process_args): Set flag_json_format.
3919 (output_intermediate_line): Remove.
3920 (output_intermediate_json_line): Likewise.
3921 (get_gcov_intermediate_filename): Return new extension
3923 (output_intermediate_file): Implement JSON emission.
3924 (output_json_intermediate_file): Implement JSON emission.
3925 (generate_results): Use ::get_name for function name.
3926 Handle JSON output file.
3927 (read_graph_file): Use ::get_name instead of cplus_demangle.
3928 (read_count_file): Likewise.
3929 (solve_flow_graph): Likewise.
3930 (add_line_counts): Likewise.
3931 (accumulate_line_counts): Use new flag_json_format.
3932 (output_function_details): Use ::get_name instead of cplus_demangle.
3933 (output_lines): Likewise.
3934 * json.cc (test_writing_literals): Add new tests.
3935 * json.h (class literal): Add new boolean constructor.
3937 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
3939 PR rtl-optimization/87701
3940 PR rtl-optimization/87780
3941 * combine.c (make_more_copies): Rewrite.
3943 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
3945 * doc/generic.texi (ABSU_EXPR): Document.
3946 * match.pd (absu(x)*absu(x) -> x*x): Handle.
3947 (absu(absu(X)) -> absu(X)): Likewise.
3948 (absu(-X) -> absu(X)): Likewise.
3949 (absu(X) where X is nonnegative -> X): Likewise.
3951 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
3953 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
3954 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
3955 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
3956 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
3957 (s-tm-texi): Also check timestamp on d-target.def.
3958 (generated_files): Add TM_D_H and d-target-hooks-def.h.
3959 (build/genhooks.o): Also depend on D_TARGET_DEF.
3960 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
3962 * config/aarch64/aarch64-d.c: New file.
3963 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
3965 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
3967 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
3968 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
3969 * config/arm/arm-d.c: New file.
3970 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
3971 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
3972 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
3973 * config/arm/t-arm (arm-d.o): New rule.
3974 * config/default-d.c: New file.
3975 * config/glibc-d.c: New file.
3976 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3977 * config/i386/i386-d.c: New file.
3978 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
3979 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
3980 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
3981 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
3982 * config/i386/t-i386 (i386-d.o): New rule.
3983 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3984 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3985 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
3986 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3987 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
3988 * config/mips/mips-d.c: New file.
3989 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
3990 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
3991 * config/mips/t-mips (mips-d.o): New rule.
3992 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3993 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3994 * config/powerpcspe/powerpcspe-d.c: New file.
3995 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
3997 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
3998 Support GNU D by using 0 as the language type.
3999 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
4000 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
4001 * config/riscv/riscv-d.c: New file.
4002 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
4004 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
4005 * config/riscv/t-riscv (riscv-d.o): New rule.
4006 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
4007 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
4008 * config/rs6000/rs6000-d.c: New file.
4009 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
4011 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
4012 Support GNU D by using 0 as the language type.
4013 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
4014 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
4015 * config/s390/s390-d.c: New file.
4016 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
4017 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
4018 * config/s390/t-s390 (s390-d.o): New rule.
4019 * config/sparc/sparc-d.c: New file.
4020 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
4022 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
4023 * config/sparc/t-sparc (sparc-d.o): New rule.
4024 * config/t-glibc (glibc-d.o): New rule.
4025 * configure: Regenerated.
4026 * configure.ac (tm_d_file): New variable.
4027 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
4028 * doc/contrib.texi (Contributors): Add self for the D frontend.
4029 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
4030 * doc/install.texi (Configuration): Mention libphobos as an option for
4031 --enable-shared. Mention d as an option for --enable-languages.
4032 (Testing): Mention check-d as a target.
4033 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
4034 name suffixes. Mention d as a -x option.
4035 * doc/sourcebuild.texi (Top Level): Mention libphobos.
4036 * doc/standards.texi (Standards): Add section on D language.
4037 * doc/tm.texi: Regenerated.
4038 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
4039 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
4040 * dwarf2out.c (is_dlang): New function.
4041 (gen_compile_unit_die): Use DW_LANG_D for D.
4042 (declare_in_namespace): Return module die for D, instead of adding
4043 extra declarations into the namespace.
4044 (gen_namespace_die): Generate DW_TAG_module for D.
4045 (gen_decl_die): Handle CONST_DECLSs for D.
4046 (dwarf2out_decl): Likewise.
4047 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
4048 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
4049 kinds of aggregates.
4050 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
4051 * genhooks.c: Include d/d-target.def.
4053 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
4056 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
4057 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
4058 (STACK_DYNAMIC_OFFSET): Likewise.
4060 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
4063 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
4066 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
4068 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
4069 return type and other typos.
4071 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
4073 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
4074 a shorter sequence with fewer branches.
4075 (emit_final_str_compare_gpr): Ditto.
4077 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
4079 * config/rs6000/tmmintrin.h: New file.
4080 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
4082 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
4084 * config/rs6000/mmintrin.h: Enable 32bit compilation.
4085 * config/rs6000/xmmintrin.h: Likewise.
4087 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
4089 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
4091 2018-10-26 Richard Biener <rguenther@suse.de>
4093 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
4095 (vect_mark_slp_stmts_relevant): Likewise.
4096 (vect_detect_hybrid_slp_stmts): Likewise.
4097 (vect_bb_slp_scalar_cost): Likewise.
4098 (vect_remove_slp_scalar_calls): Likewise.
4100 2018-10-26 Jan Hubicka <jh@suse.cz>
4102 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
4103 (warn_types_mismatch): Fix walk of DECL_NAME.
4104 (odr_types_equivalent_p): Fix overactive assert.
4106 2018-10-26 Richard Biener <rguenther@suse.de>
4108 PR tree-optimization/87105
4109 * tree-vectorizer.h (_slp_tree::refcnt): New member.
4110 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
4112 (vect_create_new_slp_node): Initialize refcnt to one.
4114 (scalar_stmts_set_t, bst_fail): Remove.
4115 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
4116 (vect_build_slp_tree): Add bst_map argument and lookup
4117 already created SLP nodes.
4118 (vect_print_slp_tree): Handle a SLP graph, print SLP node
4120 (vect_slp_rearrange_stmts): Handle a SLP graph.
4121 (vect_analyze_slp_instance): Adjust and free SLP nodes from
4122 the CSE map. Fix indenting.
4123 (vect_schedule_slp_instance): Add short-cut.
4125 2018-10-26 Martin Liska <mliska@suse.cz>
4128 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
4129 addr_expr and not with pointers.
4131 2018-10-26 Jan Hubicka <jh@suse.cz>
4133 * tree.c (free_lang_data_in_type): Only check main variants.
4134 * ipa-devirt.c (warn_odr): Make static.
4135 (types_same_for_odr): Drop strict variant.
4136 (types_odr_comparable): Likewise.
4137 (odr_or_derived_type_p): Look for main variants.
4138 (odr_name_hasher::equal): Cleanup comment.
4139 (odr_subtypes_equivalent): Add warn and warned arguments; check main
4141 (type_variants_equivalent_p): break out from ...
4142 (odr_types_equivalent): ... here; go for main variants where needed.
4143 (warn_odr): ... here; turn static.
4144 (warn_types_mismatch): Compare mangled names of main variants.
4145 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
4146 (type_with_linkage_p): Sanity check that we look at main variant.
4147 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
4148 * tree.h (types_same_for_odr): Drop strict argument.
4150 2018-10-26 Richard Biener <rguenther@suse.de>
4152 PR tree-optimization/87746
4153 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
4154 Simplify and fix WRT strided store groups with size not
4155 equal to step in element count.
4156 (vect_analyze_group_access_1): Dump the whole group.
4158 2018-10-25 Carl Love <cel@us.ibm.com>
4160 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
4161 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
4162 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
4163 precicion entry for each overloaded builtin.
4164 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
4165 VSCEDPUO): Rename overloaded name.
4166 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
4167 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
4168 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
4169 define_expand for xscmpexqp instruction.
4170 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
4172 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
4173 Jinsong Ji <jji@us.ibm.com>
4175 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
4176 function with vec_sl.
4177 (_mm_slli_epi32): Likewise.
4178 (_mm_slli_epi64): Likewise.
4179 (_mm_srai_epi16): Replace deprecated function with vec_sra.
4180 (_mm_srai_epi32): Likewise.
4181 (_mm_srli_epi16): Replace deprecated function with vec_sr.
4182 (_mm_srli_epi32): Likewise.
4183 (_mm_srli_epi64): Likewise.
4184 (_mm_sll_epi16): Replace deprecated function with vec_sl.
4185 (_mm_sll_epi32): Likewise.
4186 (_mm_sll_epi64): Likewise.
4187 (_mm_sra_epi16): Replace deprecated function with vec_sra.
4188 (_mm_sra_epi32): Likewise.
4189 (_mm_srl_epi16): Replace deprecated function with vec_sr.
4190 (_mm_srl_epi32): Likewise.
4191 (_mm_srl_epi64): Likewise.
4193 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
4194 Jinsong Ji <jji@us.ibm.com>
4196 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
4197 comparison operators with vec_cmp* for compatibility due to
4198 unfortunate history; clean up formatting and use types more
4200 (_mm_sll_epi32): Likewise.
4201 (_mm_sll_epi64): Likewise.
4202 (_mm_srl_epi16): Likewise.
4203 (_mm_srl_epi32): Likewise.
4204 (_mm_srl_epi64): Likewise.
4206 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
4207 Jinsong Ji <jji@us.ibm.com>
4209 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
4210 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
4211 __vector __bool int. Use vec_cmpgt in preference to deprecated
4212 function vec_vcmpgtfp.
4213 (_mm_max_ps): Likewise.
4215 2018-10-25 Jeff Law <law@redhat.com>
4217 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
4220 2018-10-25 Martin Sebor <msebor@redhat.com>
4222 * doc/extend.texi (aligned): Expand attribute description.
4223 (Alignment): Rename section. Discuss function arguments.
4225 2018-10-25 Jan Hubicka <jh@suse.cz>
4227 * ipa-devirt.c (main_odr_variant): Remove.
4228 (hash_odr_name, types_same_for_odr, types_odr_comparable,
4229 odr_name_hasher::equal, odr_subtypes_equivalent_p):
4230 Drop use of main_odr_variant.
4231 (add_type_duplicate): Silence confused warnings on integer types.
4232 (get_odr_type): Always look for main variant.
4233 (register_odr_type): Simplify.
4235 2018-10-25 Richard Biener <rguenther@suse.de>
4237 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
4238 Initialize ng to silence error with release checking bootstrap.
4240 2018-10-25 Richard Biener <rguenther@suse.de>
4242 * tree-if-conv.c: Include tree-ssa-sccvn.h.
4243 (tree_if_conversion): Run CSE on the if-converted loop body.
4245 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
4247 * config/s390/constraints.md (ZL): New constraint.
4248 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
4250 * config/s390/s390.md (movdi_larl): Remove.
4251 (movdi_64): Add the LARL alternative.
4253 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
4256 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
4257 (RTX_CODE_HWINT_P): New macro.
4258 (rtx_code_size): Use RTX_CODE_HWINT_P ().
4260 2018-10-25 Jan Hubicka <jh@suse.cz>
4262 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
4265 2018-10-25 Richard Biener <rguenther@suse.de>
4267 PR tree-optimization/87665
4268 PR tree-optimization/87745
4269 * tree-vectorizer.h (get_earlier_stmt): Remove.
4270 (get_later_stmt): Pick up UID from the original non-pattern stmt.
4272 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
4274 * options.texi (Deprecated): Move list to Var section.
4276 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
4277 Jinsong Ji <jji@us.ibm.com>
4279 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
4280 __vector long to __vector long long.
4281 (_mm_cvtpd_ps): Likewise.
4282 (_mm_cvttpd_epi32): Likewise.
4283 (_mm_cvtpi32_pd): Likewise.
4284 (_mm_unpackhi_epi64): Likewise.
4285 (_mm_unpacklo_epi64): Likewise.
4287 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
4289 PR rtl-optimization/87720
4290 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
4292 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
4294 * gimple-ssa-isolate-paths.c
4295 (find_implicit_erroneous_behavior): Do not change code if the
4296 pass is running for warnings only.
4297 (find_explicit_erroneous_behavior): Likewise.
4299 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
4301 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
4302 Define as rs6000_mangle_decl_assembler_name.
4303 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
4304 long double to IEEE long double, switch the names of the long
4305 double built-in functions to be <func>f128 instead of <func>l.
4307 2018-10-24 Martin Sebor <msebor@redhat.com>
4309 * doc/extend.texi (nonnull): List no-argument form. Reference
4310 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
4312 2018-10-24 Richard Biener <rguenther@suse.de>
4314 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
4316 2018-10-24 Martin Liska <mliska@suse.cz>
4318 PR tree-optimization/84436
4319 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
4321 (switch_conversion::contains_linear_function_p): New.
4322 (switch_conversion::build_one_array): Support linear
4323 transformation on input.
4324 * tree-switch-conversion.h (struct switch_conversion): Add
4325 contains_linear_function_p declaration.
4327 2018-10-24 Richard Biener <rguenther@suse.de>
4329 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
4330 if its argument is CONSTANT_CLASS_P.
4332 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
4334 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
4335 it is wrong for forward declarations.
4337 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
4339 * config/s390/s390.c (s390_check_qrst_address): Add the missing
4340 SYMBOL_REF_P () check.
4342 2018-10-24 Richard Biener <rguenther@suse.de>
4344 PR tree-optimization/87105
4345 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
4346 dump classification.
4347 (vect_analyze_data_ref_accesses): Handle duplicate loads and
4348 stores by splitting the affected group after the fact.
4349 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
4350 fail the SLP build because of size constraints.
4352 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4354 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
4355 * configure: Regenerate.
4356 * config.in: Regenerate.
4357 * varasm.c (mergeable_string_section): Use readonly_data_section
4358 if linker doesn't support SHF_MERGE with alignment > 8.
4359 (mergeable_constant_section): Likewise.
4361 2018-10-24 Richard Biener <rguenther@suse.de>
4363 PR tree-optimization/84013
4364 * tree-ssa-structalias.c (struct msdi_data): New struct for
4365 marshalling data to walk_stmt_load_store_ops.
4366 (maybe_set_dependence_info): Refactor as callback for
4367 walk_stmt_load_store_ops.
4368 (compute_dependence_clique): Set restrict info on all stmt kinds.
4370 2018-10-24 Martin Liska <mliska@suse.cz>
4372 * cgraph.c (cgraph_node::dump):
4373 Remove reduntant dumps and make tp_first_run dump more compact.
4375 2018-10-24 Richard Biener <rguenther@suse.de>
4377 PR tree-optimization/87665
4378 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
4381 2018-10-12 Jeff Law <law@redhat.com>
4383 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
4386 2018-10-23 Richard Biener <rguenther@suse.de>
4388 * tree-vrp.c (add_assert_info): Guard dump_printf with
4390 * gimple-ssa-evrp-analyze.c
4391 (evrp_range_analyzer::record_ranges_from_incoming_edge):
4392 Use value_range::ignore_equivs_equal_p.
4394 2018-10-23 Richard Biener <rguenther@suse.de>
4396 PR tree-optimization/87105
4397 PR tree-optimization/87608
4398 * passes.def (pass_all_early_optimizations): Add early phi-opt
4400 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
4401 addition to debug stmts.
4402 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
4403 and abs replacement early.
4404 * tree-cfg.c (gimple_empty_block_p): Likewise.
4406 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
4409 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
4410 specified to configure.
4411 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
4413 2018-10-23 Richard Biener <rguenther@suse.de>
4415 PR tree-optimization/87700
4416 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
4418 2018-10-23 Jakub Jelinek <jakub@redhat.com>
4421 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
4422 second argument from __mmask16 to __mmask8.
4423 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
4424 _mm_mask_packs_epi32): Likewise.
4425 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
4427 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
4429 2018-10-23 Richard Biener <rguenther@suse.de>
4431 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
4433 2018-10-23 Richard Biener <rguenther@suse.de>
4435 PR tree-optimization/86144
4436 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
4437 over simd attribute.
4439 2018-10-23 Richard Biener <rguenther@suse.de>
4441 PR tree-optimization/87693
4442 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
4443 the case we do not find the taken edge.
4445 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
4446 Jinsong Ji <jji@us.ibm.com>
4448 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
4449 (_mm_store_pd): Use unaligned vector type for pointer cast.
4450 (_mm_maskmoveu_si128): Likewise.
4451 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
4452 (_mm_store_ps): Use unaligned vector type for pointer cast.
4454 2018-10-22 Paul Koning <ni1d@arrl.net>
4456 * symtab.c (symtab_node::increase_alignment): Correct max
4459 2018-10-22 Yury Gribov <tetra2005@gmail.com>
4461 PR tree-optimization/87633
4462 * match.pd: Do not generate unordered integer comparisons.
4464 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
4466 PR rtl-optimization/87600
4467 * combine.c: Add include of expr.h.
4468 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
4469 register to a pseudo.
4470 (make_more_copies): New function, add a copy to a new pseudo after
4471 the moves from hard registers into pseudos.
4472 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
4473 later. Call make_more_copies.
4475 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
4477 * lra-constraints.c (process_alt_operands): New local array,
4478 matching_early_clobber. Check matching_early_clobber before
4479 decrementing reject, and set matching_early_clobber after.
4481 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
4484 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
4485 call output_addr_const and hope for the best.
4487 2018-10-22 Richard Biener <rguenther@suse.de>
4489 * gimple-ssa-evrp-analyze.c
4490 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
4491 smarter about what ranges to use.
4492 * tree-vrp.c (add_assert_info): Dump here.
4493 (register_edge_assert_for_2): Instead of here at multiple but
4496 * gcc.dg/tree-ssa/evrp12.c: New testcase.
4497 * gcc.dg/predict-6.c: Adjust.
4498 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
4499 * gcc.dg/tree-ssa/vrp02.c: Likewise.
4500 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
4502 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
4503 Richard Biener <rguenther@suse.de>
4505 * bitmap.h: Update data structure documentation, including a
4506 description of bitmap views as either linked-lists or splay trees.
4507 (struct bitmap_element_def): Update comments for splay tree bitmaps.
4508 (struct bitmap_head_def): Likewise.
4509 (bitmap_list_view, bitmap_tree_view): New prototypes.
4510 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
4512 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
4513 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
4514 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
4515 released bitmap element here.
4516 (bitmap_element_free): Remove.
4517 (bitmap_elt_clear_from): Work on splay tree bitmaps.
4518 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
4519 this function similar ones such that linked-list bitmap implementation
4520 functions are grouped.
4521 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
4522 and moved for grouping.
4523 (bitmap_list_insert_element_after): Renamed from
4524 bitmap_elt_insert_after, and moved for grouping.
4525 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
4526 (bitmap_tree_link_left, bitmap_tree_link_right,
4527 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
4528 bitmap_tree_link_element, bitmap_tree_unlink_element,
4529 bitmap_tree_find_element): New functions for splay-tree bitmap
4531 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
4532 Renamed and moved, see above entries.
4533 (bitmap_tree_listify_from): New function to convert part of a splay
4534 tree bitmap to a linked-list bitmap.
4535 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
4536 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
4537 (bitmap_find_bit): Remove.
4538 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
4539 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
4540 Handle splay tree bitmaps.
4541 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
4542 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
4543 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
4544 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
4545 bitmap_intersect_compl_p, bitmap_ior_and_compl,
4546 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
4547 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
4548 corresponding changes to use linked-list specific bitmap_element
4549 manipulation functions as applicable for efficiency.
4550 (bitmap_tree_to_vec): New function.
4551 (debug_bitmap_elt_file): New function split out from ...
4552 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
4553 (bitmap_print): Likewise.
4555 PR tree-optimization/63155
4556 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
4558 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
4561 2018-10-22 Martin Liska <mliska@suse.cz>
4563 PR tree-optimization/87686
4565 2018-08-29 Martin Liska <mliska@suse.cz>
4567 * tree-switch-conversion.c (switch_conversion::expand):
4568 Strenghten assumption about gswitch statements.
4570 2018-10-22 Martin Liska <mliska@suse.cz>
4572 * ipa-icf.c (sem_item::compare_attributes): Remove.
4573 (sem_item::compare_referenced_symbol_properties): Use
4574 attribute_list_equal instead.
4575 (sem_function::equals_wpa): Likewise.
4576 * ipa-icf.h: Remove compare_attributes.
4578 2018-10-22 Richard Biener <rguenther@suse.de>
4581 * mem-stats.h (mem_usage::operator==): Fix pasto.
4583 2018-10-22 Richard Biener <rguenther@suse.de>
4585 PR tree-optimization/87640
4586 * tree-vrp.c (set_value_range_with_overflow): Decompose
4588 (extract_range_from_binary_expr_1): Adjust.
4590 2018-10-22 Martin Jambor <mjambor@suse.cz>
4592 * tree-eh.h (stmt_could_throw_p): Add function parameter.
4593 (stmt_can_throw_external): Likewise.
4594 (stmt_can_throw_internal): Likewise.
4595 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
4596 (lower_eh_constructs_2): Likewise.
4597 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
4598 (stmt_can_throw_external): Likewise.
4599 (stmt_can_throw_internal): Likewise.
4600 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
4601 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
4602 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
4603 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
4604 (pass_lower_eh_dispatch::execute): Pass cfun to
4605 stmt_can_throw_external.
4606 (cleanup_empty_eh): Likewise.
4607 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
4608 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
4609 stmt_can_throw_external instead of pushing it to cfun.
4610 (symbol_table::create_edge): Likewise.
4611 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
4612 stmt_can_throw_internal.
4613 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
4614 to stmt_could_throw_p.
4615 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
4616 stmt_can_throw_internal.
4617 (pass_store_merging::execute): Likewise.
4618 * gimple-ssa-strength-reduction.c
4619 (find_candidates_dom_walker::before_dom_children): Pass cfun to
4621 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
4622 stmt_can_throw_internal.
4623 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
4624 to stmt_can_throw_external.
4625 (check_stmt): Pass cfun to stmt_could_throw_p.
4626 (check_stmt): Pass cfun to stmt_can_throw_external.
4627 (pass_nothrow::execute): Likewise.
4628 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
4629 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
4630 stmt_can_throw_internal.
4631 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
4632 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
4633 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
4634 * tree-complex.c (expand_complex_libcall): Pass cfun to
4635 stmt_could_throw_p and to stmt_can_throw_internal.
4636 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
4637 * tree-inline.c (copy_edges_for_bb): Likewise.
4638 (maybe_move_debug_stmts_to_successors): Likewise.
4639 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
4641 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
4642 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
4643 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
4644 stmt_can_throw_internal.
4645 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
4646 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
4648 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
4649 stmt_can_throw_internal.
4650 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
4651 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
4653 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
4654 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
4655 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
4656 (convert_mult_to_fma_1): Likewise.
4657 (convert_to_divmod): Likewise.
4658 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
4659 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
4660 * tree-ssa-propagate.c
4661 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
4662 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
4663 (maybe_optimize_range_tests): Likewise.
4664 (linearize_expr_tree): Likewise.
4665 (reassociate_bb): Likewise.
4666 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
4667 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
4668 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
4669 (handle_char_store): Likewise.
4670 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
4671 stmt_can_throw_internal.
4672 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
4674 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
4675 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
4676 (vectorizable_simd_clone_call): Likewise.
4677 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
4678 (gimple_stringop_fixed_value): Likewise.
4680 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
4682 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
4683 literal pool references.
4684 (s390_check_qrst_address): Adapt to the new behavior of
4685 s390_loadrelative_operand_p ().
4687 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
4690 * config/i386/sse.md (*andnot<mode>3_bcst): New.
4692 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
4695 * config/i386/sse.md (*<code><mode>3_bcst): New.
4697 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
4700 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
4701 V4DI, V16SI and V8DI.
4702 (*sub<mode>3<mask_name>_bcst): New.
4703 (*add<mode>3<mask_name>_bcst): Likewise.
4705 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
4706 Jinsong Ji <jji@us.ibm.com>
4708 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
4709 __m64 with __vector unsigned long long for compatibility.
4710 (_mm_movemask_epi8): Likewise.
4711 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
4712 (_mm_cvttps_pi32): Likewise.
4713 (_mm_cvtpi32_ps): Likewise.
4714 (_mm_cvtps_pi16): Likewise.
4715 (_mm_loadh_pi): Likewise.
4716 (_mm_storeh_pi): Likewise.
4717 (_mm_movehl_ps): Likewise.
4718 (_mm_movelh_ps): Likewise.
4719 (_mm_loadl_pi): Likewise.
4720 (_mm_storel_pi): Likewise.
4721 (_mm_movemask_ps): Likewise.
4722 (_mm_shuffle_pi16): Likewise.
4724 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
4727 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
4728 __builtin_ia32_vfnmsubpd512_mask.
4729 (_mm512_mask_fnmsub_round_pd): Likewise.
4730 (_mm512_fnmsub_pd): Likewise.
4731 (_mm512_mask_fnmsub_pd): Likewise.
4732 (_mm512_maskz_fnmsub_round_pd): Use
4733 __builtin_ia32_vfnmsubpd512_maskz.
4734 (_mm512_maskz_fnmsub_pd): Likewise.
4735 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
4736 (_mm512_mask_fnmsub_round_ps): Likewise.
4737 (_mm512_fnmsub_ps): Likewise.
4738 (_mm512_mask_fnmsub_ps): Likewise.
4739 (_mm512_maskz_fnmsub_round_ps): Use
4740 __builtin_ia32_vfnmsubps512_maskz.
4741 (_mm512_maskz_fnmsub_ps): Likewise.
4742 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
4743 __builtin_ia32_vfnmsubpd256_mask.
4744 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
4745 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
4746 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
4747 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
4748 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
4749 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
4750 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
4751 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
4752 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
4753 __builtin_ia32_vfnmsubpd.
4754 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
4755 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
4756 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
4757 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
4758 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
4759 * config/i386/i386-builtin.def: Add
4760 __builtin_ia32_vfnmsubpd256_mask,
4761 __builtin_ia32_vfnmsubpd256_maskz,
4762 __builtin_ia32_vfnmsubpd128_mask,
4763 __builtin_ia32_vfnmsubpd128_maskz,
4764 __builtin_ia32_vfnmsubps256_mask,
4765 __builtin_ia32_vfnmsubps256_maskz,
4766 __builtin_ia32_vfnmsubps128_mask,
4767 __builtin_ia32_vfnmsubps128_maskz,
4768 __builtin_ia32_vfnmsubpd512_mask,
4769 __builtin_ia32_vfnmsubpd512_maskz,
4770 __builtin_ia32_vfnmsubps512_mask,
4771 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
4772 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
4773 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
4774 __builtin_ia32_vfnmsubpd256.
4775 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
4776 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
4777 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
4779 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
4781 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
4783 (fmai_vmfnmsub_<mode><round_name>): Likewise.
4785 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
4788 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
4789 __builtin_ia32_vfnmaddpd512_mask.
4790 (_mm512_mask_fnmadd_round_pd): Likewise.
4791 (_mm512_fnmadd_pd): Likewise.
4792 (_mm512_mask_fnmadd_pd): Likewise.
4793 (_mm512_maskz_fnmadd_round_pd): Use
4794 __builtin_ia32_vfnmaddpd512_maskz.
4795 (_mm512_maskz_fnmadd_pd): Likewise.
4796 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
4797 (_mm512_mask_fnmadd_round_ps): Likewise.
4798 (_mm512_fnmadd_ps): Likewise.
4799 (_mm512_mask_fnmadd_ps): Likewise.
4800 (_mm512_maskz_fnmadd_round_ps): Use
4801 __builtin_ia32_vfnmaddps512_maskz.
4802 (_mm512_maskz_fnmadd_ps): Likewise.
4803 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
4804 __builtin_ia32_vfnmaddpd256_mask.
4805 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
4806 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
4807 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
4808 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
4809 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
4810 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
4811 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
4812 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
4813 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
4814 __builtin_ia32_vfnmaddpd.
4815 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
4816 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
4817 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
4818 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
4819 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
4820 * config/i386/i386-builtin.def: Add
4821 __builtin_ia32_vfnmaddpd256_mask,
4822 __builtin_ia32_vfnmaddpd256_maskz,
4823 __builtin_ia32_vfnmaddpd128_mask,
4824 __builtin_ia32_vfnmaddpd128_maskz,
4825 __builtin_ia32_vfnmaddps256_mask,
4826 __builtin_ia32_vfnmaddps256_maskz,
4827 __builtin_ia32_vfnmaddps128_mask,
4828 __builtin_ia32_vfnmaddps128_maskz,
4829 __builtin_ia32_vfnmaddpd512_mask,
4830 __builtin_ia32_vfnmaddpd512_maskz,
4831 __builtin_ia32_vfnmaddps512_mask,
4832 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
4833 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
4834 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
4835 __builtin_ia32_vfnmaddpd256.
4836 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
4837 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
4838 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
4840 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
4842 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
4844 (fmai_vmfnmadd_<mode><round_name>): Likewise.
4846 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
4849 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
4850 __builtin_ia32_vfmsubpd512_mask.
4851 (_mm512_mask_fmsub_round_pd): Likewise.
4852 (_mm512_fmsub_pd): Likewise.
4853 (_mm512_mask_fmsub_pd): Likewise.
4854 (_mm512_maskz_fmsub_round_pd): Use
4855 __builtin_ia32_vfmsubpd512_maskz.
4856 (_mm512_maskz_fmsub_pd): Likewise.
4857 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
4858 (_mm512_mask_fmsub_round_ps): Likewise.
4859 (_mm512_fmsub_ps): Likewise.
4860 (_mm512_mask_fmsub_ps): Likewise.
4861 (_mm512_maskz_fmsub_round_ps): Use
4862 __builtin_ia32_vfmsubps512_maskz.
4863 (_mm512_maskz_fmsub_ps): Likewise.
4864 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
4865 __builtin_ia32_vfmsubpd256_mask.
4866 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
4867 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
4868 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
4869 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
4870 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
4871 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
4872 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
4873 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
4874 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
4875 __builtin_ia32_vfmsubpd.
4876 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
4877 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
4878 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
4879 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
4880 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
4881 * config/i386/i386-builtin.def: Add
4882 __builtin_ia32_vfmsubpd256_mask,
4883 __builtin_ia32_vfmsubpd256_maskz,
4884 __builtin_ia32_vfmsubpd128_mask,
4885 __builtin_ia32_vfmsubpd128_maskz,
4886 __builtin_ia32_vfmsubps256_mask,
4887 __builtin_ia32_vfmsubps256_maskz,
4888 __builtin_ia32_vfmsubps128_mask,
4889 __builtin_ia32_vfmsubps128_maskz,
4890 __builtin_ia32_vfmsubpd512_mask,
4891 __builtin_ia32_vfmsubpd512_maskz,
4892 __builtin_ia32_vfmsubps512_mask,
4893 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
4894 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
4895 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
4896 __builtin_ia32_vfmsubpd256.
4897 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
4898 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
4899 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
4901 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
4903 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
4905 (fmai_vmfmsub_<mode><round_name>): Likewise.
4907 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
4909 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
4910 Remove plus. Renamed to ...
4911 (*sub<mode>3<mask_name>_bcst): This.
4912 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
4913 (*add<mode>3<mask_name>_bcst): This.
4915 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
4918 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
4920 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
4923 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
4924 (_mm_or_epi32): Likewise.
4925 (_mm256_xor_epi32): Likewise.
4926 (_mm_xor_epi32): Likewise.
4927 (_mm256_or_epi64): Likewise.
4928 (_mm_or_epi64): Likewise.
4929 (_mm256_xor_epi64): Likewise.
4930 (_mm_xor_epi64): Likewise.
4932 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
4935 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
4937 2018-10-20 Jakub Jelinek <jakub@redhat.com>
4940 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
4942 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
4944 * doc/ux.texi: Move @section directly after @node.
4946 2018-10-19 Jakub Jelinek <jakub@redhat.com>
4950 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
4951 depend closely nested inside of loop with ordered clause with
4954 2018-10-19 David Malcolm <dmalcolm@redhat.com>
4956 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
4957 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
4958 * doc/ux.texi: New file.
4960 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4962 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
4963 be the first CR field allocated.
4965 2018-10-19 Richard Biener <rguenther@suse.de>
4968 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
4969 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
4971 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
4974 * config/i386/sse.md
4975 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
4976 (*add<mode>3<mask_name>_bcst_2): Likewise.
4978 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
4980 * config/i386/sse.md
4981 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
4982 Replace nonimmediate_operand with register_operand.
4983 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
4985 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
4988 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
4990 PR rtl-optimization/87596
4991 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
4992 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
4993 for instructions in FROM..TO range.
4995 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
4997 * cfgexpand.c (expand_one_var): Use specific wording in error message
4998 for non-local frame variables.
4999 * stor-layout.c (layout_decl): Do not issue a warning for them.
5001 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
5003 * haifa-sched.c (priority): Add force_recompute parameter.
5004 (apply_replacement): Call priority () with force_recompute = true.
5005 (restore_pattern): Likewise.
5007 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
5009 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
5010 HOST_BITS_PER_WIDE_INT.
5011 (test_vector_ops_duplicate): Likewise.
5013 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
5016 * config/i386/sse.md (VF_AVX512): New.
5017 (avx512bcst): Likewise.
5018 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
5020 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
5022 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
5025 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
5027 * doc/invoke.texi (-dumpversion): Improve grammar.
5028 (-dumpfullversion): Make more consistent with -dumpversion.
5030 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
5032 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
5033 Set JUMP_LABEL to the jump insn.
5034 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
5035 Predict emitted jump and add label to jump insn.
5037 2018-10-18 David Malcolm <dmalcolm@redhat.com>
5039 PR tree-optimization/87562
5040 * input.c (get_substring_ranges_for_loc): Use
5041 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
5042 getting the linemap for the endpoint. Verify that it's either
5043 in the same linemap as the start point's spelling location, or
5044 at least in the same file.
5046 2018-10-18 Richard Biener <rguenther@suse.de>
5048 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
5049 feed width-specific load/store costs through ix86_vec_cost.
5050 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
5051 (k8_cost): Likewise.
5052 (bdver_cost): Likewise.
5053 (znver1_cost): Likewise.
5054 (btver1_cost): Likewise.
5055 (btver2_cost): Likewise.
5057 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
5059 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
5060 to simplify subreg of vec_merge.
5062 2018-10-18 Richard Biener <rguenther@suse.de>
5064 * config/i386/i386.c: Fix costing of vector FMA.
5066 2018-10-18 Richard Biener <rguenther@suse.de>
5068 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
5070 (ix86_builtin_vectorization_cost): For vec_construct properly
5071 cost insertion into SSE regs.
5072 (...): Adjust calls to ix86_vec_cost.
5074 2018-10-18 Richard Biener <rguenther@suse.de>
5078 2018-02-07 Richard Biener <rguenther@suse.de>
5080 PR tree-optimization/84204
5081 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
5084 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
5087 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
5089 (test_vector_ops_duplicate): Add test for a scalar subreg of a
5090 VEC_MERGE of a VEC_DUPLICATE.
5092 2018-10-17 Joseph Myers <joseph@codesourcery.com>
5094 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
5095 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
5096 * doc/standards.texi (C Language): Document C2X.
5097 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
5098 (rl78_option_override): Handle "GNU C2X" language name.
5100 2018-10-17 Joseph Myers <joseph@codesourcery.com>
5102 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
5103 Document C17 as published in 2018.
5105 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
5108 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
5109 bail out if both sides do not have the same storage order.
5111 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
5113 * bitmap.c (bitmap_head::dump): New.
5114 * bitmap.h (bitmap_head): Add dump().
5115 * gimple-ssa-evrp-analyze.c
5116 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
5117 (evrp_range_analyzer::set_ssa_range_info): Same.
5118 (evrp_range_analyzer::record_ranges_from_phis): Same.
5119 (evrp_range_analyzer::record_ranges_from_stmt): Same.
5120 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
5121 * gimple-ssa-sprintf.c (get_int_range): Same.
5122 (format_integer): Same.
5123 (sprintf_dom_walker::handle_gimple_call): Same.
5124 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
5125 (ipcp_vr_lattice::top_p): Same.
5126 (ipcp_vr_lattice::bottom_p): Same.
5127 (ipcp_vr_lattice::set_to_bottom): Same.
5128 (ipa_vr_operation_and_type_effects): Same.
5129 (propagate_vr_across_jump_function): Same.
5130 (ipcp_store_vr_results): Same.
5131 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
5132 (ipa_print_node_jump_functions_for_edge): Same.
5133 (ipa_get_value_range): Same.
5134 (ipa_compute_jump_functions_for_edge): Same.
5135 (ipa_write_jump_function): Same.
5136 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
5137 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
5139 * vr-values.c (set_value_range_to_nonnegative): Same.
5140 (set_value_range_to_truthvalue): Same.
5141 (vr_values::get_value_range): Same.
5142 (vr_values::set_defs_to_varying): Same.
5143 (vr_values::update_value_range): Same.
5144 (symbolic_range_based_on_p): Same.
5145 (vr_values::op_with_boolean_value_range_p): Same.
5146 (vr_values::extract_range_for_var_from_comparison_expr): Same.
5147 (vr_values::extract_range_from_ssa_name): Same.
5148 (vr_values::extract_range_from_binary_expr): Same.
5149 (vr_values::extract_range_from_unary_expr): Same.
5150 (vr_values::extract_range_from_cond_expr): Same.
5151 (vr_values::extract_range_from_comparison): Same.
5152 (vr_values::check_for_binary_op_overflow): Same.
5153 (vr_values::extract_range_basic): Same.
5154 (vr_values::extract_range_from_assignment): Same.
5155 (compare_ranges): Same.
5156 (compare_range_with_value): Same.
5157 (vr_values::adjust_range_with_scev): Same.
5158 (vrp_valueize): Same.
5159 (vrp_valueize_1): Same.
5160 (vr_values::get_vr_for_comparison): Same.
5161 (vr_values::compare_name_with_value): Same.
5162 (vr_values::compare_names): Same.
5163 (vr_values::vrp_evaluate_conditional): Same.
5164 (find_case_label_ranges): Same.
5165 (vr_values::vrp_visit_switch_stmt): Same.
5166 (vr_values::extract_range_from_phi_node): Same.
5167 (vr_values::simplify_div_or_mod_using_ranges): Same.
5168 (vr_values::simplify_bit_ops_using_ranges): Same.
5169 (test_for_singularity): Same.
5170 (range_fits_type_p): Same.
5171 (vr_values::simplify_cond_using_ranges_1): Same.
5172 (vr_values::simplify_switch_using_ranges): Same.
5173 (vr_values::simplify_float_conversion_using_ranges): Same.
5174 (vr_values::two_valued_val_range_p): Same.
5175 (vr_values::add_equivalence): Move to value_range::equiv_add.
5176 * vr-values.h (vr_values::add_equivalence): Remove.
5177 (VR_INITIALIZER): Remove.
5178 * tree-vrp.c (value_range::set): New.
5179 (value_range::equiv_add): New.
5180 (value_range::value_range): New.
5181 (value_range::deep_copy): New.
5182 (value_range::check): New.
5183 (value_range::equal_p): New.
5184 (value_range::ignore_equivs_equal_p): New.
5185 (value_range::operator==): New.
5186 (value_range::operator!=): New.
5187 (value_range::symbolic_p): New.
5188 (value_range::numeric_p): New.
5189 (value_range::set_undefined): New.
5190 (value_range::set_varying): New.
5191 (value_range::may_contain_p): New.
5192 (value_range::equiv_clear): New.
5193 (value_range::singleton_p): New.
5194 (value_range::intersect): New.
5195 (value_range::dump): New.
5196 (value_range::set_and_canonicalize): New.
5197 (set_value_range): Adjust for value_range API.
5198 (set_value_range_to_undefined): Same.
5199 (set_value_range_to_varying): Same.
5200 (set_and_canonicalize_value_range): Same.
5201 (set_value_range_to_nonnull): Same.
5202 (set_value_range_to_null): Same.
5203 (range_is_null): Same.
5204 (range_is_nonnull): Same.
5205 (range_int_cst_p): Same.
5206 (range_int_cst_singleton_p): Same.
5207 (symbolic_range_p): Same.
5208 (range_includes_zero_p): Same.
5209 (value_range_constant_singleton): Same.
5210 (vrp_set_zero_nonzero_bits): Same.
5211 (ranges_from_anti_range): Same.
5212 (extract_range_into_wide_ints): Same.
5213 (extract_range_from_multiplicative_op): Same.
5214 (set_value_range_with_overflow): Same.
5215 (extract_range_from_binary_expr_1): Same.
5216 (extract_range_from_unary_expr): Same.
5217 (dump_value_range): Same.
5218 (debug_value_range): Same.
5219 (vrp_prop::check_array_ref): Same.
5220 (vrp_prop::check_mem_ref): Same.
5221 (vrp_prop::vrp_initialize): Same.
5222 (vrp_prop::visit_stmt): Same.
5223 (intersect_ranges): Same.
5224 (vrp_prop::visit_phi): Same.
5225 (vrp_prop::vrp_finalize): Same.
5226 (determine_value_range_1): Same.
5227 (determine_value_range): Same.
5228 (vrp_intersect_ranges_1): Rename to...
5229 (vrp_intersect_1): this.
5230 (vrp_intersect_ranges): Rename to...
5231 (value_range::intersect_helper): ...this.
5232 (vrp_meet_1): Rename to...
5233 (value_range::union_helper): ...this.
5234 (vrp_meet): Rename to...
5235 (value_range::union_): ...this.
5236 (copy_value_range): Remove.
5237 * tree-vrp.h (struct value_range): Rewrite into a proper class.
5238 (value_range::vrtype): New.
5239 (value_range::type): New.
5240 (value_range::equiv): New.
5241 (value_range::min): New.
5242 (value_range::max): New.
5243 (value_range::varying_p): New.
5244 (value_range::undefined_p): New.
5245 (value_range::null_p): New.
5246 (value_range::equiv_add): New.
5247 (copy_value_range): Remove.
5249 2018-10-17 David Malcolm <dmalcolm@redhat.com>
5251 * Makefile.in (SELFTEST_TARGETS): New.
5252 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
5253 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
5254 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
5256 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
5257 (selftest-c++-gdb, selftest-c++-valgrind): Move to
5259 * configure: Regenerate.
5260 * configure.ac (selftest_languages): New.
5262 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
5264 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
5265 overflow wraps argument.
5266 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
5267 wide_int_range_multiplicative_op.
5268 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
5269 overflow wraps argument.
5270 (wide_int_range_multiplicative_op): Same.
5271 (wide_int_range_lshift): Same.
5272 (wide_int_range_div): Same.
5273 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
5274 (wide_int_range_lshift): Same.
5275 (wide_int_range_div): Same.
5277 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
5279 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
5280 use sign as argument.
5281 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
5282 wide_int_range_shift_undefined_p.
5284 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5286 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
5288 (@despeculate_copy<ALLI_TI:mode>): ... This.
5289 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
5292 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5294 * config.gcc: Obsolete *-*-solaris2.10*.
5295 * doc/install.texi (Specific, *-*-solaris2*): Document it.
5297 2018-10-12 Jeff Law <law@redhat.com>
5299 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
5300 reg + sym +- const_int addressing modes.
5302 2018-10-15 David Malcolm <dmalcolm@redhat.com>
5304 * common.opt (fdiagnostics-minimum-margin-width=): New option.
5305 * diagnostic-show-locus.c (layout::layout): Apply the minimum
5307 (layout::start_annotation_line): Only print up to 3 of the
5308 margin character, to avoid touching the left-hand side.
5309 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
5310 minimum margin width, as set by test_diagnostic_context's ctor.
5311 (selftest::test_fixit_insert_containing_newline): Likewise.
5312 (selftest::test_fixit_insert_containing_newline_2): Likewise.
5313 (selftest::test_line_numbers_multiline_range): Clear
5314 dc.min_margin_width.
5315 * diagnostic.c (diagnostic_initialize): Initialize
5317 * diagnostic.h (struct diagnostic_context): Add field
5319 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
5320 * opts.c (common_handle_option): Handle
5321 OPT_fdiagnostics_minimum_margin_width_.
5322 * selftest-diagnostic.c
5323 (selftest::test_diagnostic_context::test_diagnostic_context):
5324 Initialize min_margin_width to 6.
5325 * toplev.c (general_init): Initialize global_dc->min_margin_width.
5327 2018-10-15 David Malcolm <dmalcolm@redhat.com>
5329 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
5330 Fix usage of "error_at_rich_loc" in the comment.
5332 2018-10-15 Renlin Li <renlin.li@arm.com>
5335 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
5336 if-conversioned loop when it contains ifn with types not
5337 supported by backend.
5338 * internal-fn.c (expand_direct_optab_fn): Add an assert.
5339 (direct_internal_fn_supported_p): New helper function.
5340 * internal-fn.h (direct_internal_fn_supported_p): Declare.
5342 2018-10-15 Jakub Jelinek <jakub@redhat.com>
5345 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
5346 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
5347 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
5349 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
5351 PR tree-optimization/87022
5352 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
5353 bits in dist vector rather than the first one.
5355 2018-10-15 Richard Biener <rguenther@suse.de>
5358 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
5359 (visit_loadstore): When a used restrict tag escaped verify that
5360 the points-to solution of "other" pointers do not include
5362 (compute_dependence_clique): If a used restrict tag escaped
5363 communicated that down to visit_loadstore.
5365 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
5367 * config/s390/s390.c (s390_expand_vec_init): Force vector element
5368 into reg if it isn't a general operand.
5370 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
5373 * config/i386/sse.md (*vec_dupv2di): Add register source to
5376 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
5379 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
5380 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
5381 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
5382 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
5384 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
5386 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
5387 (notice_args_size): Set it in the current trace if no insn that can
5388 throw internally has been seen yet.
5389 (connect_traces): When connecting args_size between traces, allow the
5390 incoming values not to match if there is an insn setting it before the
5391 first insn that can throw internally; in that case, force the creation
5392 of a CFI note on this latter insn.
5394 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
5396 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
5397 the base class, because using the injected-class-name was not clearly
5398 specified until DR 176.
5400 2018-10-12 Paul Koning <ni1d@arrl.net>
5402 * config/pdp11/pdp11.md (doloop_end): New expander.
5403 (doloop_end_insn): renamed from "doloop_end".
5404 (addqi3): New pattern.
5405 (subqi3): New pattern.
5406 * config/pdp11/predicates.md (incdec_operand): New predicate.
5408 2018-10-12 Yury Gribov <tetra2005@gmail.com>
5411 * real.c (format_helper::can_represent_integral_type_p): New function
5412 * real.h (format_helper::can_represent_integral_type_p): Ditto.
5413 * match.pd: New pattern.
5415 2018-10-12 Alexandre Oliva <oliva@adacore.com>
5417 * configure.ac: Introduce --enable-large-address-aware
5418 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
5419 * doc/install.texi: Document it.
5420 * configure, config.in: Rebuilt.
5421 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
5422 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
5423 (LINK_SPEC): Insert it.
5424 * config/i386/mingw-w64.h: Likewise.
5426 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
5428 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
5430 PR rtl-optimization/87600
5431 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
5433 2018-10-12 Paul Koning <ni1d@arrl.net>
5435 * doc/md.texi (doloop_end): Document that the pattern code may
5436 need to check operand mode.
5438 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
5440 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
5441 to zero-extend between int and floating-point registers.
5442 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
5443 ldp into floating-point registers. Add type and arch attributes.
5444 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
5445 Use f_loads for type attribute.
5447 2018-10-11 Martin Sebor <msebor@redhat.com>
5449 * doc/extend.texi (attribute packed): Correct typos.
5451 2018-10-11 Martin Sebor <msebor@redhat.com>
5453 * doc/extend.texi (attribute flatten): Mention interaction with
5456 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
5459 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
5460 Set new_decl virtual flag to zero.
5462 2018-10-11 Martin Sebor <msebor@redhat.com>
5465 * doc/extend.texi (attribute format_arg): Discuss using multiple
5466 attributes on a single function.
5468 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
5470 PR tree-optimization/86829
5471 * match.pd (sin (atan (x))): New simplification rules.
5472 (cos (atan (x))): Likewise.
5473 * real.c (build_sinatan_real): New function.
5474 * real.h (build_sinatan_real): Prototype.
5476 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
5478 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
5480 (fold_mergeeo_helper): New helper function.
5481 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
5482 intrinsics. Correct some whitespace indentation issues.
5484 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
5487 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
5488 Use HOST_WIDE_INT_1U for shift.
5490 2018-10-11 Doug Rupp <rupp@adacore.com>
5491 Olivier Hainque <hainque@adacore.com>
5493 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
5494 Pass --relax to the linker for RTPs.
5495 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
5497 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
5498 Jan Hubicka <jh@suse.cz>
5499 Martin Jambor <mjambor@suse.cz>
5501 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
5502 the same elements are repeated rather than printing all of them.
5503 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
5504 "repeated" elements.
5505 * read-rtl-function.c (test_loading_repeat): New function.
5506 (read_rtl_function_c_tests): Call test_loading_repeat.
5507 * rtl-tests.c (test_dumping_repeat): New function.
5508 (rtl_tests_c_tests): Call test_dumping_repeat.
5510 2018-10-11 Richard Biener <rguenther@suse.de>
5512 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
5513 bdver?_cost): Unify to ...
5514 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
5515 * config/i386/i386.c (processor_cost_table): Adjust.
5517 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
5520 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
5521 the thunk when expanding to GIMPLE.
5523 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
5525 * varasm.c (mergeable_string_section): Don't try to move zero-length
5526 strings to the merge section.
5528 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
5531 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
5533 2018-10-10 Jakub Jelinek <jakub@redhat.com>
5536 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
5537 to special_args set.
5539 2018-10-10 Richard Biener <rguenther@suse.de>
5541 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
5542 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
5543 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
5544 and recursing and pattern terminating the recursion on SSE
5545 vector width using ix86_expand_reduc.
5546 (reduc_sminmax_scal_<mode>): Split into part reducing to half
5547 width and recursing and SSE2 vector variant doing the final
5548 reduction with ix86_expand_reduc.
5549 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
5550 with terminating the recursion at AVX level, splitting that
5553 2018-10-09 David Malcolm <dmalcolm@redhat.com>
5555 * genmatch.c (error_cb): Rename to...
5556 (diagnostic_cb): ...this, converting int params to enums.
5557 (fatal_at): Update for renaming.
5558 (warning_at): Likewise.
5560 * input.c (selftest::ebcdic_execution_charset::apply):
5561 Update for renaming of...
5562 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
5564 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
5565 converting level and reason to enums.
5566 (class selftest::lexer_error_sink): Rename to...
5567 (class selftest::lexer_test_options): ...this, renaming field
5568 "m_errors" to "m_diagnostics".
5569 (selftest::lexer_test_options::apply): Update for renaming of...
5570 (selftest::lexer_test_options::on_error): ...this, renaming to...
5571 (selftest::lexer_test_options::on_diagnostic): ...this
5572 converting level and reason to enums.
5573 (selftest::test_lexer_string_locations_raw_string_unterminated):
5574 Update for renamings.
5575 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
5578 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
5580 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
5581 * config/rs6000/pmmintrin.h: New file.
5583 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
5585 PR tree-optimization/86659
5586 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
5588 2018-10-09 Richard Biener <rguenther@suse.de>
5590 PR tree-optimization/63155
5591 * tree-ssa-structalias.c: Include tree-ssa.h.
5592 (get_constraint_for_ssa_var): For undefs return nothing_id.
5593 (find_func_aliases): Cleanup PHI handling.
5595 2018-10-09 Richard Biener <rguenther@suse.de>
5597 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
5600 2018-10-09 Martin Liska <mliska@suse.cz>
5602 * asan.c (asan_emit_stack_protection): If a stack variable
5603 is located in a same file as current function, then emit
5604 line info into variable definition string.
5606 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
5608 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
5611 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
5613 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
5616 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
5618 PR tree-optimization/86659
5619 * gimple-match.h (struct gimple_match_op): Add reverse field.
5620 (gimple_match_op::set_op): New overloaded method.
5621 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
5622 the REF_REVERSE_STORAGE_ORDER flag on the value.
5623 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
5624 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
5626 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
5629 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
5630 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
5632 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
5635 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
5636 Defined with __builtin_ia32_vfmaddsubpd512_mask.
5638 2018-10-08 Richard Biener <rguenther@suse.de>
5640 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
5641 cost the same as AVX128 ones.
5643 2018-10-08 Paul Koning <ni1d@arrl.net>
5645 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
5646 (expand_block_move): New function.
5647 * config/pdp11/pdp11.c (output_block_move): Remove.
5648 (expand_block_move): New function.
5649 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
5650 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
5651 (*movmemhi1): Remove.
5653 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
5655 * config/s390/2827.md: Increase latencies for some FP instructions.
5657 2018-10-08 Richard Biener <rguenther@suse.de>
5659 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
5661 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
5662 * tree-vectorizer.h (dump_stmt_cost): Adjust.
5663 (add_stmt_cost): Dump return value of the hook.
5665 2018-10-08 Richard Biener <rguenther@suse.de>
5667 PR tree-optimization/63155
5668 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
5669 (ssa_propagation_engine::ssa_propagate): Remove redundant
5670 bitmap bit clearing.
5672 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
5674 PR rtl-optimization/86939
5675 PR rtl-optimization/87479
5676 * ira.h (non_conflicting_reg_copy_p): New prototype.
5677 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
5678 (make_hard_regno_dead): Don't add conflicts for register
5679 ignore_reg_for_conflicts.
5680 (make_object_dead): Likewise.
5681 (non_conflicting_reg_copy_p): New function.
5682 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
5683 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
5684 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
5685 (make_hard_regno_dead): Don't add conflicts for register
5686 ignore_reg_for_conflicts. Remove special conflict handling of
5687 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
5688 check_pic_pseudo_p and update callers.
5689 (mark_pseudo_dead): Don't add conflicts for register
5690 ignore_reg_for_conflicts.
5691 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
5693 2018-10-05 Andrew Waterman <andrew@sifive.com>
5694 Jim Wilson <jimw@sifive.com>
5696 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
5697 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
5698 new pattern using HONOR_SNANS that emits one extra instruction.
5700 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5702 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
5703 patterns): Merge SI and DI patterns to a GPR pattern.
5704 (unnamed define_insn and define_split for record form of that): Merge
5705 to a single define_insn_and_split pattern.
5707 2018-10-05 David Malcolm <dmalcolm@redhat.com>
5710 * input.c (expand_location_to_spelling_point): Add param "aspect"
5711 and use rather than hardcoding LOCATION_ASPECT_CARET.
5712 (get_substring_ranges_for_loc): Handle the case of a single token
5713 within a macro expansion.
5714 * input.h (expand_location_to_spelling_point): Add "aspect" param,
5715 defaulting to LOCATION_ASPECT_CARET.
5717 2018-10-05 Paul Koning <ni1d@arrl.net>
5719 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
5720 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
5721 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
5722 (pdp11_guard_type): New function.
5724 2018-10-05 Paul Koning <ni1d@arrl.net>
5726 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
5727 * config/pdp11/pdp11.opt (mfloat32): Remove.
5729 * doc/invoke.texi (pdp11 -mfloat32): Remove:
5730 (pdp11 -mfloat64): Remove.
5732 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
5734 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
5735 (*cmp<mode>_cc_i387): Ditto.
5736 (*cmpu<mode>_cc_i387): Ditto.
5737 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5738 * config/i386/i386.c (ix86_expand_fp_compare): Remove
5740 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
5741 Emit x86_sahf_1 pattern.
5742 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
5743 (ix86_expand_carry_flag_compare): Ditto.
5745 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
5747 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
5748 to reg_or_0_operand. Add "C" constraint.
5749 (*cmpxf_cc_i387): Ditto.
5750 (*cmp<mode>_i387): Change operand 2 predicate
5751 to nonimm_or_0_operand. Add "C" constraint.
5752 (*cmp<mode>_cc_i387): Ditto.
5753 (*cmp<mode>_0_i387): Remove insn pattern.
5754 (*cmp<mode>_0_cc_i387): Ditto.
5756 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
5758 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
5759 * config/i386/predicates.md (nonimm_or_0_operand): Rename
5760 from vector_move_operand. Update all uses.
5762 2018-10-05 Martin Sebor <msebor@redhat.com>
5764 PR tree-optimization/87490
5765 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
5768 2018-10-05 Richard Biener <rguenther@suse.de>
5770 PR tree-optimization/63155
5771 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
5772 vertical space in dumpfiles.
5773 * tree-ssa-propagate.h
5774 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
5775 * tree-ssa-propagate.c (cfg_blocks_back): New global.
5776 (ssa_edge_worklist_back): Likewise.
5777 (curr_order): Likewise.
5778 (cfg_blocks_get): Remove abstraction.
5779 (cfg_blocks_add): Likewise.
5780 (cfg_blocks_empty_p): Likewise.
5781 (add_ssa_edge): Add to current or next worklist based on
5783 (add_control_edge): Likewise.
5784 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
5786 (ssa_propagation_engine::ssa_propagate): ... here. Unify
5787 iteration from CFG and SSA edge worklist so we process
5788 everything in RPO order, prioritizing forward progress
5790 (ssa_prop_init): Allocate new worklists, do not dump
5792 (ssa_prop_fini): Free new worklists.
5794 2018-10-05 Richard Biener <rguenther@suse.de>
5796 * tree-core.h (tree_block::abstract_flag): Remove.
5797 (tree_block::block_num): Make full 32bits.
5798 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
5799 * tree.h (BLOCK_ABSTRACT): Remove.
5800 * dwarf2out.c (gen_lexical_block_die): Remove dead code
5801 resulting from BLOCK_ABSTRACT being always false.
5802 (gen_inlined_subroutine_die): Likewise.
5803 (gen_block_die): Likewise.
5804 * tree.c (block_ultimate_origin): Likewise.
5805 * tree-pretty-print.c (dump_block_node): Remove code dealing
5806 with BLOCK_ABSTRACT.
5807 * tree-ssa-live.c (dump_scope_block): Likewise.
5808 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
5809 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
5811 2018-10-05 Richard Biener <rguenther@suse.de>
5813 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
5814 is asked for initialize mode to the component mode of the
5817 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
5820 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
5821 assembler for -mavx.
5822 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
5824 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5827 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
5829 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
5830 to be DImode. When using mffscrn, force the operand to a register.
5832 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
5834 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
5835 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
5836 X87MODEF mode iterator.
5837 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
5838 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
5839 X87MODEF mode iterator.
5841 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
5843 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
5844 -Wno-prio-ctor-dtor.
5846 2018-10-04 David Malcolm <dmalcolm@redhat.com>
5848 * Makefile.in (OBJS): Add opt-problem.o.
5849 * dump-context.h: Include "selftest.h.
5850 (selftest::temp_dump_context): New forward decl.
5851 (class dump_context): Make friend of class
5852 selftest::temp_dump_context.
5853 (dump_context::dump_loc_immediate): New decl.
5854 (class dump_pretty_printer): Move here from dumpfile.c.
5855 (class temp_dump_context): Move to namespace selftest.
5856 (temp_dump_context::temp_dump_context): Add param
5857 "forcibly_enable_dumping".
5858 (selftest::verify_dumped_text):
5859 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
5860 (selftest::verify_item):
5861 (ASSERT_IS_TEXT): Move here from dumpfile.c.
5862 (ASSERT_IS_TREE): Likewise.
5863 (ASSERT_IS_GIMPLE): Likewise.
5864 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
5866 (dump_context::dump_loc_immediate): ...this new function.
5867 (class dump_pretty_printer): Move to dump-context.h.
5868 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
5869 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
5870 (temp_dump_context::temp_dump_context): Move to "selftest"
5871 namespace. Add param "forcibly_enable_dumping", and use it to
5872 conditionalize the use of m_pp;
5873 (selftest::verify_dumped_text): Make non-static.
5874 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
5875 (selftest::verify_item): Make non-static.
5876 (ASSERT_IS_TEXT): Move to dump-context.h.
5877 (ASSERT_IS_TREE): Likewise.
5878 (ASSERT_IS_GIMPLE): Likewise.
5879 (selftest::test_capture_of_dump_calls): Pass "true" for new
5880 param of temp_dump_context.
5881 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
5882 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
5884 * opt-problem.cc: New file.
5885 * opt-problem.h: New file.
5886 * optinfo-emit-json.cc
5887 (selftest::test_building_json_from_dump_calls): Pass "true" for
5888 new param of temp_dump_context.
5889 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
5890 (optinfo::emit_for_opt_problem): New function.
5891 (optinfo::emit): Clarity which emit_item is used.
5892 * optinfo.h (optinfo::get_dump_location): New accessor.
5893 (optinfo::emit_for_opt_problem): New decl.
5894 (optinfo::emit): Make const.
5895 * selftest-run-tests.c (selftest::run_tests): Call
5896 selftest::opt_problem_cc_tests.
5897 * selftest.h (selftest::opt_problem_cc_tests): New decl.
5898 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
5899 bool to opt_result, converting fprintf messages to
5900 opt_result::failure_at calls. Add "stmt" param for use by the
5902 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
5903 (runtime_alias_check_p): Convert return type from bool to
5904 opt_result, converting dump_printf calls to
5905 opt_result::failure_at, using the statement DDR_A for their
5907 (find_data_references_in_stmt): Convert return type from bool to
5908 opt_result, converting "return false" to opt_result::failure_at
5910 * tree-data-ref.h: Include "opt-problem.h".
5911 (dr_analyze_innermost): Convert return type from bool to opt_result,
5912 and add a const gimple * param.
5913 (find_data_references_in_stmt): Convert return type from bool to
5915 (runtime_alias_check_p): Likewise.
5916 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
5917 dr_analyze_innermost.
5918 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
5919 Convert return type from bool to opt_result, adding a message for
5920 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
5921 (vect_analyze_data_ref_dependence): Convert return type from bool
5922 to opt_result. Change sense of return type from "false"
5923 effectively meaning "no problems" to "false" meaning a problem,
5924 so that "return false" becomes "return opt_result::success".
5925 Convert "return true" calls to opt_result::failure_at, using
5926 the location of statement A rather than vect_location.
5927 (vect_analyze_data_ref_dependences): Convert return type from bool
5929 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
5930 calls to opt_result::failure_at, using the stmt location rather
5932 (vect_verify_datarefs_alignment): Convert return type from bool
5934 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
5935 into multiple more-tightly-scoped copies.
5936 (vect_analyze_data_refs_alignment): Convert return type from bool
5938 (vect_analyze_data_ref_accesses): Likewise, converting a
5939 "return false" to a "return opt_result::failure_at", adding a
5941 (vect_prune_runtime_alias_test_list): Convert return type from
5942 bool to opt_result, converting dump_printf_loc to
5943 opt_result::failure_at. Add a %G to show the pertinent statement,
5944 and use the stmt's location rather than vect_location.
5945 (vect_find_stmt_data_reference): Convert return type from
5946 bool to opt_result, converting dump_printf_loc to
5947 opt_result::failure_at, using stmt's location.
5948 (vect_analyze_data_refs): Convert return type from bool to
5949 opt_result. Convert "return false" to "return
5950 opt_result::failure_at", adding messages as needed.
5951 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
5952 type from bool to opt_result.
5953 (vect_determine_vf_for_stmt): Likewise.
5954 (vect_determine_vectorization_factor): Likewise, converting
5955 dump_printf_loc to opt_result::failure_at, using location of phi
5956 rather than vect_location.
5957 (vect_analyze_loop_form_1): Convert return type from bool to
5958 opt_result, converting dump_printf_loc calls, retaining the use of
5960 (vect_analyze_loop_form): Convert return type from loop_vec_info
5961 to opt_loop_vec_info.
5962 (vect_analyze_loop_operations): Convert return type from bool to
5963 opt_result, converting dump_printf_loc calls, using the location
5964 of phi/stmt rather than vect_location where available. Convert
5965 various "return false" to "return opt_result::failure_at" with
5966 "unsupported phi" messages.
5967 (vect_get_datarefs_in_loop): Convert return type from bool to
5968 opt_result. Add a message for the
5969 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
5970 (vect_analyze_loop_2): Convert return type from bool to
5971 opt_result. Ensure "ok" is set to a opt_result::failure_at before
5972 each "goto again;", adding new messages where needed.
5973 Add "unsupported grouped {store|load}" messages.
5974 (vect_analyze_loop): Convert return type from loop_vec_info to
5976 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
5978 * tree-vect-stmts.c (process_use): Likewise, converting
5979 dump_printf_loc call and using stmt location, rather than
5981 (vect_mark_stmts_to_be_vectorized): Likeise.
5982 (vect_analyze_stmt): Likewise, adding a %G.
5983 (vect_get_vector_types_for_stmt): Convert return type from bool to
5984 opt_result, converting dump_printf_loc calls and using stmt
5985 location, rather than vect_location.
5986 (vect_get_mask_type_for_stmt): Convert return type from tree to
5987 opt_tree, converting dump_printf_loc calls and using stmt location.
5988 * tree-vectorizer.c: Include "opt-problem.h.
5989 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
5990 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
5991 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
5992 enabled, use it to report at the top level "couldn't vectorize
5993 loop" followed by the problem.
5994 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
5995 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
5997 (vect_analyze_stmt): Likewise.
5998 (vect_get_vector_types_for_stmt): Likewise.
5999 (tree vect_get_mask_type_for_stmt): Likewise.
6000 (vect_analyze_data_ref_dependences): Likewise.
6001 (vect_enhance_data_refs_alignment): Likewise.
6002 (vect_analyze_data_refs_alignment): Likewise.
6003 (vect_verify_datarefs_alignment): Likewise.
6004 (vect_analyze_data_ref_accesses): Likewise.
6005 (vect_prune_runtime_alias_test_list): Likewise.
6006 (vect_find_stmt_data_reference): Likewise.
6007 (vect_analyze_data_refs): Likewise.
6008 (vect_analyze_loop): Convert return type from loop_vec_info to
6010 (vect_analyze_loop_form): Likewise.
6011 (vect_analyze_slp): Convert return type from bool to opt_result.
6013 2018-10-04 David Malcolm <dmalcolm@redhat.com>
6015 * doc/invoke.texi (-fopt-info): Document new "internals"
6017 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
6018 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
6020 (optinfo_verbosity_options): Add "internals".
6021 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
6022 (dump_context::apply_dump_filter_p): New member function.
6023 (dump_context::dump_loc): Use apply_dump_filter_p rather than
6024 explicitly masking the dump_kind.
6025 (dump_context::begin_scope): Increment the scope depth first. Use
6026 apply_dump_filter_p rather than explicitly masking the dump_kind.
6027 (dump_context::emit_item): Use apply_dump_filter_p rather than
6028 explicitly masking the dump_kind.
6029 (dump_dec): Likewise.
6030 (dump_hex): Likewise.
6031 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
6032 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
6033 (opt_info_switch_p): Update handling of default
6034 MSG_OPTIMIZED_LOCATIONS to cope with default of
6035 MSG_PRIORITY_USER_FACING.
6036 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
6037 masking the dump_kind.
6038 (selftest::test_capture_of_dump_calls): Update test_dump_context
6039 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
6040 than MSG_ALL. Generalize scope test to be run at all four
6041 combinations of with/without MSG_PRIORITY_USER_FACING and
6042 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
6043 for each of the two values.
6044 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
6045 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
6046 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
6047 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
6048 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
6049 with MSG_PRIORITY_*.
6050 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
6051 dump messages as MSG_PRIORITY_USER_FACING.
6052 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
6053 about the interaction with MSG_PRIORITY_*.
6055 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
6057 * varasm.c (output_constant): Add new parameter merge_strings.
6058 Make strings properly zero terminated in merge string sections.
6059 (mergeable_string_section): Don't fail if the last char is non-zero.
6060 (assemble_variable_contents): Handle merge string sections.
6061 (assemble_variable): Likewise.
6062 (assemble_constant_contents): Likewise.
6063 (output_constant_def_contents): Likewise.
6064 (output_constructor_array_range,
6065 output_constructor_regular_field): Adjust call to output_constant.
6066 (output_object_block): Adjust call to assemble_constant_contents
6067 and assemble_variable_contents.
6069 2018-10-04 Martin Liska <mliska@suse.cz>
6072 * cgraphunit.c (process_function_and_variable_attributes):
6073 Warn about a function with alias attribute and a body.
6075 2018-10-04 Martin Liska <mliska@suse.cz>
6078 * multiple_target.c (redirect_to_specific_clone): New function.
6079 (ipa_target_clone): Use it.
6080 * tree-inline.c: Fix comment.
6082 2018-10-04 David Malcolm <dmalcolm@redhat.com>
6084 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
6086 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
6087 (gcc::dump_manager::register_pass): New member function, adapted
6088 from loop body in gcc::pass_manager::register_pass, adding a
6089 call to update_dfi_for_opt_info.
6090 (gcc::dump_manager::opt_info_enable_passes): Store the
6091 -fopt-info options into the new fields. Move the loop
6093 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
6095 * dumpfile.h (struct opt_pass): New forward decl.
6096 (gcc::dump_manager::register_pass): New decl.
6097 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
6098 (class gcc::dump_manager): Add fields "m_optgroup_flags",
6099 "m_optinfo_flags", and "m_optinfo_filename".
6100 * passes.c (gcc::pass_manager::register_pass): Move all of the
6101 dump-handling code to gcc::dump_manager::register_pass.
6103 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
6105 PR rtl-optimization/87466
6106 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
6107 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
6108 * doc/tm.texi: Regenerate.
6109 * ira-lives.c (process_bb_node_lives): Use the new target hook.
6110 * lra-lives.c (process_bb_lives): Likewise.
6111 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
6114 2018-10-04 Tamar Christina <tamar.christina@arm.com>
6116 * params.c (add_params): Fix initialization.
6118 2018-10-04 Martin Liska <mliska@suse.cz>
6120 PR gcov-profile/84107
6121 * tree-profile.c (init_ic_make_global_vars):
6122 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
6123 Come up with new ic_tuple* variables. Emit
6124 __gcov_indirect_call{,_topn} variables.
6125 (gimple_gen_ic_profiler): Access the variable
6127 (gimple_gen_ic_func_profiler): Access
6128 __gcov_indirect_call.callee field.
6129 (gimple_init_gcov_profiler): Use ptr_type_node.
6130 * value-prof.c (gimple_ic): Use ptr_type_node.
6132 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6134 PR tree-optimization/85787
6135 * ipa-pure-const.c (malloc_candidate_p_1): Move most of
6136 malloc_candidate_p into this function and add support for
6137 detecting multiple phis.
6138 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
6140 2018-10-04 Martin Liska <mliska@suse.cz>
6143 * ipa-inline.c (inline_to_all_callers_1):
6144 Call ultimate_alias_target for node being inlined.
6146 2018-10-03 Jeff Law <law@redhat.com>
6148 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
6150 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
6151 * tree.h (get_typenode_from_name): Prototype.
6153 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
6155 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
6156 Change operand 2 predicate to nonimmediate_operand.
6157 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6159 2018-10-03 Martin Sebor <msebor@redhat.com>
6160 Jeff Law <law@redhat.com>
6162 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
6164 (get_string_length): Detect unterminated arrays.
6165 (format_string): Same.
6166 (format_directive): Warn about unterminated arrays.
6167 (handle_gimple_call): Mark statements with no_warning as needed.
6169 2018-10-03 Jim Wilson <jimw@sifive.com>
6171 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
6172 also define __riscv_abi_rve. Delete trailing white space.
6174 2018-10-03 Paul Koning <ni1d@arrl.net>
6176 Enable LRA register allocator for PDP11.
6177 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
6180 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
6181 * config/pdp11/pdp11.opt (-mlra): New option.
6182 * doc/invoke.texi (PDP-11 Options): Document -mlra.
6184 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
6186 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
6187 (*<absneg:code>extend<mode>xf2): Ditto.
6189 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
6191 PR tree-optimization/87415
6192 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
6195 2018-10-02 Jeff Law <law@redhat.com>
6197 * gimple-fold.c (get_range_strlen): Only set *nonstr when
6198 an unterminated string is discovered. Bubble up range
6199 even for unterminated strings.
6200 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
6201 indicates the string was not terminated via NONSTR.
6203 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
6205 * tree-vrp.c (extract_range_from_unary_expr): Special case all
6206 pointer conversions.
6207 Do not do anything special for anti-ranges.
6209 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
6211 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
6212 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
6214 2018-10-03 Martin Liska <mliska@suse.cz>
6216 PR gcov-profile/86109
6217 * coverage.c (coverage_begin_function): Do not
6218 mark lambdas as artificial.
6219 * tree-core.h (struct GTY): Remove tm_clone_flag
6220 and introduce new lambda_function.
6221 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
6223 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
6226 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
6227 P8_VECTOR and VSX are enabled.
6229 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
6231 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
6232 0x3907 as CPU model number.
6234 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
6236 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
6237 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
6238 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
6239 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
6240 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
6241 * config/s390/s390.md: Likewise. Rename also the cpu attribute
6242 value from arch12 to z14.
6244 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
6246 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
6248 (isinf<mode>2): Ditto.
6250 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
6252 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
6253 before emitting fxam. Perform calculations in XFmode.
6255 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
6257 * match.pd (((X /[ex] A) +- B) * A): New transformation.
6259 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
6262 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
6263 * tree.c (signed_or_unsigned_type_for): Handle complex.
6265 2018-10-02 Jeff Law <law@redhat.com>
6267 * gimple-fold.c (get_range_strlen): Remove dead code.
6269 2018-10-02 Martin Sebor <msebor@redhat.com>
6270 Jeff Law <law@redhat.com>
6272 * builtins.c (unterminated_array): Add new arguments.
6273 If argument is not terminated, bubble up size and exact
6275 (expand_builtin_strnlen): Detect, avoid expanding
6276 and diagnose unterminated arrays.
6277 (c_strlen): Fill in offset of start of unterminated strings.
6278 * builtins.h (unterminated_array): Update prototype.
6280 2018-10-02 Richard Biener <rguenther@suse.de>
6282 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
6283 of haddv4df, first reduce to SSE width and exploit the fact
6284 that we only need element zero with the reduction result.
6285 (reduc_plus_scal_v2df): Likewise.
6287 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
6289 * dojump.h (do_jump): Delete.
6290 (do_jump_1): Likewise.
6291 (split_comparison): Move around.
6292 * dojump.c (do_jump): Make static.
6293 (do_jump_1): Likewise.
6294 (jumpifnot): Move around.
6295 (jumpifnot_1): Likewise.
6297 (jumpif_1): Likewise.
6298 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
6300 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
6302 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
6303 insns in the delay slot and add_insn_after for the jump insn.
6305 2018-10-02 Richard Biener <rguenther@suse.de>
6307 * tree-inline.c (expand_call_inline): Use the location of
6308 the callee declaration for the inline-entry marker.
6309 * final.c (notice_source_line): Remove special-casing of
6310 NOTE_INSN_INLINE_ENTRY.
6312 2018-10-01 Carl Love <cel@us.ibm.com>
6315 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
6316 (__builtin_mtfsb0): New.
6317 (__builtin_mtfsb1): New.
6318 ( __builtin_set_fpscr_rn): New.
6319 (__builtin_set_fpscr_drn): New.
6320 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
6321 (rs6000_expand_set_fpscr_rn_builtin): Add.
6322 (rs6000_expand_set_fpscr_drn_builtin): Add.
6323 (rs6000_expand_builtin): Add case statement entries for
6324 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
6325 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
6326 RS6000_BUILTIN_MFFSL.
6327 (rs6000_init_builtins): Add ftype initialization and def_builtin
6328 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
6329 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
6330 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
6331 rs6000_mffscdrn): Add define_insn.
6332 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
6333 * doc/extend.texi: Add documentation for the builtins.
6335 2018-10-01 Richard Biener <rguenther@suse.de>
6337 PR tree-optimization/87465
6338 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
6339 causing branch miscounts.
6341 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6343 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
6344 aarch64_option_default_param): New.
6345 (params.h): Include.
6346 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
6347 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
6348 stack-clash protection validation code.
6350 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6352 * params.c (validate_param): New.
6353 (add_params): Use it.
6354 (set_param_value): Refactor param validation into validate_param.
6355 (diagnostic.h): Include.
6356 * diagnostic.h (diagnostic_ready_p): New.
6358 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6360 * params.c (set_param_value):
6361 Add index of parameter being validated.
6362 * common/common-target.def (option_validate_param): New.
6363 * common/common-targhooks.h (default_option_validate_param): New.
6364 * common/common-targhooks.c (default_option_validate_param): New.
6365 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
6366 * doc/tm.texi: Regenerate.
6368 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6371 * config/aarch64/aarch64.c (aarch64_override_options_internal):
6372 Add validation for stack-clash parameters and set defaults.
6374 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6377 * configure.ac: Add stack-clash-protection-guard-size.
6378 * doc/install.texi: Document it.
6379 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
6380 * params.def: Update comment for guard-size.
6381 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
6382 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
6383 * configure: Regenerate.
6385 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6388 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
6389 STACK_DYNAMIC_OFFSET): New.
6390 * config/aarch64/aarch64.c (aarch64_layout_frame):
6391 Update outgoing args size.
6392 (aarch64_stack_clash_protection_alloca_probe_range,
6393 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
6395 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6398 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
6400 * target.def (stack_clash_protection_alloca_probe_range): New.
6401 (stack_clash_protection_final_dynamic_probe): Remove.
6402 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
6403 (default_stack_clash_protection_final_dynamic_probe): Remove.
6404 * targhooks.c: Likewise.
6405 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
6406 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
6407 * doc/tm.texi: Regenerate.
6409 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6412 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
6413 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
6414 aarch64_clamp_to_uimm12_shift): New.
6415 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
6416 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
6418 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6421 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
6423 2018-10-01 Jeff Law <law@redhat.com>
6424 Richard Sandiford <richard.sandiford@linaro.org>
6425 Tamar Christina <tamar.christina@arm.com>
6428 * config/aarch64/aarch64.md
6429 (probe_stack_range): Add k (SP) constraint.
6430 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
6431 STACK_CLASH_MAX_UNROLL_PAGES): New.
6432 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
6433 stack probes for stack clash.
6434 (aarch64_allocate_and_probe_stack_space): New.
6435 (aarch64_expand_prologue): Use it.
6436 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
6437 (aarch64_sub_sp): Add emit_move_imm optional param.
6439 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
6441 PR tree-optimization/87261
6442 * match.pd: Remove trailing whitespace.
6443 Add (x & y) | ~(x | y) -> ~(x ^ y),
6444 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
6446 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
6448 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
6450 (ashlsi3_insn): Update instruction constraints.
6451 (ashrsi3_insn): Likewise.
6452 (rotrsi3): Likewise.
6453 (add_shift): Likewise.
6454 * config/arc/constraints.md (Csz): New 32 bit constraint. It
6455 avoids placing in the limm field small constants which, otherwise,
6456 could end into a small instruction.
6458 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
6460 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
6461 destination register is not odd-even.
6462 (umaddsidi4_split): Likewise.
6464 2018-10-01 Richard Biener <rguenther@suse.de>
6466 * tree-inline.c (expand_call_inline): Store origin of fn
6467 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
6468 * tree.c (block_ultimate_origin): Simplify and do some
6471 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
6473 * config/i386/mmx.md (EMMS): New int iterator.
6474 (emms): New int attribute.
6475 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
6476 EMMS int iterator. Explicitly declare clobbers.
6477 (mmx_emms): Remove expander.
6479 * config/i386/predicates.md (emms_operation): Remove predicate.
6480 (vzeroall_pattern): New predicate.
6481 (vzeroupper_pattern): Rename from vzeroupper_operation.
6482 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
6483 vzeroupper_pattern and vzeroall_pattern predicates.
6485 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
6487 PR rtl-optimization/86939
6488 * ira-lives.c (make_hard_regno_born): Rename from this...
6489 (make_hard_regno_live): ... to this. Remove update to conflict
6490 information. Update function comment.
6491 (make_hard_regno_dead): Add conflict information update. Update
6493 (make_object_born): Rename from this...
6494 (make_object_live): ... to this. Remove update to conflict information.
6495 Update function comment.
6496 (make_object_dead): Add conflict information update. Update function
6498 (mark_pseudo_regno_live): Call make_object_live.
6499 (mark_pseudo_regno_subword_live): Likewise.
6500 (mark_hard_reg_dead): Update function comment.
6501 (mark_hard_reg_live): Call make_hard_regno_live.
6502 (process_bb_node_lives): Likewise.
6503 * lra-lives.c (make_hard_regno_born): Rename from this...
6504 (make_hard_regno_live): ... to this. Remove update to conflict
6505 information. Remove now uneeded check_pic_pseudo_p argument.
6506 Update function comment.
6507 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
6508 to conflict information. Update function comment.
6509 (mark_pseudo_live): Remove update to conflict information. Update
6511 (mark_pseudo_dead): Add conflict information update.
6512 (mark_regno_live): Call make_hard_regno_live.
6513 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
6514 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
6516 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
6519 * config/i386/i386.c (construct_container): Use TImode for
6520 BLKmode values in 2 integer registers.
6522 2018-09-29 Jeff Law <law@redhat.com>
6524 * builtins.c (unterminated_array): Pass in c_strlen_data * to
6525 c_strlen rather than just a tree *.
6526 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
6527 Update recursive calls appropriately. If caller did not provide a
6528 suitable data pointer, create a local one. When a non-terminated
6529 string is discovered, bubble up information about the string via the
6530 c_strlen_data object.
6531 * builtins.h (c_strlen): Update prototype.
6532 (c_strlen_data): New structure.
6533 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
6534 For a type 2 call, if c_strlen indicates a non-terminated string
6535 use the length of the non-terminated string.
6536 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
6538 2018-09-29 Jakub Jelinek <jakub@redhat.com>
6541 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
6542 __m512d type for __A argument rather than __m512.
6544 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
6546 * match.pd (simple_comparison): Don't optimize if either operand is
6547 a function pointer when target needs function pointer canonicalization.
6549 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
6551 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
6552 power5 .. power9 to remove indirection.
6553 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
6554 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
6555 ASM_CPU_476_SPEC): Delete.
6556 (ASM_CPU_SPEC): Adjust.
6557 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
6558 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
6560 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
6562 * config.in: Delete HAVE_AS_DCI.
6563 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
6564 * config/rs6000/rs6000.h: Ditto.
6565 * configure.ac: Delete HAVE_AS_DCI.
6566 * configure: Regenerate.
6568 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
6570 * config.in (HAVE_AS_LWSYNC): Delete.
6571 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
6572 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
6574 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
6575 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
6577 * configure.ac: Delete HAVE_AS_LWSYNC.
6578 * configure: Regenerate.
6580 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
6581 Pierre-Marie de Rodat <derodat@adacore.com>
6583 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
6584 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
6585 (cgraph_node::create_thunk): Add indirect_offset parameter.
6586 (thunk_adjust): Likewise.
6587 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
6588 and initialize the corresponding field with it.
6589 (cgraph_node::dump): Dump indirect_offset field.
6590 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
6591 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
6592 (thunk_adjust): Add indirect_offset parameter and deal with it.
6593 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
6594 pass it to thunk_adjust. Do not call the target hook if it's non-zero
6595 or if the thunk is external or local. Fix formatting. Do not chain
6596 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
6597 if any, in the GIMPLE representation.
6598 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
6599 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
6600 (input_node): Read indirect_offset field.
6601 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
6602 call to thunk_adjust.
6603 * tree-nested.c (struct nesting_info): Add thunk_p field.
6604 (create_nesting_tree): Set it.
6605 (convert_all_function_calls): Copy static chain from targets to thunks.
6606 (finalize_nesting_tree_1): Return early for thunks.
6607 (unnest_nesting_tree_1): Do not finalize thunks.
6608 (gimplify_all_functions): Do not gimplify thunks.
6610 2018-09-28 David Malcolm <dmalcolm@redhat.com>
6612 * opt-suggestions.c (option_proposer::build_option_suggestions):
6613 Release "option_values".
6615 2018-09-28 David Malcolm <dmalcolm@redhat.com>
6617 * coverage.c (get_coverage_counts): Convert problem-reporting dump
6618 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
6619 * dumpfile.c (kind_as_string): New function.
6620 (dump_loc): Rather than a hardcoded prefix of "note: ", use
6621 kind_as_string to vary the prefix based on dump_kind.
6622 (selftest::test_capture_of_dump_calls): Update for above.
6624 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
6626 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
6627 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
6629 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
6631 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
6632 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
6633 INVALID_REGNUM instead of FPSR_REG.
6634 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
6635 * config/i386/i386.md: Update comment of FP compares.
6636 (fldenv): Do not clobber FPSR_REG.
6638 2018-09-28 Richard Biener <rguenther@suse.de>
6640 * tree.h (BLOCK_ORIGIN): New.
6641 * omp-expand.c (grid_expand_target_grid_body): Assign
6642 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
6643 * tree-inline.c (remap_block): Likewise.
6644 * auto-profile.c (get_function_decl_from_block): Simplify
6645 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
6646 * langhooks.c (lhd_print_error_function): Likewise.
6647 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
6649 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
6650 * tree.c (block_nonartificial_location): Likewise.
6651 (block_ultimate_origin): Likewise.
6652 * tree-pretty-print.c (percent_K_format): Likewise. Remove
6653 no longer needed LTO case.
6655 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
6656 Jan Hubicka <jh@suse.cz>
6657 Martin Jambor <mjambor@suse.cz>
6659 * simplify-rtx.c (simplify_merge_mask): New function.
6660 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
6661 same masks are used in op1 or op2.
6662 (test_vec_merge): New function.
6663 (test_vector_ops): Call test_vec_merge.
6665 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
6667 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
6668 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
6669 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
6670 (cypress_costs): Set it.
6671 (supersparc_costs): Likewise.
6672 (hypersparc_costs): Likewise.
6673 (leon_cost): Likewise.
6674 (leon3_costs): Likewise.
6675 (sparclet_costs): Likewise.
6676 (ultrasparc_costs): Likewise.
6677 (ultrasparc_costs): Likewise.
6678 (niagara_costs): Likewise.
6679 (niagara2_costs): Likewise.
6680 (niagara3_costs): Likewise.
6681 (niagara4_costs): Likewise.
6682 (niagara7_costs): Likewise.
6683 (m8_costs): Likewise.
6684 (TARGET_CAN_FOLLOW_JUMP): Define.
6685 (pass_work_around_errata::gate): Minor tweak.
6686 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
6687 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
6688 Automaitcally clear MASK_FSMULD mask for V7 processors.
6689 (sparc_can_follow_jump): New static function.
6690 (output_ubranch): Deal with CROSSING_JUMP_P.
6691 (sparc_use_sched_lookahead): Rewrite using switch statement.
6692 (sparc_issue_rate): Reorder.
6693 (sparc_branch_cost): New function.
6695 2018-09-27 Martin Sebor <msebor@redhat.com>
6697 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
6698 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
6699 (int_fits_type_p): Same.
6701 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
6703 * config/i386/i386.md (FPCR_REG): Remove.
6704 (UNSPEC_FLDCW): Remove.
6705 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
6706 (x86_fldcw_1): Remove insn pattern.
6707 (fnstenv): Do not clobber FPCR_REG.
6709 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
6710 (CALL_USED_REGISTERS): Ditto.
6711 (REG_ALLOC_ORDER): Ditto.
6712 (REG_CLASS_CONTENTS): Ditto.
6713 (HI_REGISTER_NAMES): Ditto.
6714 (ADDITIONAL_REGISTER_NAMES): Use defines instead
6715 of numerical constants.
6716 * config/i386/i386.c (regclass_map): Remove fpsr register.
6717 (dbx_register_map): Ditto.
6718 (dbx64_register_map): Ditto.
6719 (svr4_dbx_register_map): Ditto.
6720 (print_reg): Do not handle FPCR_REG.
6722 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
6725 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
6726 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
6727 Delete, always treat as true.
6728 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
6729 Ditto. Simplify remaining code.
6730 * config/powerpcspe/powerpcspe.h: Ditto.
6731 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
6732 Simplify remaining code.
6733 (rs6000_expand_builtin): Ditto.
6734 * config/rs6000/rs6000.h: Ditto.
6735 * configure.ac: Ditto.
6736 * configure: Regenerate.
6738 2018-09-27 Martin Liska <mliska@suse.cz>
6740 * coverage.c (get_coverage_counts): Revert the formatting
6741 of missing profile opt info.
6743 2018-09-27 Richard Biener <rguenther@suse.de>
6747 * dwarf2out.c (set_block_origin_self): Do not mark outermost
6748 block as we do not output that.
6749 (gen_inlined_subroutine_die): Elide the originally outermost
6750 block, matching what we do for concrete instances.
6751 (decls_for_scope): Add parameter specifying whether to recurse
6754 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
6755 Tom de Vries <tom@codesourcery.com>
6759 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
6760 STORE_FLAG_VALUE == 1.
6762 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
6764 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
6765 constant definitions.
6766 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
6767 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
6768 ("speculation_barrier"): New expander definition.
6770 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
6772 PR gcov-profile/86957
6773 * common.opt: New warning option -Wmissing-profile.
6774 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
6775 * doc/invoke.texi: Document -Wmissing-profile.
6777 2018-09-26 Jim Wilson <jimw@sifive.com>
6779 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
6780 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
6782 2018-09-26 Martin Sebor <msebor@redhat.com>
6784 * tree.c (zerop): Change return type to bool.
6785 (integer_zerop, integer_onep, integer_each_onep): Same.
6786 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
6787 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
6788 (real_onep, real_minus_onep, chain_index): Same.
6789 (print_type_hash_statistics, type_list_equal): Same.
6790 * tree.h (zerop): Same.
6791 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
6792 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
6793 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
6794 (real_onep, real_minus_onep, chain_index): Same.
6795 (print_type_hash_statistics, type_list_equal): Same.
6797 2018-09-26 Jim Wilson <jimw@sifive.com>
6799 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
6801 2018-09-26 Jakub Jelinek <jakub@redhat.com>
6804 * config/i386/i386.c: Include debug.h and dwarf2out.h.
6805 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
6808 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
6810 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
6812 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
6814 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
6815 and FP_SECOND_SSE_REGS.
6816 (REG_CLASS_NAMES): Ditto.
6817 (REG_CLASS_CONTENTS): Ditto.
6818 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
6819 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
6820 (ix86_preferred_output_reload_class): Ditto.
6821 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
6822 clobber constraint to "=&f".
6823 (fix_truncdi_i387): Ditto.
6824 (lrintxfdi2): Ditto.
6825 (fistdi2_<rounding>): Ditto.
6826 (fpremxf4_i387): Change "=u" constraint to "=f".
6827 (fprem1xf4_i387): Ditto.
6829 (fptanxf4_i387): Ditto.
6830 (fxtractxf3_i387): Ditto.
6831 (fscalexf4_i387): Ditto.
6832 (atan2xf3): Change "u" constraint to "f".
6833 (fyl2xxf3_i387): Ditto.
6834 (fyl2xp1xf3_i387): Ditto.
6836 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
6839 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
6840 for removed I387_MASK_PM entity.
6843 2018-09-26 Jeff Law <law@redhat.com>
6845 2018-09-26 Alexey Neyman <stilor@att.net>
6847 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
6848 headers are no longer pulled in by <isl/val.h>.
6850 2018-09-26 Richard Biener <rguenther@suse.de>
6853 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
6854 or concrete instance DIE to the tree. Create abstract origin
6855 attributes also for concrete instances.
6857 2018-09-26 Alexey Neyman <stilor@att.net>
6859 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
6860 headers are no longer pulled in by <isl/val.h>.
6862 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
6864 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
6865 Use new helper functions.
6866 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
6867 Use new helper functions.
6868 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
6869 aarch_mm_needs_release): New declarations.
6870 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
6871 aarch_mm_needs_release): New.
6873 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
6875 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
6876 (arm32_output_mi_thunk): Deal with long calls.
6878 2018-09-26 Richard Biener <rguenther@suse.de>
6882 * tree-inline.c (expand_call_inline): When the location
6883 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
6884 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
6885 the inserted BLOCK to make inlined_function_outer_scope_p
6887 * dwarf2out.c (add_call_src_coords_attributes): Do not add
6888 coords for reserved locations.
6890 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
6892 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
6893 (*call_indirect_nonlocal_sysv<mode>): Ditto.
6894 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
6895 (*sibcall_nonlocal_sysv<mode>): Ditto.
6896 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
6897 (<bd>_<mode>): Ditto.
6898 (<bd>tf_<mode>): Ditto.
6900 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
6902 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
6903 control string as a list of templates instead of as C code.
6904 (*altivec_movti): Ditto.
6905 * config/rs6000/darwin.md (movdf_low_di): Ditto.
6907 2018-09-25 Jim Wilson <jimw@sifive.com>
6909 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
6910 when target symbol is weak.
6912 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
6915 * builtins.c (unterminated_array): Simplify.
6916 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
6917 where pointer arithmetic is safe.
6919 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
6922 * config/rs6000/altivec.md (altivec_vspltb): Use
6923 const_0_to_15_operand instead of u5bit_cint_operand.
6924 (*altivec_vspltb_internal): Ditto.
6925 (altivec_vspltb_direct): Ditto.
6926 (altivec_vsplth): Use const_0_to_7_operand instead of
6928 (*altivec_vsplth_internal): Ditto.
6929 (altivec_vsplth_direct): Ditto.
6930 (altivec_vspltw): Use const_0_to_3_operand instead of
6932 (*altivec_vspltw_internal): Ditto.
6933 (altivec_vspltw_direct): Ditto.
6934 (altivec_vspltsf): Ditto.
6935 (*altivec_vspltsf_internal): Ditto.
6936 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
6937 various splats with the proper size immediate. Reorder the various
6938 cases by ascending size of immediate, and put all such together.
6940 2018-09-25 Richard Biener <rguenther@suse.de>
6943 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
6945 (maybe_create_die_with_external_ref): Do not create
6946 DW_TAG_imported_unit here.
6947 (add_abstract_origin_attribute): Handle external BLOCK refs.
6948 (dwarf2out_abstract_function): Simplify LTO case.
6949 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
6950 rather than using maybe_create_die_with_external_ref.
6952 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
6955 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
6956 (frndintxf2_mask_pm_i387): Ditto.
6957 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
6958 Enable for !flag_trapping_math.
6959 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
6960 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
6961 Change operand 1 predicate to nonimmediate_operand.
6962 (attr "i387_cw"): Remove mask_pm.
6963 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
6964 (enum ix86_entity): Remove I387_MASK_PM.
6965 * config/i386/i386.c (ix86_i387_mode_needed): Do not
6966 handle I387_MASK_PM.
6967 (ix86_mode_needed): Ditto.
6968 (ix86_mode_after): Ditto.
6969 (ix86_mode_entry): Ditto.
6970 (ix86_mode_exit): Ditto.
6971 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
6973 2018-09-25 Jakub Jelinek <jakub@redhat.com>
6975 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
6976 to_update_switch_stmts to vNULL instead of calling create on them
6979 2018-09-25 Richard Biener <rguenther@suse.de>
6981 PR tree-optimization/87402
6982 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
6983 (visit_phi): Re-instantiate handling of supposed to be VARYING
6984 but non-VARYING backedge value.
6986 2018-09-25 Richard Biener <rguenther@suse.de>
6989 * dwarf2out.c (struct sym_off_pair): New.
6990 (external_die_map): New global.
6991 (lookup_decl_die): When in LTO create DIEs lazily from the
6993 (lookup_block_die): New function, create DIEs lazily in LTO.
6994 (equate_block_to_die): New function.
6995 (dwarf2out_die_ref_for_decl): During WPA get the association
6996 from the external DIE map.
6997 (dwarf2out_register_external_die): Record mapping into the
6999 (maybe_create_die_with_external_ref): New function split out from
7000 DIE generation part of old dwarf2out_register_external_die.
7001 (add_abstract_origin_attribute): Do not return the DIE. When
7002 in LTO reference externals directly.
7003 (dwarf2out_abstract_function): When in LTO ignore calls for
7004 decls with external DIEs (already present abstract instances).
7005 (gen_call_site_die): Adjust.
7006 (add_high_low_attributes): Likewise.
7007 (gen_lexical_block_die): Likewise.
7008 (gen_inlined_subroutine_die): Likewie.
7009 (gen_block_die): Likewise.
7010 (dwarf2out_inline_entry): Likewise.
7011 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
7014 2018-09-25 Martin Liska <mliska@suse.cz>
7016 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
7017 integers and not by a float value.
7019 2018-09-25 Martin Liska <mliska@suse.cz>
7022 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
7024 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
7025 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
7026 * godump.c (go_format_type): Remove extra parenthesis.
7028 2018-09-25 Martin Liska <mliska@suse.cz>
7030 * alias.c (set_dest_equal_p): Remove unused function.
7031 * config/i386/i386.c (def_builtin_pure2): Likewise.
7032 * diagnostic-show-locus.c (class layout): Remove
7034 (layout::layout): Likewise here.
7035 * dump-context.h (class temp_dump_context): Likewise.
7036 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
7037 (add_AT_loclistsptr): Likewise.
7038 (add_AT_offset): Likewise.
7039 (get_AT_hi_pc): Likewise.
7040 (is_comdat_die): Likewise.
7041 (type_is_enum): Likewise.
7042 (ceiling): Likewise.
7043 (add_AT_vms_delta): Likewise.
7044 (is_class_die): Likewise.
7045 * edit-context.c (class line_event): Remove unused field.
7046 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
7048 * ipa-cp.c (ipa_get_vr_lat): Likewise.
7049 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
7050 (ok_for_base_p_nonstrict): Likewise.
7051 * tree-chrec.c (is_not_constant_evolution): Likewise.
7052 (chrec_fold_poly_cst): Likewise.
7053 * tree-if-conv.c (has_pred_critical_p): Likewise.
7054 * tree-ssa-coalesce.c (print_exprs): Likewise.
7055 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
7056 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
7057 * tree-vrp.c (value_ranges_intersect_p): Likewise.
7058 (value_range_nonnegative_p): Likewise.
7060 2018-09-25 Martin Liska <mliska@suse.cz>
7062 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
7063 Do not handle "GNU Pascal".
7064 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
7066 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
7067 from documentation. Likewise.
7068 * dbxout.c (dbxout_range_type): Likewise.
7069 * doc/cpp.texi: Likewise.
7070 * doc/extend.texi: Likewise.
7071 * doc/frontends.texi: Likewise.
7072 * doc/invoke.texi: Remove Pascal entry.
7073 * tree.def (CLEANUP_POINT_EXPR): Likewise.
7074 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
7076 2018-09-25 Martin Liska <mliska@suse.cz>
7079 * doc/invoke.texi: Document all parameters and remove default
7082 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
7085 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
7086 contains HOST_WIDE_INTs when computing its size.
7088 2018-09-24 Jim Wilson <jimw@sifive.com>
7091 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
7093 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
7095 2018-09-24 Andrew Pinski <apinski@marvell.com>
7097 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
7098 access prev before checking it for NULLness in the
7099 AARCH64_FUSE_CMP_BRANCH case.
7101 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
7104 * config/i386/i386.c (rest_of_insert_endbranch): Set
7105 endbr_queued_at_entrance to true and don't insert ENDBR if
7106 x86_function_profiler will be called.
7107 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
7109 * config/i386/i386.h (machine_function): Add
7110 endbr_queued_at_entrance.
7112 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
7114 * genattrtab.c (mk_attr_alt): Use alternative_mask.
7115 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
7117 (check_attr_test): Use alternative_mask.
7118 (get_attr_value): Likewise.
7119 (compute_alternative_mask): Use alternative_mask and XWINT.
7120 (make_alternative_compare): Use alternative_mask.
7121 (attr_alt_subset_p): Use XWINT.
7122 (attr_alt_subset_of_compl_p): Likewise.
7123 (attr_alt_intersection): Use alternative_mask and XWINT.
7124 (attr_alt_union): Likewise.
7125 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
7126 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
7127 (simplify_test_exp): Use alternative_mask and XWINT.
7128 (write_test_expr): Use alternative_mask and XWINT, adjust bit
7129 number calculation to support 64 bits. Generate code that
7130 checks 64-bit masks.
7131 (main): Use alternative_mask.
7132 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
7134 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
7137 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
7138 RETURN+USE when returning via %r14.
7140 2018-09-24 Martin Liska <mliska@suse.cz>
7142 * gcov.c (output_lines): Print colorization legend
7143 for both flag_use_colors and flag_use_hotness_colors.
7146 2018-09-24 Martin Liska <mliska@suse.cz>
7148 * coverage.c (get_coverage_counts): Use warning_at
7149 with current_function_decl location. Use %qD in warning
7152 2018-09-24 Martin Liska <mliska@suse.cz>
7154 * memory-block.h (memory_block_pool::release): Annotate with
7155 valgrind that the memory is not accessible.
7157 2018-09-24 Martin Liska <mliska@suse.cz>
7160 * asan.c: Make asan_handled_variables extern.
7162 * cfgexpand.c (expand_stack_vars): Make sure
7163 a representative is unpoison if another
7164 variable in the partition is handled by
7165 use-after-scope sanitization.
7167 2018-09-24 Richard Biener <rguenther@suse.de>
7169 PR tree-optimization/63155
7170 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
7171 the worklist when the edge of the respective argument isn't
7174 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
7176 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
7177 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
7178 (MASK_CLASS_P): Update for rename.
7179 (MAYBE_MASK_CLASS_P): Ditto.
7180 (REG_CLASS_NAMES): Update.
7181 (REG_CLASS_CONTENT): Update.
7182 * config/i386/i386.c (regclass_map): Update for MASK_REG
7183 and ALL_MASK_REGS rename.
7184 * config/i386/constraints.md (Yk): Update for rename.
7187 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
7189 * config/i386/i386.h (enum reg_class): Remove
7190 EVEX_SSE_REGS and MOD4_SSE_REGS.
7191 (REG_CLASS_NAMES): Update.
7192 (REG_CLASS_CONTENT): Update.
7193 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
7194 registers as ALL_SSE_REGS.
7195 (ix86_additional_allocno_class_p): Remove.
7196 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
7197 (ix86_register_priority): Lower priority of EVEX SSE registers.
7198 Use IN_RANGE macro where appropriate.
7199 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
7200 AVX-5124VNNIW checks.
7201 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
7202 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
7203 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
7204 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
7205 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
7206 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
7207 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
7208 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
7209 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
7210 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
7211 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
7212 * config/i386/constraints.md (Yh): Remove.
7214 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
7216 * config/i386/i386.c (regclass_map): Declare integer REX registers
7219 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
7221 * doc/service.texi (Service): Switch the fsf.org link to https.
7223 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
7226 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7227 Define to speculation_safe_value_not_needed.
7229 2018-09-21 Florian Weimer <fweimer@redhat.com>
7232 * doc/extend.texi (Common Function Attributes): Mention that
7233 noreturn suppresses tail call optimization.
7235 2018-09-21 Jeff Law <law@redhat.com>
7237 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
7238 vr_values::cleanup_edges_and_switches.
7239 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
7241 (identify_jump_threads): Remove EDGE_IGNORE handling.
7242 (execute_vrp): Move handling of to_remove_edges and
7243 to_update_switch_stmts into vr_values class member functions.
7244 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
7245 (to_update_switch_stmts): Likewise.
7246 * vr-values.c: Include cfghooks.h.
7247 (vr_values::vr_values): Initialize to_remove_edges and
7248 to_update_switch_stmts.
7249 (vr_values::~vr_values): Verify to_remove_edges and
7250 to_update_switch_stmts are empty.
7251 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
7252 (vr_values::cleanup_edges_and_switches): New member function.
7253 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
7254 function. Add new data members.
7256 2018-09-21 David Malcolm <dmalcolm@redhat.com>
7258 PR tree-optimization/87309
7259 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
7260 calls with pflags and alt_flags.
7261 (selftest::test_capture_of_dump_calls): Add test of interaction of
7262 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
7264 2018-09-21 Olivier Hainque <hainque@adacore.com>
7266 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
7268 2018-09-21 Olivier Hainque <hainque@adacore.com>
7270 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
7272 2018-09-21 Olivier Hainque <hainque@adacore.com>
7274 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
7275 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
7276 (VXWORKS_LIBS_RTP): Minor reordering.
7278 2018-09-21 Olivier Hainque <hainque@adacore.com>
7280 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
7281 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
7283 2018-09-21 Olivier Hainque <hainque@adacore.com>
7285 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
7286 (PTRDIFF_TYPE): Likewise.
7288 2018-09-21 Olivier Hainque <hainque@adacore.com>
7290 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
7291 triplet, similar to support for VxWorks7.
7292 * config/vxworks-dummy.h: Provide a default definition
7293 of TARGET_VXWORKS64 to 0.
7295 2018-09-21 Olivier Hainque <hainque@adacore.com>
7297 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
7298 * config/vxworks-dummy.h: here.
7300 2018-09-21 Olivier Hainque <hainque@adacore.com>
7302 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
7304 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
7305 Bo Zhou <zbo.zhou@hisilicon.com>
7307 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
7308 * config/aarch64/aarch64-tune.md: Regenerated.
7309 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
7310 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
7311 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
7313 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
7314 Julian Brown <julian@codesourcery.com>
7316 * builtins.c (get_builtin_sync_mem): Handle address spaces.
7318 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
7320 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
7321 if the call takes a static chain.
7323 2018-09-21 Martin Liska <mliska@suse.cz>
7325 * auto-profile.c (autofdo_source_profile::read): Do not
7327 (read_profile): Do not add working sets.
7328 (read_autofdo_file): Remove sum_all.
7329 (afdo_callsite_hot_enough_for_early_inline): Remove const
7331 * coverage.c (struct counts_entry): Remove gcov_summary.
7332 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
7333 do not support GCOV_TAG_PROGRAM_SUMMARY.
7334 (get_coverage_counts): Remove summary and expected
7336 * coverage.h (get_coverage_counts): Likewise.
7337 * doc/gcov-dump.texi: Remove -w option.
7338 * gcov-dump.c (dump_working_sets): Remove.
7339 (main): Do not support '-w' option.
7340 (print_usage): Likewise.
7341 (tag_summary): Likewise.
7342 * gcov-io.c (gcov_write_summary): Do not dump
7344 (gcov_read_summary): Likewise.
7345 (gcov_histo_index): Remove.
7346 (gcov_histogram_merge): Likewise.
7347 (compute_working_sets): Likewise.
7348 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
7350 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
7351 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
7352 (GCOV_HISTOGRAM_SIZE): Remove.
7353 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
7354 (struct gcov_summary): Simplify rapidly just
7355 to runs and sum_max fields.
7356 (gcov_histo_index): Remove.
7357 (NUM_GCOV_WORKING_SETS): Likewise.
7358 (compute_working_sets): Likewise.
7359 * gcov-tool.c (print_overlap_usage_message): Remove
7360 trailing empty line.
7361 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
7362 (output_lines): Remove program related line.
7363 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
7364 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
7366 (input_profile_summary): Do not read it.
7367 (merge_profile_summaries): And do not merge it.
7368 (input_symtab): Do not call removed function.
7369 * modulo-sched.c (sms_schedule): Do not print sum_max.
7370 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
7371 removed when histogram method was invented.
7372 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
7374 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
7376 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
7377 and dump selected value.
7378 * profile.c (add_working_set): Remove.
7379 (get_working_sets): Likewise.
7380 (find_working_set): Likewise.
7381 (get_exec_counts): Do not work with working sets.
7382 (read_profile_edge_counts): Do not inform as sum_max is removed.
7383 (compute_branch_probabilities): Likewise.
7384 (compute_value_histograms): Remove argument for call of
7385 get_coverage_counts.
7386 * profile.h: Do not make gcov_summary const.
7388 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
7390 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
7392 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
7394 PR tree-optimization/86990
7395 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
7396 Check that the entire merged store group is made of constants only for
7399 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
7401 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
7402 (VTABLE_VERIFICATION_SPEC): Likewise.
7403 (SANITIZER_EARLY_SPEC): Likewise.
7404 (SANITIZER_SPEC): Likewise.
7405 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
7406 * doc/invoke.texi (Link Options): Document -r.
7408 2018-09-20 Richard Biener <rguenther@suse.de>
7411 * gimplify.c (gimplify_expr): Retain alignment of
7412 addressable lvalue in dereference.
7414 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
7417 * configure.ac: Check for .loc is_stmt support.
7418 * configure, config.in: Rebuilt.
7419 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
7422 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7424 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
7425 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
7428 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7430 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
7432 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
7435 2018-09-20 Marek Polacek <polacek@redhat.com>
7437 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
7439 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
7441 PR tree-optimization/87288
7442 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
7443 into account when determining PEELING_FOR_NITERS.
7445 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
7447 PR tree-optimization/86877
7448 * tree-vect-loop.c (vect_analyze_loop_2): Call
7449 vect_verify_datarefs_alignment.
7451 2018-09-19 Marek Polacek <polacek@redhat.com>
7453 * doc/invoke.texi: Document -Wclass-conversion.
7455 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
7457 * config/pa/pa.c (pa_adjust_priority): Delete.
7458 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
7460 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
7461 (atomic_storehi): Likewise.
7462 (atomic_storesi): Likewise.
7463 (atomic_loaddi): Restore compare and swap exchange loop code.
7465 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
7467 PR rtl-optimization/86902
7468 * combine.c (try_combine): When changing the CC mode used, don't change
7469 an unrelated mode in other_insn to that new CC mode.
7471 2018-09-19 David Malcolm <dmalcolm@redhat.com>
7473 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
7474 with %T in place of calls to dump_generic_expr.
7475 (prune_runtime_alias_test_list): Likewise.
7476 (create_runtime_alias_checks): Likewise.
7477 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
7478 (vect_analyze_data_ref_dependence): Likewise.
7479 (vect_slp_analyze_data_ref_dependence): Likewise.
7480 (vect_record_base_alignment): Likewise. Use %G in place of call
7481 to dump_gimple_stmt.
7482 (vect_compute_data_ref_alignment): Likewise.
7483 (verify_data_ref_alignment): Likewise.
7484 (vect_find_same_alignment_drs): Likewise.
7485 (vect_analyze_group_access_1): Likewise.
7486 (vect_analyze_data_ref_accesses): Likewise.
7487 (dependence_distance_ge_vf): Likewise.
7488 (dump_lower_bound): Likewise.
7489 (vect_prune_runtime_alias_test_list): Likewise.
7490 (vect_find_stmt_data_reference): Likewise.
7491 (vect_analyze_data_refs): Likewise.
7492 (vect_create_addr_base_for_vector_ref): Likewise.
7493 (vect_create_data_ref_ptr): Likewise.
7494 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
7495 (vect_can_advance_ivs_p): Likewise.
7496 (vect_update_ivs_after_vectorizer): Likewise.
7497 (vect_gen_prolog_loop_niters): Likewise.
7498 (vect_prepare_for_masked_peels): Likewise.
7499 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
7500 (vect_determine_vectorization_factor): Likewise.
7501 (vect_is_simple_iv_evolution): Likewise.
7502 (vect_analyze_scalar_cycles_1): Likewise.
7503 (vect_analyze_loop_operations): Likewise.
7504 (report_vect_op): Likewise.
7505 (vect_is_slp_reduction): Likewise.
7506 (check_reduction_path): Likewise.
7507 (vect_is_simple_reduction): Likewise.
7508 (vect_create_epilog_for_reduction): Likewise.
7509 (vect_finalize_reduction:): Likewise.
7510 (vectorizable_induction): Likewise.
7511 (vect_transform_loop_stmt): Likewise.
7512 (vect_transform_loop): Likewise.
7513 (optimize_mask_stores): Likewise.
7514 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
7515 (vect_split_statement): Likewise.
7516 (vect_recog_over_widening_pattern): Likewise.
7517 (vect_recog_average_pattern): Likewise.
7518 (vect_determine_min_output_precision_1): Likewise.
7519 (vect_determine_precisions_from_range): Likewise.
7520 (vect_determine_precisions_from_users): Likewise.
7521 (vect_mark_pattern_stmts): Likewise.
7522 (vect_pattern_recog_1): Likewise.
7523 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
7524 (vect_record_max_nunits): Likewise.
7525 (vect_build_slp_tree_1): Likewise.
7526 (vect_build_slp_tree_2): Likewise.
7527 (vect_print_slp_tree): Likewise.
7528 (vect_analyze_slp_instance): Likewise.
7529 (vect_detect_hybrid_slp_stmts): Likewise.
7530 (vect_detect_hybrid_slp_1): Likewise.
7531 (vect_slp_analyze_operations): Likewise.
7532 (vect_slp_analyze_bb_1): Likewise.
7533 (vect_transform_slp_perm_load): Likewise.
7534 (vect_schedule_slp_instance): Likewise.
7535 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
7536 (vect_mark_stmts_to_be_vectorized): Likewise.
7537 (vect_init_vector_1): Likewise.
7538 (vect_get_vec_def_for_operand): Likewise.
7539 (vect_finish_stmt_generation_1): Likewise.
7540 (vect_check_load_store_mask): Likewise.
7541 (vectorizable_call): Likewise.
7542 (vectorizable_conversion): Likewise.
7543 (vectorizable_operation): Likewise.
7544 (vectorizable_load): Likewise.
7545 (vect_analyze_stmt): Likewise.
7546 (vect_is_simple_use): Likewise.
7547 (vect_get_vector_types_for_stmt): Likewise.
7548 (vect_get_mask_type_for_stmt): Likewise.
7549 * tree-vectorizer.c (increase_alignment): Likewise.
7551 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
7553 * doc/rtl.texi: Adjust vec_select description.
7554 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
7555 non-constant selectors.
7557 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
7559 * config/aarch64/aarch64-protos.h
7560 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
7561 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
7562 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
7563 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
7564 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
7565 (AARCH64_FL_PROFILE): Move index so flags are ordered.
7566 (AARCH64_ISA_RCPC8_4): New flag.
7567 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
7568 to aarch64_offset_9bit_signed_unscaled_p.
7569 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
7571 * config/aarch64/constraints.md (Ust): New constraint.
7572 * config/aarch64/predicates.md.
7573 (aarch64_9bit_offset_memory_operand): New predicate.
7574 (aarch64_rcpc_memory_operand): New predicate.
7576 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
7578 PR rtl-optimization/87361
7579 * rtlanal.c (nonzero_bits1): Revert accidental change.
7581 2018-09-19 Richard Biener <rguenther@suse.de>
7583 PR tree-optimization/87349
7584 PR tree-optimization/87342
7585 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
7587 2018-09-18 Marek Polacek <polacek@redhat.com>
7589 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
7590 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
7592 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
7594 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
7596 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
7598 PR rtl-optimization/86882
7599 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
7601 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
7603 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
7604 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
7606 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
7609 * doc/invoke.texi (Link Options): Fix formatting and grammar.
7611 2018-09-18 Richard Biener <rguenther@suse.de>
7614 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
7615 (compute_samebase_partition_bases): Likewise.
7616 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
7617 (gimple_can_coalesce_p): Simplify.
7619 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
7621 Handle a library implementation of ffs calling __builtin_ffs.
7622 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
7623 (mmix_init_libfuncs): New function: make __builtin_ffs expand
7626 2018-09-17 David Malcolm <dmalcolm@redhat.com>
7628 * diagnostic-show-locus.c (class layout_range): Add field
7630 (layout_range::layout_range): Add "original_idx" param and use it
7631 to initialize new field.
7632 (make_range): Use 0 for original_idx.
7633 (layout::layout): Pass in index to calls to
7634 maybe_add_location_range.
7635 (layout::maybe_add_location_range): Add param "original_idx" and
7636 pass it on to layout_range.
7637 (layout::print_any_labels): Pass on range->m_original_idx to
7639 (gcc_rich_location::add_location_if_nearby): Use 0 for
7641 * gcc-rich-location.h (text_range_label::get_text): Update for new
7643 (range_label_for_type_mismatch::get_text): Likewise.
7645 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
7647 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
7649 2018-09-17 David Malcolm <dmalcolm@redhat.com>
7651 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
7652 format_string_diagnostic_t.
7653 (fmtwarn_n): Likewise.
7654 * substring-locations.c
7655 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
7656 (format_warning_n_va): Convert to...
7657 (format_string_diagnostic_t::emit_warning_n_va): ...this.
7658 (format_warning_va): Convert to...
7659 (format_string_diagnostic_t::emit_warning_va): ...this.
7660 (format_warning_at_substring): Convert to...
7661 (format_string_diagnostic_t::emit_warning): ...this.
7662 (format_warning_at_substring_n): Convert to...
7663 (format_string_diagnostic_t::emit_warning_n): ...this.
7664 * substring-locations.h (class format_string_diagnostic_t): New
7666 (format_warning_va): Convert to
7667 format_string_diagnostic_t::emit_warning_va.
7668 (format_warning_n_va): Convert to
7669 format_string_diagnostic_t::emit_warning_n_va.
7670 (format_warning_at_substring): Convert to
7671 format_string_diagnostic_t::emit_warning.
7672 (format_warning_at_substring_n): Convert to
7673 format_string_diagnostic_t::emit_warning_n.
7675 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
7676 Bernd Schmidt <bernds_cb1@t-online.de>
7678 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
7681 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
7683 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
7684 operand 0 predicate to nonimmediate operand.
7685 (rint<mode>2_frndint): Remove insn pattern.
7686 (rint<mode>2): Change operand 1 predicate to general_operand.
7687 Extend operand 1 to XFmode and generate rintxf2 insn.
7688 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
7689 Do not use X87MODEF mode macro.
7690 (frndintxf2_<rounding>_i387): Rename from
7691 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
7692 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
7693 to XFmode and generate significandxf3 insn.
7695 2018-09-17 Richard Biener <rguenther@suse.de>
7697 PR tree-optimization/87328
7698 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
7699 visiting unexecutable backedges when not iterating.
7700 (do_rpo_vn): Mark all edges not executable even when not
7703 2018-09-17 Martin Jambor <mjambor@suse.cz>
7706 * doc/invoke.texi (Warning Options): Likewise.
7708 2018-09-17 Richard Biener <rguenther@suse.de>
7710 PR tree-optimization/87301
7711 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
7712 clean EH info from leftover copy assignments.
7714 2018-09-17 Martin Liska <mliska@suse.cz>
7716 PR gcov-profile/85871
7717 * gcov.c (output_intermediate_file): Fix out of bounds
7720 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
7722 * config/arc/arc.c: Object attributes for core4 not reflected
7724 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
7727 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
7729 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
7731 2018-09-17 Martin Liska <mliska@suse.cz>
7733 * doc/gcov.texi: Document new option --use-hotness-colors.
7734 * gcov.c (struct source_info): Declare new field.
7735 (source_info::source_info): Set default for maximum_count.
7736 (print_usage): Add new -q option.
7737 (process_args): Process it.
7738 (accumulate_line_info): Save src->maximum_count.
7739 (output_line_beginning): Make color line number if
7740 flag_use_hotness_colors is set.
7741 (output_line_details): Pass default argument value.
7742 (output_lines): Pass src->maximum_count.
7744 2018-09-17 Martin Liska <mliska@suse.cz>
7746 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
7747 Use processor_names table.
7748 * config/i386/i386.c (ix86_default_align): Use
7749 processor_cost_table for alignment values.
7750 (ix86_option_override_internal): Use processor_names.
7751 (ix86_function_specific_print): Likewise.
7752 * config/i386/i386.h (struct processor_costs):
7753 Add alignment values.
7754 (struct ptt): Remove and replace with const char *.
7755 * config/i386/x86-tune-costs.h (struct processor_costs):
7756 Declare default alignments for all costs.
7758 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
7760 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
7761 symbolics or VR_VARYING ranges for ABS_EXPR.
7762 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
7763 when range will wrap.
7765 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
7768 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
7769 before and after a JUMP_TABLE_DATA.
7771 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
7774 * dojump.c (do_compare_and_jump): Canonicalize function pointers
7775 when one operand is a function pointer. Use POINTER_TYPE_P and
7776 FUNC_OR_METHOD_TYPE_P.
7777 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
7778 * fold-const.c (build_range_check): Likewise.
7779 * match.pd (simple_comparison): Likewise.
7781 2018-09-14 David Malcolm <dmalcolm@redhat.com>
7784 * spellcheck.c (get_edit_distance_cutoff): New function.
7785 (selftest::test_edit_distance_unit_test_oneway): Rename to...
7786 (selftest::test_get_edit_distance_one_way): ...this.
7787 (selftest::test_get_edit_distance_unit): Rename to...
7788 (selftest::test_get_edit_distance_both_ways): ...this.
7789 (selftest::test_edit_distances): Move tests to this new function,
7790 and test some more pairs of strings. Update for above renaming.
7791 (selftest::get_old_cutoff): New function.
7792 (selftest::test_get_edit_distance_cutoff): New function.
7793 (selftest::assert_suggested_for): New function.
7794 (ASSERT_SUGGESTED_FOR): New macro.
7795 (selftest::assert_not_suggested_for): New function.
7796 (ASSERT_NOT_SUGGESTED_FOR): New macro.
7797 (selftest::test_suggestions): New function.
7798 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
7799 tests to selftest::test_edit_distances and call it. Add calls to
7800 selftest::test_get_edit_distance_cutoff and
7801 selftest::test_suggestions.
7802 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
7803 (best_match::consider): Replace hard-coded cutoff calculation with
7805 (best_match::get_cutoff): New declaration.
7806 (best_match::get_best_meaningful_candidate): Likewise.
7808 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
7810 * builtins.c (fold_builtin_strlen): Remove TODO comment.
7812 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
7815 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
7817 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
7818 terminated string literal.
7820 2018-09-14 Martin Sebor <msebor@redhat.com>
7822 * builtins.c (unterminated_array): Handle ARRAY_REF.
7823 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
7824 * builtins.h (unterminated_array): Declare extern.
7825 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
7827 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
7830 2018-09-14 Martin Sebor <msebor@redhat.com>
7831 Jeff Law <law@redhat.com>
7833 * builtins.c (unterminated_array): New.
7834 (expand_builtin_strcpy): Adjust.
7835 (expand_builtin_strcpy_args): Detect unterminated arrays.
7836 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
7837 unterminated arrays.
7838 * gimple-fold.h (get_maxval_strlen): Add argument.
7839 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
7841 * gimple-fold.c (get_range_strlen): Add argument.
7842 (get_maxval_strlen): Adjust.
7843 * gimple-fold.h (get_range_strlen): Add argument.
7845 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
7847 * config/i386/movdirintrin.h: Fix copyright year.
7849 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
7851 * reg-stack.c: Include regs.h.
7852 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
7853 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
7854 FIRST_STACK_REG, not DFmode.
7855 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
7856 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
7857 (change stack): Default register mode to the reg_raw_mode of
7858 FIRST_STACK_REG, not DFmode.
7859 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
7860 (*swapxf): Rename from swapxf.
7862 2018-09-14 Carl Love <cel@us.ibm.com>
7864 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
7865 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
7867 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
7870 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
7873 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
7876 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
7878 2018-09-14 Jason Merrill <jason@redhat.com>
7880 Fix --enable-gather-detailed-mem-stats.
7881 * hash-table.c (hash_table_usage): Change from variable to function.
7882 * hash-table.h: Adjust.
7883 * Makefile.in: Add missing dependencies on hash-table.h.
7885 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7887 PR tree-optimization/87259
7889 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
7890 execute_cse_reciprocals_1 has tried transforming.
7892 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
7894 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
7895 VR_VARYING for PLUS/MINUS_EXPR.
7897 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
7899 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
7902 2018-09-14 Richard Biener <rguenther@suse.de>
7905 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
7906 bits for the merged partition.
7908 2018-09-13 Martin Sebor <msebor@redhat.com>
7909 Bernd Edlinger <bernd.edlinger@hotmail.de>
7911 * builtins.h (c_srlen): Add argument.
7912 * builtins.c (warn_string_no_nul): New function.
7913 (c_strlen): Add argument and use it. Update recursive calls.
7914 Pass DECL argument to string_constant to get info on non
7915 terminated strings. Update *NONSTR as needed.
7916 (fold_builtin_strlen): Add argument to calls to c_strlen.
7917 Warn for unterminated arrays.
7918 (warn_string_no_null): Add prototype.
7919 * expr.c (string_constant): Update arguments. Update recursive
7920 calls appropriately. Detect missing NUL terminator and outermost
7921 declaration its missing in.
7922 Improve checks for arrays with nonzero lower bound or elements
7923 that are not a single byte. Simplify offset computation.
7924 Simplify checks for non-NUL terminated strings.
7925 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
7926 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
7928 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
7930 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
7932 * fold-const.c (c_getstr): Fix function comment. Remove unused third
7933 argument. Fix range checks.
7934 * fold-const.h (c_getstr): Adjust protoype.
7935 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
7936 string is constant but contains no NUL byte.
7938 * expr.c (string_constant): Adjust function comment.
7939 Remove bogus check for zero termination.
7941 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
7943 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
7944 (get_constant_size): Don't make STRING_CSTs larger than they are.
7945 (check_string_literal): New check function for STRING_CSTs.
7946 (output_constant): Use it.
7948 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
7951 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
7953 2018-09-13 Richard Biener <rguenther@suse.de>
7955 PR tree-optimization/87263
7956 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
7957 (struct unwind_state): Add max_rpo field.
7958 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
7959 Compute max_rpo, the max RPO number a block can be backwards reached
7960 from. Re-write non-iterating mode to a RPO ordered worklist approach,
7961 separating it from the iterating mode.
7963 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
7965 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
7966 (rfs_decision): New scheduling decision.
7968 2018-09-13 Richard Biener <rguenther@suse.de>
7971 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
7972 (vn_nary_op_insert_pieces_predicated): Do not write useless
7973 valid_dominated_by_p entry outside of the allocated storage.
7975 2018-09-13 Omar Sandoval <osandov@osandov.com>
7976 Tom de Vries <tdevries@suse.de>
7979 * dwarf2out.c (is_c): New function.
7980 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
7982 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
7985 * config/aarch64/aarch64.md (*aarch64_bfxil):
7987 * config/aarch64/constraints.md (Ulc): Define.
7988 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
7990 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
7993 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
7995 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
7996 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
7997 aarch64_layout_frame call.
7998 (aarch64_expand_epilogue): Likewise.
7999 (aarch64_initial_elimination_offset): Likewise.
8000 (aarch64_get_separate_components): Likewise.
8001 (aarch64_use_return_insn_p): Likewise.
8002 (aarch64_layout_frame): Remove unneeded check.
8004 2018-09-13 Jakub Jelinek <jakub@redhat.com>
8006 * configure.ac: Only append
8007 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
8008 gcc_config_arguments if it was never reconfigured or last reconfigure
8009 was with different arguments.
8010 * configure: Regenerated.
8012 2018-09-13 Jakub Jelinek <jakub@redhat.com>
8013 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8016 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
8017 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
8019 2018-09-13 Jakub Jelinek <jakub@redhat.com>
8021 PR tree-optimization/87287
8022 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
8023 X % C == 0 to X % (unsigned) C == 0 optimization to ...
8024 * match.pd (X % C == 0): ... here. New optimization.
8026 2018-09-12 Jakub Jelinek <jakub@redhat.com>
8029 * expr.h (maybe_optimize_mod_cmp): Declare.
8030 * expr.c (mod_inv): New function.
8031 (maybe_optimize_mod_cmp): New function.
8032 (do_store_flag): Use it.
8033 * cfgexpand.c (expand_gimple_cond): Likewise.
8035 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
8036 Julian Brown <julian@codesourcery.com>
8039 * gimplify.c (gimplify_scan_omp_clauses): Set
8040 target_firstprivatize_array_bases in OpenACC parallel and kernels
8041 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
8042 OpenACC data regions.
8044 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
8046 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
8047 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
8049 2018-09-12 Richard Biener <rguenther@suse.de>
8051 PR tree-optimization/87280
8052 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
8053 edge but unreachable target.
8054 (do_rpo_vn): For conservatively handling a PHI only mark
8055 the backedge executable but not the block reachable.
8057 2018-09-12 Richard Biener <rguenther@suse.de>
8059 PR tree-optimization/87266
8060 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
8063 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
8065 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
8067 ("trunc<BFP:mode><DFP_ALL:mode>2")
8068 ("trunc<DFP_ALL:mode><BFP:mode>2")
8069 ("extend<BFP:mode><DFP_ALL:mode>2")
8070 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
8071 according to the target operand type.
8073 2018-09-12 Jakub Jelinek <jakub@redhat.com>
8074 Andreas Krebbel <krebbel@linux.ibm.com>
8076 PR tree-optimization/86844
8077 * gimple-ssa-store-merging.c
8078 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
8079 there are any overlapping stores in between them, make sure they are
8080 also coalesced or we give up completely.
8082 2018-09-12 Jakub Jelinek <jakub@redhat.com>
8085 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
8086 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
8088 2018-09-12 Tom de Vries <tdevries@suse.de>
8090 * common.opt (gdescribe-dies): Add option.
8091 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
8092 attribute for artifical and nameless decls.
8093 (dwarf2out_register_external_die): Add description attribute to
8094 external reference die.
8095 (add_desc_attribute): New functions.
8096 (gen_subprogram_die): Add description attribute to
8097 DW_TAG_call_site_parameter.
8098 * tree-pretty-print.c (print_generic_expr_to_str): New function.
8099 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
8100 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
8102 (@item -gdescribe-dies): Add.
8104 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
8106 * tree-vrp.c (vrp_shift_undefined_p): Remove.
8107 (extract_range_from_binary_expr_1: Call
8108 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
8109 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
8112 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
8114 * gimple-ssa-warn-alloca.c
8115 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
8116 field for ALLOCA_BOUND_*_LARGE.
8118 2018-09-11 Nathan Sidwell <nathan@acm.org>
8120 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
8122 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
8124 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
8125 for clobbers. Remove obsolete comment.
8127 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
8129 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
8130 mpxchk, mpxld and mpxst types.
8131 (define_attr length_immediate): Remove all processing of mpx types.
8132 (define_attr prefix_0f): Ditto.
8133 (define_attr memory): Ditto.
8135 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
8137 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
8138 (log<mode>2): Change operand 1 predicate to general_operand.
8139 Extend operand 1 to XFmode and generate logxf3 insn.
8140 (log10<mode>2): Change operand 1 predicate to general_operand.
8141 Extend operand 1 to XFmode and generate log10xf3 insn.
8142 (log2<mode>2): Change operand 1 predicate to general_operand.
8143 Extend operand 1 to XFmode and generate log2xf3 insn.
8144 (fyl2xp1_extend<mode>xf3_i387): Remove.
8145 (log1p<mode>2): Change operand 1 predicate to general_operand.
8146 Extend operand 1 to XFmode and generate log1pxf3 insn.
8147 (fxtract_extend<mode>xf3_i387): Remove.
8148 (logb<mode>2): Change operand 1 predicate to general_operand.
8149 Extend operand 1 to XFmode and generate logbxf3 insn.
8150 (ilogb<mode>2): Change operand 1 predicate to general_operand.
8151 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
8152 (significand<mode>2): Change operand 1 predicate to general_operand.
8153 Extend operand 1 to XFmode and generate significandxf3 insn.
8155 2018-09-11 Nathan Sidwell <nathan@acm.org>
8157 * gcc.c (perror_with_name, pfatal_with_name): Delete.
8158 (load_specs): Use fatal_error.
8159 (DELETE_IF_ORDINARY, process_command): Use error.
8160 (execute, run_attempt): Use fatal_error.
8162 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
8164 * diagnostic-core.h (sorry_at): New prototype.
8165 * diagnostic.c (sorry_at): New function.
8167 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
8169 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
8170 by zero as VR_UNDEFINED.
8172 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
8174 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
8175 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
8176 (<sincos>mode2): New expander.
8177 (sincos_extend<mode>xf3_i387): Remove insn pattern.
8178 (sincos -> sin, cos splitters): Remove splitter patterns.
8179 (sincos<mode>3): Change operand 2 predicate to general_operand.
8180 Extend operand 2 to XFmode and generate sincosxf3 insn.
8181 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
8182 Change operand 3 predicate to const1_operand.
8183 (fptan_extend<mode>xf4_i387): Remove insn pattern.
8184 (tanxf2): Update operands in the call to fptanxf4_i387.
8185 (tan<mode>2): Change operand 1 predicate to general_operand.
8186 Extend operand 1 to XFmode and generate tanxf3 insn.
8187 (atan2xf3): Rename from *fpatanxf3_i387.
8188 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
8189 (atan2xf3): Remove expander.
8190 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
8191 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
8192 (atan<mode>2): Change operand 1 predicate to general_operand.
8193 Extend operand 1 to XFmode and generate atanxf3 insn.
8195 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
8197 * config/i386/i386.md (x87/SSE constant load splitter): Use
8198 memory_operand instead of nonimmediate_operand for input operand
8201 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
8203 * config/i386/i386.md (float partial SSE register stall splitter): Move
8204 splitter near its instruction pattern.
8205 (float_extend partial SSE register stall splitter): Ditto.
8206 (float_truncate partial SSE register stall splitter): Ditto.
8208 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
8211 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
8212 to speculation_safe_value_not_needed.
8215 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
8216 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
8217 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
8218 leaf_function_p, instead use has_hard_reg_initial_val.
8220 2018-09-09 Nathan Sidwell <nathan@acm.org>
8222 * gcc.h (pfatal_with_name): Don't declare here.
8223 * gcc.c (pfatal_with_name): Make static.
8225 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
8227 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
8230 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
8232 PR rtl-optimization/85458
8233 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
8234 priority hook to reduce the priority of EXPR.
8236 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
8238 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
8239 DImode for x87 on 32bit targets. Conditionally disable x87 modes
8240 with X87_ENABLE_FLOAT. Remove preparation code.
8241 (*float<SWI48:mode><MODEF:mode>2): Rename from
8242 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
8243 math using "enabled" attribute.
8244 (*floatdi<MODEF:mode>2_i387): Rename from
8245 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
8246 enable for 32bit targets only.
8247 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
8249 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
8250 as operand 1 predicate. Rewrite as define_insn_and_split.
8251 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
8253 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
8255 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
8256 to fallthru to FLOAT case.
8258 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
8261 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
8262 around folding of vec_sl to handle out of range shift values.
8264 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
8266 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
8267 Update callers to gen_fix_trunc<mode>_i387_fisttp
8268 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
8269 nonimmediate_operand.
8270 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
8271 and corresponding splitters.
8272 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
8273 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
8274 (fix_truncdi_i387_with_temp): Remove insn pattern
8275 and corresponding splitters.
8276 (fix_trunc<mode>_i387): Change operand 0 predicate to
8277 nonimmediate_operand.
8278 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
8279 and corresponding splitters.
8280 (*fistdi2_1): Remove.
8282 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
8283 (lrintxfdi2): Remove expander. Reimplement as define_insn.
8284 (*fist<mode>2_1): Remove.
8285 (fist<mode>2): Ditto.
8286 (fist<mode>2_with_temp): Remove insn pattern and corresponding
8288 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
8289 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
8290 (fistdi2_<rounding>): Change operand 0 predicate to
8291 nonimmediate_operand.
8292 (fistdi2_<rounding>_with_temp): Remove insn pattern
8293 and corresponding splitters.
8294 (fist<mode>2_<rounding>): Change operand 0 predicate to
8295 nonimmediate_operand.
8296 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
8297 and corresponding splitters.
8299 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
8301 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8303 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
8306 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
8308 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
8309 early gimple folding of vec_splat().
8310 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
8311 * gimple-fold.h: Add an extern define for tree_vec_extract().
8313 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
8315 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
8316 wrappers around TREE_TYPE comparisons.
8318 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
8321 * config/s390/predicates.md: Add nonsym_memory_operand.
8322 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
8323 contains a SYMBOL_REF, load it into an intermediate pseudo.
8324 (s390_emit_compare_and_swap): Legitimize operand.
8325 * config/s390/s390.md: Use the new nonsym_memory_operand
8326 with UNSPECV_CAS patterns.
8328 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
8331 * config/s390/s390-passes.def: New file.
8332 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
8334 (make_pass_s390_early_mach): Add declaration.
8335 * config/s390/s390.c (make_pass_s390_early_mach):
8336 (s390_option_override): Remove dynamic registration.
8337 * config/s390/t-s390: Add s390-passes.def.
8339 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
8341 * config/s390/s390.c (s390_decompose_constant_pool_ref):
8342 Remove UNSPEC_LTREL_BASE check.
8343 (annotate_constant_pool_refs): Likewise.
8344 (find_constant_pool_ref): Likewise.
8345 (find_ltrel_base): Removed.
8346 (replace_ltrel_base): Removed.
8347 (s390_mainpool_finish): Remove replace_ltrel_base call.
8348 (s390_chunkify_start): Remove pending LTREL_BASE logic.
8349 (s390_chunkify_finish): Remove replace_ltrel_base call.
8350 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
8352 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
8355 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
8356 to speculation_safe_value_not_needed.
8358 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
8359 Bernd Schmidt <bernds_cb1@t-online.de>
8361 * config/nvptx/nvptx-opts.h: New file.
8362 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
8363 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
8365 (TARGET_SM35): New macro.
8366 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
8368 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
8370 (misa=): New option.
8371 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
8373 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
8375 * config/i386/i386.md (truncdfsf2): Remove expander.
8376 (truncdfsf2_with_temp): Ditto.
8377 (truncxf<mode>2): Ditto.
8378 (*truncdfsf_fast_mixed): Remove insn pattern.
8379 (*truncdfsf_fast_i387): Ditto.
8380 (*truncdfsf_mixed): Ditto.
8381 (*truncdfsf_i387): Ditto.
8382 (*truncdfsf2_i387_1): Ditto.
8383 (*truncxfsf2_mixed): Ditto.
8384 (*truncxfdf2_mixed): Ditto.
8385 (*truncxf<mode>2_i387_noop): Ditto. Update callers
8386 to call gen_truncxf<mode>2 instead.
8387 (*truncxf<mode>2_i387): Remove.
8388 (reg->reg splitters): Remove splitter pattern.
8389 (reg->mem splitters): Ditto.
8391 (truncdfsf2): New insn pattern.
8392 (truncxf<mode>2): Ditto.
8394 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8396 * tree-ssa-math-opts.c (is_mult_by): New function.
8397 (is_square_of): Use the above.
8398 (optimize_recip_sqrt): New function.
8399 (pass_cse_reciprocals::execute): Use the above.
8401 2018-09-05 Richard Biener <rguenther@suse.de>
8404 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
8405 to zero-init the emplaced vec.
8407 2018-09-05 Martin Liska <mliska@suse.cz>
8409 PR tree-optimization/87205
8410 * tree-switch-conversion.c (pass_lower_switch::execute):
8411 Group cases for switch statements.
8413 2018-09-05 Richard Biener <rguenther@suse.de>
8415 PR tree-optimization/87217
8416 * tree-ssa-sccvn.c (vuse_valueize): New.
8417 (vn_reference_lookup_pieces): Use it.
8418 (vn_reference_lookup): Likewise.
8420 2018-09-05 Nathan Sidwell <nathan@acm.org>
8423 * stor-layout.c (place_field): Scan forwards to check last
8424 bitfield when ms_bitfield_placement is in effect.
8426 2018-09-05 Richard Biener <rguenther@suse.de>
8429 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
8432 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
8433 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
8435 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
8436 * config.gcc (extra_objs): Build it.
8437 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
8439 * config/aarch64/aarch64-passes.def
8440 (pass_tag_collision_avoidance): New pass.
8441 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
8442 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
8443 (aarch64_classify_address): Remove static qualifier.
8444 (aarch64_address_info, aarch64_address_type): Move to...
8445 * config/aarch64/aarch64-protos.h: ... here.
8446 (make_pass_tag_collision_avoidance): New function.
8447 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
8450 2018-09-05 Martin Liska <mliska@suse.cz>
8452 * doc/gcov.texi: Update documentation of humar
8454 * gcov.c (format_count): Print one decimal place, it provides
8455 more fine number of situations like '1G' vs. '1.4G'.
8457 2018-09-05 Martin Liska <mliska@suse.cz>
8460 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
8461 * optc-gen.awk: Allow 'Var' for Deprecated options in order
8462 to generate a MASK value.
8464 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
8467 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
8468 if frame pointer isn't used.
8469 (compute_frame_pointer_to_fb_displacement): Likewise.
8471 2018-09-04 Jakub Jelinek <jakub@redhat.com>
8474 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
8475 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
8476 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
8477 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
8478 and OPTION_MASK_ISA_XSAVEC_UNSET.
8480 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
8482 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
8483 XOR operations in NAND case.
8485 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
8487 * wide-int-range.cc (wide_int_range_convert): New.
8488 * wide-int-range.h (wide_int_range_convert): New.
8489 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
8490 code into wide_int_range_convert.
8491 (extract_range_into_wide_ints): Do not munge anti range constants
8492 into the entire domain. Just return the range back.
8494 2018-09-04 Martin Liska <mliska@suse.cz>
8496 * genmatch.c (output_line_directive): Add new argument
8498 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
8500 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
8502 * doc/invoke.texi (Option Summary): Add whitespace.
8504 * doc/invoke.texi (Option Summary): Add -Waligned-new.
8506 2018-09-04 Richard Biener <rguenther@suse.de>
8508 PR tree-optimization/87211
8509 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
8510 backedge value we're supposed to treat as VARYING also number
8511 the PHI to VARYING in case it got a different value-number already.
8513 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
8515 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
8516 (extract_range_from_binary_expr_1): Do not call
8517 vrp_can_optimize_bit_op.
8518 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
8520 (wide_int_range_get_mask_and_bounds): New.
8521 (wide_int_range_optimize_bit_op): New.
8522 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
8523 (wide_int_range_bit_and): Same.
8524 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
8525 (wide_int_range_optimize_bit_op): New.
8526 (wide_int_range_get_mask_and_bounds): New.
8528 2018-09-04 Richard Biener <rguenther@suse.de>
8530 PR tree-optimization/87176
8531 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
8532 variable. When value-numbering a virtual PHI node make sure
8533 to not value-number to the backedge value.
8535 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
8537 * doc/extend.texi (Long Long, Hex Floats): Document support for
8538 long long and hex floats in more recent versions of ISO C++.
8540 2018-09-03 Richard Biener <rguenther@suse.de>
8542 PR tree-optimization/87177
8543 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
8546 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
8548 * bb-reorder.c (edge_order): Convert to C-qsort-style
8549 tri-state comparator.
8550 (reorder_basic_blocks_simple): Change std::stable_sort to
8553 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
8555 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
8556 tri-state comparator.
8557 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
8559 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
8561 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
8562 (mergesort): ... here as maximum count for using netsort.
8563 (gcc_qsort): Set nlim to 3 if stable sort is requested.
8564 (gcc_stablesort): New.
8565 * system.h (gcc_stablesort): Declare.
8567 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
8569 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
8570 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
8571 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
8573 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
8575 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
8576 lxsdx and stxsdx alternatives.
8577 (*mov<mode>_hardfloat64): Ditto.
8578 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
8580 2018-09-03 Richard Biener <rguenther@suse.de>
8582 PR tree-optimization/87200
8583 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
8586 2018-09-03 Martin Liska <mliska@suse.cz>
8588 PR tree-optimization/87201
8589 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
8590 Fix parenthesis in an expression.
8592 2018-09-03 Richard Biener <rguenther@suse.de>
8594 PR tree-optimization/87197
8595 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
8596 visited. CSE the VN_INFO hashtable lookup.
8598 PR tree-optimization/87169
8599 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
8600 iterating make sure there's no extra backedges from irreducible
8601 regions feeding the header. Mark the destination block
8604 2018-09-03 Martin Liska <mliska@suse.cz>
8607 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
8608 * common/common-targhooks.c (default_get_valid_option_values):
8610 * common/common-targhooks.h (default_get_valid_option_values):
8612 * common/config/i386/i386-common.c: Move processor_target_table
8614 (ix86_get_valid_option_values): New function.
8615 (TARGET_GET_VALID_OPTION_VALUES): New macro.
8616 * config/i386/i386.c (struct ptt): Move to i386-common.c.
8617 (PTA_*): Move all defined masks into i386-common.c.
8618 (ix86_function_specific_restore): Use new processor_cost_table.
8619 * config/i386/i386.h (struct ptt): Moved from i386.c.
8620 (struct pta): Likewise.
8621 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
8622 * doc/tm.texi.in: Likewise.
8623 * opt-suggestions.c (option_proposer::suggest_option):
8624 Pass prefix to build_option_suggestions.
8625 (option_proposer::get_completions): Likewise.
8626 (option_proposer::build_option_suggestions): Use the new target
8628 * opts.c (struct option_help_tuple): New struct.
8629 (print_filtered_help): Use the new target hook.
8631 2018-09-03 Martin Liska <mliska@suse.cz>
8634 * predict.c (set_even_probabilities): Add likely_edges
8635 argument and handle cases where we have precisely one
8637 (combine_predictions_for_bb): Catch also likely_edges.
8638 (tree_predict_by_opcode): Handle gswitch statements.
8639 * tree-cfg.h (find_case_label_for_value): New declaration.
8640 (find_taken_edge_switch_expr): Likewise.
8641 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
8642 Find pivot in decision tree based on probabily, not by number of
8645 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
8647 * doc/standards.texi (Standards): Update Objective-C reference.
8649 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
8651 * doc/install.texi (Prerequisites): Update link for MPC.
8653 2018-09-01 Michael Matz <matz@suse.de>
8655 PR tree-optimization/87074
8656 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
8657 PHIs for outer-loop uses.
8659 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
8661 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
8662 * doc/invoke.texi (C Dialect Options): Ditto.
8664 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
8666 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
8668 2018-08-31 Richard Biener <rguenther@suse.de>
8670 PR tree-optimization/87168
8671 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
8672 (rpo_elim::eliminate_avail): When OP was not visited it must
8675 2018-08-31 David Malcolm <dmalcolm@redhat.com>
8677 * tree-vrp.c (copy_value_range): Convert param "from" from
8678 "value_range *" to "const value_range *".
8679 (range_is_null): Likewise for param "vr".
8680 (range_int_cst_p): Likewise.
8681 (range_int_cst_singleton_p): Likewise.
8682 (symbolic_range_p): Likewise.
8683 (value_ranges_intersect_p): Likewise for both params.
8684 (value_range_nonnegative_p): Likewise for param "vr".
8685 (value_range_constant_singleton): Likewise.
8686 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
8687 (extract_range_into_wide_ints): Likewise for param "vr".
8688 (extract_range_from_multiplicative_op): Likewise for params "vr0"
8690 (vrp_can_optimize_bit_op): Likewise.
8691 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
8693 (extract_range_from_unary_expr): Likewise.
8694 (debug_value_range): Likewise for param "vr".
8695 (value_range::dump): Add "const" qualifier.
8696 (vrp_prop::check_array_ref): Convert local "vr" from
8697 "value_range *" to "const value_range *".
8698 (vrp_prop::check_mem_ref): Likewise.
8699 (vrp_prop::visit_stmt): Likewise for local "old_vr".
8700 (vrp_intersect_ranges_1): Likewise for param "vr_1".
8701 (vrp_intersect_ranges): Likewise.
8702 (simplify_stmt_for_jump_threading): Likewise for local "vr".
8703 (vrp_prop::vrp_finalize): Likewise.
8704 * tree-vrp.h (value_range::dump): Add "const" qualifier.
8705 (vrp_intersect_ranges): Add "const" qualifier to params as above.
8706 (extract_range_from_unary_expr): Likewise.
8707 (value_range_constant_singleton): Likewise.
8708 (symbolic_range_p): Likewise.
8709 (copy_value_range): Likewise.
8710 (extract_range_from_binary_expr_1): Likewise.
8711 (range_int_cst_p): Likewise.
8712 (vrp_set_zero_nonzero_bits): Likewise.
8713 (range_int_cst_singleton_p): Likewise.
8715 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
8717 * config/aarch64/arm_neon.h (vabsd_s64): New.
8718 (vnegd_s64): Likewise.
8720 2018-08-31 Martin Jambor <mjambor@suse.cz>
8722 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
8724 2018-08-31 Martin Liska <mliska@suse.cz>
8726 * ipa-icf.c (sem_item::add_type): Use
8727 sem_item::m_type_hash_cache.
8728 * ipa-icf.h: Move the cache from sem_item_optimizer
8731 2018-08-31 Nathan Sidwell <nathan@acm.org>
8733 * doc/extend.texi (Backwards Compatibility): Remove implicit
8734 extern C leeway of () being (...).
8736 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8738 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
8740 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
8744 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
8746 2018-08-31 Jakub Jelinek <jakub@redhat.com>
8749 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
8750 gen_int_mode. Formatting fixes.
8752 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
8754 * target.def (custom_function_descriptors): Improve documentation.
8755 * doc/tm.texi.in (Trampolines): Expand discussion of function
8756 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
8757 beginning of the section.
8758 * doc/tm.texi: Regenerated.
8760 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
8762 * cfg.h (class auto_edge_flag): Spell out the template-id of the
8763 base class in the initializer list. This is a workaround for
8764 building with older GCC.
8765 (class auto_bb_flag): Likewise.
8767 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
8769 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
8770 (altivec_vcmpequ<VI_char>_p): Remove star.
8771 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
8773 (expand_strncmp_vec_sequence): New function.
8774 (emit_final_str_compare_vec): New function.
8775 (expand_strn_compare): Add support for vector strncmp.
8776 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
8777 length specification to bytes.
8778 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
8779 (vcmpnezb_p): New pattern.
8780 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
8781 for option -mstring-compare-inline-limit.
8783 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
8785 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
8786 (PTA_SKYLAKE): Add PTA_AES.
8787 (PTA_GOLDMONT): Likewise.
8789 2018-08-29 Jan Hubicka <jh@suse.cz>
8792 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
8793 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
8795 2018-08-29 Jan Hubicka <jh@suse.cz>
8797 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
8799 (hash_tree): Do not visit TYPE_STUB_DECL.
8800 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
8801 stream TYPE_STUB_DECL.
8802 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
8803 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
8804 after free_lang_data.
8805 (type_in_anonymous_namespace_p): Likewise.
8807 2018-08-29 Jan Hubicka <jh@suse.cz>
8809 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
8810 comment that it has to be even number.
8811 (class sreal): Change m_sig type to int32_t.
8812 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
8813 int64_t for temporary calculations.
8814 (sreal_verify_basics): Drop one bit from minimum and maximum.
8816 2018-08-30 Richard Biener <rguenther@suse.de>
8818 PR tree-optimization/87147
8819 * tree-ssa-sccvn.c (SSA_VISITED): New function.
8820 (visit_phi): When the degenerate result is from the backedge and
8821 we didn't visit its definition yet drop to VARYING.
8822 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
8824 2018-08-29 Jan Hubicka <jh@suse.cz>
8826 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
8828 (hash_tree): Likewise.
8830 2018-08-29 Jan Hubicka <jh@suse.cz>
8832 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
8833 and TYPE_NEXT_REF_TO.
8835 2018-08-29 Jan Hubicka <jh@suse.cz>
8837 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
8838 comment that it has to be even number.
8839 (class sreal): Change m_sig type to int32_t.
8840 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
8841 int64_t for temporary calculations.
8842 (sreal_verify_basics): Drop one bit from minimum and maximum.
8844 2018-08-30 Tamar Christina <tamar.christina@arm.com>
8846 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
8848 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
8851 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
8852 if to_add is negative.
8854 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
8857 * builtins.c (c_strlen): Improve range checks.
8859 2018-08-29 Martin Sebor <msebor@redhat.com>
8860 Jeff Law <law@redhat.com>
8862 PR tree-optimization/86714
8863 PR tree-optimization/86711
8864 * builtins.c (c_strlen): Add arguments to call to string_constant.
8865 * expr.c (string_constant): Add argument. Detect missing nul
8866 terminator and outermost declaration it's missing in.
8867 * expr.h (string_constant): Add argument.
8868 * fold-const.c (read_from_constant_string): Add arguments to call to
8870 (c_getstr): Likewise.
8871 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
8873 * tree-ssa-strlen.c (get_stridx): Likewise.
8875 2018-08-29 Jan Hubicka <jh@suse.cz>
8877 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
8878 Do not stream DECL_VINDEX.
8879 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
8880 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
8881 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
8883 2018-08-29 Richard Biener <rguenther@suse.de>
8885 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
8886 virtual operands that are not default defs to honor region
8888 (rpo_vn_valueize): Remove ineffective code here.
8890 2018-08-29 Richard Biener <rguenther@suse.de>
8892 PR tree-optimization/87132
8893 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
8894 when skipping defs reachable over backedges.
8896 2018-08-29 Richard Biener <rguenther@suse.de>
8898 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
8899 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
8900 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
8901 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
8902 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
8903 (vn_reference_lookup_3): Remove use of const_parms.
8904 (free_rpo_vn): Do not free const_parms.
8905 (do_rpo_vn): Do not call init_const_parms.
8906 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
8907 SSA_NAME_POINTS_TO_READONLY_MEMORY.
8908 (call_may_clobber_ref_p_1): Likewise.
8910 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
8913 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
8915 (-ftree-scev-cprop): Document.
8917 2018-08-29 Jan Hubicka <jh@suse.cz>
8919 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
8921 (sreal constructor): Update.
8922 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
8923 sreal:operator/): Update.
8925 2018-08-29 Martin Liska <mliska@suse.cz>
8927 * tree-switch-conversion.c (switch_conversion::expand):
8928 Strenghten assumption about gswitch statements.
8930 2018-08-29 Richard Biener <rguenther@suse.de>
8932 PR tree-optimization/87117
8933 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
8934 re-value-number released SSA VDEFs.
8936 2018-08-29 Richard Biener <rguenther@suse.de>
8938 PR tree-optimization/87126
8939 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
8941 2018-08-28 Jim Wilson <jimw@sifive.com>
8943 * config/riscv/pic.md: Rewrite.
8944 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
8946 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
8947 (SOFTF, default_load, softload, softstore): New.
8949 2018-08-28 Jeff Law <law@redhat.com>
8951 * fold-const.c (fold_binary_loc): Remove recently added assert.
8953 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
8955 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
8956 to OP parmeter of generated function.
8958 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
8960 PR tree-optimization/87009
8961 * match.pd: Add boolean optimizations.
8963 2018-08-28 Martin Sebor <msebor@redhat.com>
8966 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
8967 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
8968 (pass_walloca::gate): Use it.
8969 (alloca_call_type): Same.
8970 (pass_walloca::execute): Same.
8971 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
8973 2018-08-28 David Malcolm <dmalcolm@redhat.com>
8975 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
8976 GCC_VERSION for usage of "__gcc_dump_printf__" format from
8979 2018-08-28 Richard Biener <rguenther@suse.de>
8981 PR tree-optimization/87124
8982 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
8983 constants before looking up avail.
8985 2018-08-28 Jakub Jelinek <jakub@redhat.com>
8988 * calls.c (maybe_warn_nonstring_arg): Punt early if
8989 warn_stringop_overflow is zero. Don't call get_range_strlen
8990 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
8991 Swap comparison operands to have constants on rhs. Only use
8992 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
8993 increment lenrng[0].
8995 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
8997 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
8998 use of tree_to_shwi. Remove duplicated test for the size being
8999 a whole number of bytes.
9001 2018-08-28 Richard Biener <rguenther@suse.de>
9003 PR tree-optimization/87117
9004 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
9005 Handle removed stmt without LHS (GIMPLE_NOP).
9007 2018-08-28 Richard Biener <rguenther@suse.de>
9009 PR tree-optimization/87117
9010 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
9011 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
9013 2018-08-28 Richard Biener <rguenther@suse.de>
9015 PR tree-optimization/87117
9016 * tree-ssa-pre.c (compute_avail): Do not make expressions
9017 with predicated values available.
9018 (get_expr_value_id): Assert we do not run into predicated value
9021 2018-08-28 Richard Biener <rguenther@suse.de>
9023 PR tree-optimization/87117
9024 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
9025 get virtual operands.
9026 (get_expr_operands): Handle STRING_CST like other decls.
9028 2018-08-28 Martin Liska <mliska@suse.cz>
9030 * tree.h: Update documentation of fndecl_built_in_p
9034 2018-08-27 Jeff Law <law@redhat.com>
9035 PR tree-optimization/87110
9036 * tree-ssa-dse.c (compute_trims): Handle non-constant
9039 2018-08-27 Martin Sebor <msebor@redhat.com>
9041 PR tree-optimization/86914
9042 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
9044 2018-08-27 Martin Sebor <msebor@redhat.com>
9046 PR tree-optimization/87112
9047 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
9048 the type of the bound argument.
9050 2018-08-27 Jeff Law <law@redhat.com>
9052 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
9053 type does not have a TYPE_SIZE_UNIT.
9055 2018-08-27 Steve Ellcey <sellcey@cavium.com>
9057 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
9058 with include of backend.h.
9060 2018-08-27 Richard Biener <rguenther@suse.de>
9062 PR tree-optimization/86927
9063 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
9064 use const cond reduction code.
9066 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
9068 PR tree-optimization/85758
9069 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
9071 2018-08-27 David Malcolm <dmalcolm@redhat.com>
9074 * diagnostic-show-locus.c (class layout_range): Update for
9075 conversion of show_caret_p to a tri-state.
9076 (layout_range::layout_range): Likewise.
9077 (make_range): Likewise.
9078 (layout::maybe_add_location_range): Likewise.
9079 (layout::should_print_annotation_line_p): Don't show annotation
9080 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
9081 (layout::get_state_at_point): Update for conversion of
9082 show_caret_p to a tri-state. Bail out early for
9083 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
9084 underlining or source colorization.
9085 (gcc_rich_location::add_location_if_nearby): Update for conversion
9086 of show_caret_p to a tri-state.
9087 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
9088 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
9090 (selftest::test_one_liner_labels): Likewise.
9091 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
9092 conversion of show_caret_p to a tri-state.
9093 * pretty-print.c (text_info::set_location): Likewise.
9094 * pretty-print.h (text_info::set_location): Likewise.
9095 * substring-locations.c (format_warning_n_va): Likewise.
9096 * tree-diagnostic.c (default_tree_printer): Likewise.
9097 * tree-pretty-print.c (newline_and_indent): Likewise.
9099 2018-08-27 David Malcolm <dmalcolm@redhat.com>
9102 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
9103 line above for line-insertion fix-it hints.
9104 (selftest::test_fixit_insert_containing_newline): Update the
9105 expected results, and add a test with line-numbering enabled.
9107 2018-08-27 Martin Liska <mliska@suse.cz>
9110 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
9111 params with DECL_HAS_VALUE_EXPR_P.
9113 2018-08-27 Martin Liska <mliska@suse.cz>
9115 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
9116 selected expansion strategy.
9118 2018-08-27 Martin Liska <mliska@suse.cz>
9120 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
9121 * builtins.c (is_builtin_fn): Likewise.
9122 * attribs.c (diag_attr_exclusions): Use new function
9123 fndecl_built_in_p and remove check for FUNCTION_DECL if
9125 (builtin_mathfn_code): Likewise.
9126 (fold_builtin_expect): Likewise.
9127 (fold_call_expr): Likewise.
9128 (fold_builtin_call_array): Likewise.
9129 (fold_call_stmt): Likewise.
9130 (set_builtin_user_assembler_name): Likewise.
9131 (is_simple_builtin): Likewise.
9132 * calls.c (gimple_alloca_call_p): Likewise.
9133 (maybe_warn_nonstring_arg): Likewise.
9134 * cfgexpand.c (expand_call_stmt): Likewise.
9135 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
9136 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
9137 (cgraph_node::verify_node): Likewise.
9138 * cgraphclones.c (build_function_decl_skip_args): Likewise.
9139 (cgraph_node::create_clone): Likewise.
9140 * config/arm/arm.c (arm_insert_attributes): Likewise.
9141 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
9142 * dse.c (scan_insn): Likewise.
9143 * expr.c (expand_expr_real_1): Likewise.
9144 * fold-const.c (operand_equal_p): Likewise.
9145 (fold_binary_loc): Likewise.
9146 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
9147 * gimple-low.c (lower_stmt): Likewise.
9148 * gimple-pretty-print.c (dump_gimple_call): Likewise.
9149 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
9151 * gimple.c (gimple_build_call_from_tree): Likewise.
9152 (gimple_call_builtin_p): Likewise.
9153 (gimple_call_combined_fn): Likewise.
9154 * gimplify.c (gimplify_call_expr): Likewise.
9155 (gimple_boolify): Likewise.
9156 (gimplify_modify_expr): Likewise.
9157 (gimplify_addr_expr): Likewise.
9158 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
9159 * ipa-cp.c (determine_versionability): Likewise.
9160 * ipa-fnsummary.c (compute_fn_summary): Likewise.
9161 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
9162 * ipa-split.c (visit_bb): Likewise.
9163 (split_function): Likewise.
9164 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
9165 * lto-cgraph.c (input_node): Likewise.
9166 * lto-streamer-out.c (write_symbol): Likewise.
9167 * omp-low.c (setjmp_or_longjmp_p): Likewise.
9168 (lower_omp_1): Likewise.
9169 * predict.c (strip_predict_hints): Likewise.
9170 * print-tree.c (print_node): Likewise.
9171 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
9172 * trans-mem.c (is_tm_irrevocable): Likewise.
9173 (is_tm_load): Likewise.
9174 (is_tm_simple_load): Likewise.
9175 (is_tm_store): Likewise.
9176 (is_tm_simple_store): Likewise.
9177 (is_tm_abort): Likewise.
9178 (tm_region_init_1): Likewise.
9179 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
9180 * tree-cfg.c (verify_gimple_call): Likewise.
9181 (move_stmt_r): Likewise.
9182 (stmt_can_terminate_bb_p): Likewise.
9183 * tree-eh.c (lower_eh_constructs_2): Likewise.
9184 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
9185 * tree-inline.c (remap_gimple_stmt): Likewise.
9186 (copy_bb): Likewise.
9187 (estimate_num_insns): Likewise.
9188 (fold_marked_statements): Likewise.
9189 * tree-sra.c (scan_function): Likewise.
9190 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
9191 (optimize_stack_restore): Likewise.
9192 (pass_fold_builtins::execute): Likewise.
9193 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
9194 (mark_all_reaching_defs_necessary_1): Likewise.
9195 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
9196 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
9197 (pass_forwprop::execute): Likewise.
9198 * tree-ssa-loop-im.c (stmt_cost): Likewise.
9199 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
9200 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
9201 * tree-ssa-strlen.c (get_string_length): Likewise.
9202 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
9203 (find_func_aliases_for_call): Likewise.
9204 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
9205 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
9206 * tree-tailcall.c (find_tail_calls): Likewise.
9207 * tree.c (need_assembler_name_p): Likewise.
9208 (free_lang_data_in_decl): Likewise.
9209 (get_call_combined_fn): Likewise.
9210 * ubsan.c (is_ubsan_builtin_p): Likewise.
9211 * varasm.c (incorporeal_function_p): Likewise.
9212 * tree.h (DECL_BUILT_IN): Remove and replace with
9214 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
9215 (fndecl_built_in_p): New.
9217 2018-08-27 Martin Liska <mliska@suse.cz>
9219 PR tree-optimization/86847
9220 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
9221 Dump also subtree probability.
9222 (switch_decision_tree::do_jump_if_equal): New function.
9223 (switch_decision_tree::emit_case_nodes): Handle special
9224 situations in balanced tree that can be emitted much simpler.
9225 Fix calculation of probabilities that happen in tree expansion.
9226 * tree-switch-conversion.h (struct cluster): Add
9228 (struct simple_cluster): Likewise.
9229 (struct case_tree_node): Add new function has_child.
9230 (do_jump_if_equal): New.
9232 2018-08-27 Martin Liska <mliska@suse.cz>
9234 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
9235 Add new argument to bit_test_cluster constructor.
9236 (bit_test_cluster::emit): Set bits really number of values
9238 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
9239 probability argument.
9240 * tree-switch-conversion.h (struct bit_test_cluster):
9241 Add m_handles_entire_switch.
9243 2018-08-27 Martin Liska <mliska@suse.cz>
9245 PR tree-optimization/86702
9246 * tree-switch-conversion.c (jump_table_cluster::emit):
9247 Make probabilities even for values in jump table
9248 according to number of cases handled.
9249 (switch_decision_tree::compute_cases_per_edge): Pass
9250 argument to reset_out_edges_aux function.
9251 (switch_decision_tree::analyze_switch_statement): Likewise.
9252 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
9255 2018-08-27 Martin Liska <mliska@suse.cz>
9257 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
9258 cfun argument explicitly.
9259 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
9260 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
9261 function gimple_switch_default_bb.
9262 (convert_switch_statements):
9264 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
9265 * stmt.c (label_to_block_fn): Use label_to_block and pass
9266 cfun argument explicitly and use gimple_switch_label_bb.
9267 (expand_case): Likewise.
9268 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
9269 cfun argument explicitly. Likewise.
9270 (make_edges_bb): Likewise.
9271 (make_cond_expr_edges): Likewise.
9272 (get_cases_for_edge): Likewise.
9273 (make_gimple_switch_edges): Likewise.
9274 (label_to_block_fn): Likewise.
9275 (label_to_block): Likewise.
9276 (make_goto_expr_edges): Likewise.
9277 (make_gimple_asm_edges): Likewise.
9278 (main_block_label): Likewise.
9279 (group_case_labels_stmt): Likewise.
9280 (find_taken_edge_computed_goto): Likewise.
9281 (find_taken_edge_switch_expr): Likewise.
9282 (gimple_verify_flow_info): Likewise.
9283 (gimple_redirect_edge_and_branch): Likewise.
9284 (gimple_switch_label_bb): New function.
9285 (gimple_switch_default_bb): Likewise.
9286 (gimple_switch_edge): Likewise.
9287 (gimple_switch_default_edge): Likewise.
9288 * tree-cfg.h (label_to_block_fn): Remove and replace ...
9289 (label_to_block): ... with this.
9290 (gimple_switch_label_bb): New.
9291 (gimple_switch_default_bb): Likewise.
9292 (gimple_switch_edge): Likewise.
9293 (gimple_switch_default_edge): Likewise.
9294 * tree-cfgcleanup.c (convert_single_case_switch): Use
9295 new gimple functions and pass new argument to label_to_block.
9296 (cleanup_control_flow_bb):
9297 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
9298 cfun argument explicitly.
9299 (make_eh_edges): Likewise.
9300 (redirect_eh_dispatch_edge): Likewise.
9301 (lower_resx): Likewise.
9302 (lower_eh_dispatch): Likewise.
9303 (maybe_remove_unreachable_handlers): Likewise.
9304 (unsplit_eh): Likewise.
9305 (cleanup_empty_eh): Likewise.
9306 (verify_eh_edges): Likewise.
9307 (verify_eh_dispatch_edge): Likewise.
9308 * tree-ssa-dom.c (record_edge_info): Likewise.
9309 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
9310 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
9311 (thread_through_normal_block): Likewise.
9312 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
9313 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
9314 * tree-switch-conversion.c (switch_conversion::collect): Use new
9316 (switch_conversion::check_final_bb): Likewise.
9317 (switch_conversion::gather_default_values): Pass new argument
9319 (switch_conversion::build_constructors): Likewise.
9320 (switch_decision_tree::compute_cases_per_edge): Use new
9321 gimple_switch_edge function.
9322 (switch_decision_tree::analyze_switch_statement): Pass new argument
9324 (switch_decision_tree::try_switch_expansion): Use
9325 gimple_switch_default_edge.
9326 * tree-vrp.c (find_switch_asserts): Pass new argument
9328 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
9329 (vr_values::simplify_switch_using_ranges): Likewise.
9331 2018-08-27 Richard Biener <rguenther@suse.de>
9333 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
9334 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
9336 * tree-ssa-sccvn.h (struct vn_pval): New structure.
9337 (struct vn_nary_op_s): Add unwind_to member. Add
9338 predicated_values flag and put result into a union together
9339 with a linked list of vn_pval.
9340 (struct vn_ssa_aux): Add name member to make maintaining
9341 a map of SSA name to vn_ssa_aux possible. Remove no longer
9342 needed info, dfsnum, low, visited, on_sccstack, use_processed
9343 and range_info_anti_range_p members.
9344 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
9345 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
9347 (vn_valueize): New global.
9348 (vn_context_bb): Likewise.
9349 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
9350 VN_INFO_PTR_INFO): Remove.
9351 * tree-ssa-sccvn.c: ... (rewrite)
9352 (pass_fre::execute): For -O2+ initialize loops and run
9353 RPO VN in optimistic mode (iterating). For -O1 and -Og
9354 run RPO VN in non-optimistic mode.
9355 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
9356 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
9357 * doc/invoke.texi (sccvn-max-scc-size): Remove.
9358 (rpo-vn-max-loop-depth): Document.
9359 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
9360 when valuezing the VUSE signals we walked out of the region.
9361 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
9362 (phi_translate): Set VN context block to use for availability
9364 (compute_avail): Likewise.
9365 (pre_valueize): New function.
9366 (pass_pre::execute): Adjust to the RPO VN API.
9368 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
9369 (propagate_constants_for_unrolling): Remove.
9370 (tree_unroll_loops_completely): Perform value-numbering
9371 on the unrolled bodies loop parent.
9373 2018-08-27 Richard Biener <rguenther@suse.de>
9375 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
9376 for partial antic compute.
9378 2018-08-27 Jakub Jelinek <jakub@redhat.com>
9380 PR rtl-optimization/87065
9381 * combine.c (simplify_if_then_else): Formatting fix.
9382 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
9384 (known_cond): Don't return const_true_rtx for vector modes. Use
9385 CONST0_RTX instead of const0_rtx. Formatting fixes.
9387 2018-08-27 Martin Liska <mliska@suse.cz>
9389 PR gcov-profile/87069
9390 * gcov.c (process_file): Record files already processed
9391 and warn about a file being processed multiple times.
9393 2018-08-27 Martin Liska <mliska@suse.cz>
9396 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9397 Set default values for x_aarch64_*_string strings.
9398 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
9399 prefix. For -mabi do not print '=ABI' in help and use
9400 <option_value> format for -msve-vector-bits and -moverride
9403 2018-08-26 Jeff Law <law@redhat.com>
9405 * config/mips/frame-header-opt.c: Include "backend.h" rather than
9408 2018-08-26 Marek Polacek <polacek@redhat.com>
9410 PR c++/87029, Implement -Wredundant-move.
9411 * doc/invoke.texi: Document -Wredundant-move.
9413 2018-08-25 Martin Sebor <msebor@redhat.com>
9415 PR tree-optimization/87059
9416 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
9417 to the same type as the other.
9418 * fold-const.c (fold_binary_loc): Assert expectation.
9420 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
9422 * config/darwin.c (machopic_legitimize_pic_address): Clean up
9423 extraneous parentheses, dead code section and formatting.
9425 2018-08-24 David Malcolm <dmalcolm@redhat.com>
9428 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
9429 wide enough for jumps in the line-numbering to be visible.
9430 (layout::print_gap_in_line_numbering): New member function.
9431 (layout::calculate_line_spans): When using line numbering, merge
9432 line spans that are only 1 line apart.
9433 (diagnostic_show_locus): When printing line numbers, show gaps in
9434 line numbering directly, rather than printing headers.
9435 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
9436 line-numbering with multiple line spans.
9437 (selftest::test_fixit_insert_containing_newline_2): Add test of
9438 line-numbering, in which the spans are close enough to be merged.
9440 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
9442 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
9443 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
9444 * tree-vrp.c (range_is_nonnull): Remove.
9445 (range_includes_zero_p): Accept value_range instead of min/max.
9446 (extract_range_from_binary_expr_1): Do not early bail on
9448 Use range_includes_zero_p instead of range_is_nonnull.
9449 (extract_range_from_unary_expr): Use range_includes_zero_p instead
9450 of range_is_nonnull.
9451 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
9452 special case VR_ANTI_RANGE.
9453 (vrp_finalize): Same.
9454 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
9456 (range_is_nonnull): Remove.
9457 * vr-values.c (vrp_stmt_computes_nonzero): Use
9458 range_includes_zero_p instead of range_is_nonnull.
9459 (extract_range_basic): Pass value_range to range_includes_zero_p
9460 instead of range_is_nonnull.
9462 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
9464 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
9465 * emit-rtl.h (rtl_data): Remove return_bnd.
9466 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
9467 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
9468 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
9469 (POINTER_BOUNDS_MODE): Remove definition.
9470 (make_pointer_bounds_mode): Remove.
9471 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
9472 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
9473 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
9474 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
9475 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
9476 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
9477 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
9479 * config/i386/i386-modes.def (BND32, BND64): Remove.
9480 * config/i386/i386.c (dbx_register_map): Remove bound registers.
9481 (dbx64_register_map): Ditto.
9482 (svr4_dbx_register_map): Ditto.
9483 (indirect_thunk_bnd_needed): Remove.
9484 (indirect_thunks_bnd_used): Ditto.
9485 (indirect_return_bnd_needed): Ditto.
9486 (indirect_return_via_cx_bnd): Ditto.
9487 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
9488 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
9489 (output_indirect_thunk): Ditto. Remove need_prefix argument.
9490 (output_indirect_thunk_function): Remove handling of
9491 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
9492 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
9493 (ix86_save_reg): Remove handling of crtl->return_bnd.
9494 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
9495 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
9496 and UNSPEC_BNDLX_ADDR.
9497 (ix86_output_indirect_branch_via_reg): Remove handling of
9498 indirect_thunk_prefix_bnd.
9499 (ix86_output_indirect_branch_via_push): Ditto.
9500 (ix86_output_function_return): Ditto.
9501 (ix86_output_indirect_function_return): Ditto.
9502 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
9503 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
9504 (CALL_USED_REGISTERS): Ditto.
9505 (REG_ALLOC_ORDER): Update for removal of bound registers.
9506 (HI_REGISTER_NAMES): Ditto.
9507 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
9508 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
9509 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
9510 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
9511 (FIRST_PSEUDO_REG): Update.
9512 (BND): Remove mode iterator.
9513 * config/i386/predicates.md (bnd_mem_operator): Remove.
9515 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
9517 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
9520 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
9522 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
9523 the case in which the permute needs only a single element and
9524 repeats for every vector of the result. Extend that case to
9525 handle variable-length vectors.
9526 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
9528 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
9531 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
9532 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
9534 2018-08-24 Richard Biener <rguenther@suse.de>
9536 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
9537 bb_flags_allocated members.
9538 (auto_flag): New RAII class for allocating flags.
9539 (auto_edge_flag): New RAII class for allocating edge flags.
9540 (auto_bb_flag): New RAII class for allocating bb flags.
9541 * cfgloop.c (verify_loop_structure): Allocate temporary edge
9543 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
9544 in favor of temporarily allocated BB flag.
9545 * hsa-brig.c: Re-order includes.
9546 * hsa-dump.c: Likewise.
9547 * hsa-regalloc.c: Likewise.
9548 * print-rtl.c: Likewise.
9549 * profile-count.c: Likewise.
9551 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
9554 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
9557 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
9560 * wide-int-range.cc (wide_int_range_div): Do not ignore result
9561 from wide_int_range_multiplicative_op.
9563 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9565 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
9568 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
9570 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
9573 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
9575 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
9576 full GENERIC expression used for replacement.
9578 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
9580 * tree-vrp.c (abs_extent_range): Remove.
9581 (extract_range_into_wide_ints): Pass wide ints by reference.
9582 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
9583 Pass wide ints by reference in all calls to
9584 extract_range_into_wide_ints.
9585 * wide-int-range.cc (wide_int_range_div): New.
9586 * wide-int-range.h (wide_int_range_div): New.
9587 (wide_int_range_includes_zero_p): New.
9588 (wide_int_range_zero_p): New.
9590 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
9592 * config/aarch64/aarch64.md (arches): New enum.
9593 (arch): New enum attr.
9594 (arch_enabled): New attr.
9595 (enabled): Now uses arch_enabled only.
9596 (simd, sve, fp16): Removed attribute.
9597 (fp): Attr now defined in terms of 'arch'.
9598 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
9599 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
9600 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
9601 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
9602 attributes into 'arch'.
9603 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
9604 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
9605 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
9606 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
9607 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
9608 'simd' attribute into 'arch'.
9609 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
9610 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
9611 Convert use of 'fp' attribute to 'arch'.
9612 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
9613 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
9615 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
9616 (different modes) Merge 'fp' and 'simd' into 'arch'.
9617 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
9620 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
9622 PR rtl-optimization/87026
9623 * expmed.c (canonicalize_comparison): If we can no longer create
9624 pseudoregisters, don't.
9626 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
9629 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
9631 * config/arm/arm.c (speculation_barrier_libfunc): New static
9633 (arm_init_libfuncs): Initialize it.
9634 (arm_emit_speculation_barrier): New function.
9635 * config/arm/arm.md (speculation_barrier): Call
9636 arm_emit_speculation_barrier for architectures that do not have
9638 (speculation_barrier_insn): Only match on Armv7 or later.
9640 2018-08-23 Richard Biener <rguenther@suse.de>
9643 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
9646 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
9648 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
9649 of single-vector TBLs.
9650 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
9653 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
9656 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
9657 aarch64_evpc_tbl guard.
9659 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
9661 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
9664 2018-08-22 Martin Sebor <msebor@redhat.com>
9667 * tree-pretty-print.c (pretty_print_string): Add argument.
9668 (dump_generic_node): Call to pretty_print_string with string size.
9670 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
9672 PR rtl-optimization/86771
9673 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
9674 of two SETs into those two SETs, one to be placed at i2, if that SETs
9675 destination is modified between i2 and i3.
9677 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
9679 PR tree-optimization/86725
9680 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
9682 (vect_analyze_scalar_cycles_1): Check it.
9684 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
9686 PR tree-optimization/86725
9687 * tree-vect-loop.c (vect_is_simple_reduction): When treating
9688 an outer loop phi as a double reduction, make sure that the
9689 single user of the phi result is an inner loop phi.
9691 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
9693 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
9694 grouped stores with gaps to a strided group.
9696 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
9698 * tree-vect-stmts.c (get_group_load_store_type)
9699 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
9700 first statement in a group.
9702 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
9704 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
9705 the sequence used in gcc/gcc.c.
9707 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
9710 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
9713 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
9715 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
9716 Darwin10-specific unwinder-shim.
9717 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
9718 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
9719 New to cater for Darwin10 Rosetta.
9721 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
9723 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
9724 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
9726 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
9731 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
9732 (dwarf2out_switch_text_section): Generate a local label for the
9733 second function sub-section and apply it as the second FDE start
9735 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
9736 second sub-section start.
9738 2018-08-22 Richard Biener <rguenther@suse.de>
9740 PR tree-optimization/86988
9741 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
9743 2018-08-22 Richard Biener <rguenther@suse.de>
9745 PR tree-optimization/86945
9746 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
9748 2018-08-22 Alexandre Oliva <oliva@adacore.com>
9750 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
9751 a comment about how uses of r2 for .sdata2 come about.
9753 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
9755 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
9757 2018-08-21 Marek Polacek <polacek@redhat.com>
9759 PR c++/86981, Implement -Wpessimizing-move.
9760 * doc/invoke.texi: Document -Wpessimizing-move.
9762 2018-08-21 Jan Hubicka <jh@suse.cz>
9764 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
9765 * tree.h (is_redundant_typedef): Remove.
9766 * dwarf2out.c (is_redundant_typedef): Turn into static function.
9768 2018-08-21 Jan Hubicka <jh@suse.cz>
9770 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
9773 2018-08-21 Tamar Christina <tamar.christina@arm.com>
9775 * expmed.c (extract_low_bits): Reject invalid subregs early.
9777 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
9780 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
9783 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
9785 * config/vxworks.h: Guard vxworks_asm_out_constructor and
9786 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
9787 * config/vxworks.c: Likewise.
9789 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
9791 * config/vxworks.c: Set targetm.have_ctors_dtors
9792 if HAVE_INITFINI_ARRAY_SUPPORT.
9793 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
9794 if HAVE_INITFINI_ARRAY_SUPPORT.
9796 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
9798 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
9799 default search path for VxWorks < 7.
9801 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
9803 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
9804 (get_format_string): Refer to c_getstr.
9806 2018-08-21 Tom de Vries <tdevries@suse.de>
9808 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
9809 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
9810 (debuginfo_early_stop): Declare.
9811 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
9812 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
9813 (debuginfo_early_stop): New function.
9814 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
9815 and debuginfo_early_stop.
9816 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
9817 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
9818 (general_init): Call debuginfo_early_init.
9819 (finalize): Call debuginfo_fini.
9820 (do_compile): Call debuginfo_init.
9821 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
9823 (@item -fdump-debug, @item -fdump-earlydebug): Add.
9825 2018-08-21 Tom de Vries <tdevries@suse.de>
9827 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
9828 flag_dump_noaddr and flag_dump_unnumbered.
9830 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
9832 * wide-int-range.cc (wide_int_range_abs): New.
9833 (wide_int_range_order_set): Rename from wide_int_range_min_max.
9834 * wide-int-range.h (wide_int_range_abs): New.
9835 (wide_int_range_min_max): New.
9836 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
9837 case to call wide_int_range_abs.
9838 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
9839 (extract_range_from_abs_expr): Delete.
9841 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
9844 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
9845 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
9848 2018-08-20 Nathan Sidwell <nathan@acm.org>
9849 Jeff Law <law@redhat.com>
9851 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
9852 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
9854 2018-08-20 David Malcolm <dmalcolm@redhat.com>
9857 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
9858 (decl_attributes): Likewise.
9859 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
9861 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
9862 * diagnostic-core.h (class auto_diagnostic_group): New class.
9863 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
9864 (diagnostic_report_diagnostic): Handle the first diagnostics within
9866 (emit_diagnostic): Add auto_diagnostic_group instance.
9868 (inform_n): Likewise.
9869 (warning): Likewise.
9870 (warning_at): Likewise.
9871 (warning_n): Likewise.
9872 (pedwarn): Likewise.
9873 (permerror): Likewise.
9875 (error_n): Likewise.
9876 (error_at): Likewise.
9878 (fatal_error): Likewise.
9879 (internal_error): Likewise.
9880 (internal_error_no_backtrace): Likewise.
9881 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
9882 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
9883 * diagnostic.h (struct diagnostic_context): Add fields
9884 "diagnostic_group_nesting_depth",
9885 "diagnostic_group_emission_count", "begin_group_cb",
9887 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
9888 Add auto_diagnostic_group instance(s).
9889 (find_explicit_erroneous_behavior): Likewise.
9890 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
9891 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
9892 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
9893 (gimplify_va_arg_expr): Likewise.
9894 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
9895 (HSA_SORRY_AT): Likewise.
9896 * ipa-devirt.c (compare_virtual_tables): Likewise.
9897 (warn_odr): Likewise.
9898 * multiple_target.c (expand_target_clones): Likewise.
9899 * opts-common.c (cmdline_handle_error): Likewise.
9900 * reginfo.c (globalize_reg): Likewise.
9901 * substring-locations.c (format_warning_n_va): Likewise.
9902 * tree-inline.c (expand_call_inline): Likewise.
9903 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
9904 * tree-ssa-loop-niter.c
9905 (do_warn_aggressive_loop_optimizations): Likewise.
9906 * tree-ssa-uninit.c (warn_uninit): Likewise.
9907 * tree.c (warn_deprecated_use): Likewise.
9909 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
9912 * config/i386/i386.md (eh_return): Always update EH return
9913 address in word_mode.
9915 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
9917 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
9918 TARGET_SPLIT_COMPLEX_ARG is defined.
9920 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
9922 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
9924 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
9927 * expr.c (expand_assignment): Assert that bitpos is positive.
9928 (store_field): Likewise
9929 (expand_expr_real_1): Make sure that bitpos is positive.
9930 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
9933 2018-08-20 Nathan Sidwell <nathan@acm.org>
9935 * Makefile.in (CPP_ID_DATA_H): Delete.
9936 (CPP_INTERNAL_H): Don't add it.
9937 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
9938 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
9940 2018-08-20 Richard Biener <rguenther@suse.de>
9942 PR tree-optimization/78655
9943 * tree-vrp.c (extract_range_from_binary_expr_1): Make
9944 pointer + offset nonnull if either operand is nonnull work.
9946 2018-08-20 Tom de Vries <tdevries@suse.de>
9948 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
9949 unless the referenced die describes the added property using
9950 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
9951 Otherwise, add a DW_AT_location to the referenced die.
9953 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
9956 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
9957 register_operand when calling ix86_set_reg_reg_cost.
9958 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
9959 Set *total to 0 for operands that satisfy x86_64_immediate_operand
9960 predicate and to 1 otherwise.
9962 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
9964 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
9965 emit a diagnostic that it is not supported and reset the option.
9966 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
9967 supported and consume the option. (ASM_FINAL_SPEC): New.
9969 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
9971 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
9974 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
9976 C-SKY port: Documentation
9978 * doc/extend.texi (C-SKY Function Attributes): New section.
9979 * doc/invoke.texi (Option Summary): Add C-SKY options.
9980 (C-SKY Options): New section.
9981 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
9983 2018-08-17 Jojo <jijie_rong@c-sky.com>
9984 Huibin Wang <huibin_wang@c-sky.com>
9985 Sandra Loosemore <sandra@codesourcery.com>
9986 Chung-Lin Tang <cltang@codesourcery.com>
9988 C-SKY port: Backend implementation
9990 * config/csky/*: New.
9991 * common/config/csky/*: New.
9993 2018-08-17 Jojo <jijie_rong@c-sky.com>
9994 Huibin Wang <huibin_wang@c-sky.com>
9995 Sandra Loosemore <sandra@codesourcery.com>
9996 Chung-Lin Tang <cltang@codesourcery.com>
9997 Andrew Jenner <andrew@codesourcery.com>
9999 C-SKY port: Configury
10001 * config.gcc (csky-*-*): New.
10002 * configure.ac: Add csky to targets for dwarf2 debug_line support.
10003 * configure: Regenerated.
10005 2018-08-17 David Malcolm <dmalcolm@redhat.com>
10007 * dump-context.h: Include "dumpfile.h".
10008 (dump_context::dump_printf_va): Convert final param from va_list
10009 to va_list *. Convert from ATTRIBUTE_PRINTF to
10010 ATTRIBUTE_GCC_DUMP_PRINTF.
10011 (dump_context::dump_printf_loc_va): Likewise.
10012 * dumpfile.c: Include "stringpool.h".
10013 (make_item_for_dump_printf_va): Delete.
10014 (make_item_for_dump_printf): Delete.
10015 (class dump_pretty_printer): New class.
10016 (dump_pretty_printer::dump_pretty_printer): New ctor.
10017 (dump_pretty_printer::emit_items): New member function.
10018 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
10020 (dump_pretty_printer::emit_item): New member function.
10021 (dump_pretty_printer::stash_item): New member function.
10022 (dump_pretty_printer::format_decoder_cb): New member function.
10023 (dump_pretty_printer::decode_format): New member function.
10024 (dump_context::dump_printf_va): Reimplement in terms of
10025 dump_pretty_printer.
10026 (dump_context::dump_printf_loc_va): Convert final param from va_list
10028 (dump_context::begin_scope): Reimplement call to
10029 make_item_for_dump_printf.
10030 (dump_printf): Update for change to dump_printf_va.
10031 (dump_printf_loc): Likewise.
10032 (selftest::test_capture_of_dump_calls): Convert "stmt" from
10033 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
10034 with %T, %E, and %G.
10035 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
10036 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
10037 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
10038 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
10039 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
10040 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
10041 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
10042 within a dump_printf_loc call to "%wu".
10043 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
10044 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
10045 missing space after "=".
10046 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
10047 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
10048 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
10049 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
10050 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
10051 duplicate "vectorized" from message.
10053 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
10055 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
10056 polyNxK_t element's TYPE_STRING_FLAG.
10058 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
10060 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
10061 (they were unnamed before). Fix comments.
10063 2018-08-17 Nathan Sidwell <nathan@acm.org>
10065 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
10067 2018-08-17 Richard Biener <rguenther@suse.de>
10069 PR tree-optimization/86841
10070 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
10072 2018-08-17 Martin Liska <mliska@suse.cz>
10074 * common.opt: Remove Warn, Init and Report for options with
10075 Ignore/Deprecated flag. Warning is done automatically for
10077 * config/i386/i386.opt: Likewise.
10078 * config/ia64/ia64.opt: Likewise.
10079 * config/rs6000/rs6000.opt: Likewise.
10080 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
10081 Remove usage of flag_check_pointer_bounds.
10082 * lto-wrapper.c (merge_and_complain): Do not handle
10083 OPT_fcheck_pointer_bounds.
10084 (append_compiler_options): Likewise.
10085 * opt-functions.awk: Do not handle Deprecated.
10086 * optc-gen.awk: Check that Var, Report and Init are not
10087 used for an option with Ignore/Deprecated flag.
10088 * opts-common.c (decode_cmdline_option): Do not report
10090 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
10092 * opts.h (struct cl_option): Remove cl_deprecated flag.
10093 (CL_ERR_DEPRECATED): Remove error enum value.
10095 2018-08-17 Richard Biener <rguenther@suse.de>
10097 PR middle-end/86505
10098 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
10099 across a va-arg-pack using call adjust its return value accordingly.
10101 2018-08-16 Martin Sebor <msebor@redhat.com>
10103 PR tree-optimization/86853
10104 * gimple-ssa-sprintf.c (struct format_result): Rename member.
10105 (struct fmtresult): Add member and initialize it in ctors.
10106 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
10107 (format_string): Handle %S the same as %ls. Set MAYFAIL.
10108 (format_directive): Set POSUNDER4K when MAYFAIL is set.
10109 (parse_directive): Handle %C same as %c.
10110 (sprintf_dom_walker::compute_format_length): Adjust.
10111 (is_call_safe): Adjust.
10113 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
10115 * builtins.c (c_strlen): Add new parameter eltsize. Use it
10116 for determining how to count the elements.
10117 * builtins.h (c_strlen): Adjust prototype.
10118 * expr.c (string_constant): Add new parameter mem_size.
10119 Set *mem_size appropriately.
10120 * expr.h (string_constant): Adjust protoype.
10121 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
10122 * gimple-fold.h (get_range_strlen): Adjust prototype.
10123 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
10124 (format_string): Call get_string_length with eltsize.
10126 2018-08-16 David Malcolm <dmalcolm@redhat.com>
10128 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
10129 to emit the span, rather than setting it as the prefix.
10131 2018-08-16 David Malcolm <dmalcolm@redhat.com>
10133 * diagnostic-show-locus.c (layout::start_annotation_line): Add
10134 "margin_char" parameter, defaulting to space. Use it in place
10135 of pp_space for the initial part of the margin.
10136 (layout::print_leading_fixits): Use '+' when filling the margin
10137 of line-insertion fix-it hints.
10139 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
10141 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
10144 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
10146 * config/rs6000/altivec.md: Don't set length attribute to the default
10148 * config/rs6000/darwin.md: Ditto.
10149 * config/rs6000/dfp.md: Ditto.
10150 * config/rs6000/htm.md: Ditto.
10151 * config/rs6000/rs6000.md: Ditto.
10152 * config/rs6000/sync.md: Ditto.
10153 * config/rs6000/vsx.md: Ditto.
10155 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
10157 * config/rs6000/altivec.md: Don't set length attribute to the default
10158 value, for branch instructions.
10159 * config/rs6000/darwin.md: Ditto.
10160 * config/rs6000/rs6000.md: Ditto.
10162 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
10164 * config/rs6000/rs6000.md (length): Always define as const_int 4.
10165 (unnamed conditional branch define_insn): Set length to 4 or 8
10166 depending on offset.
10167 (<bd>_<mode>): Similar, for alternative 0.
10168 (<bd>tf_<mode>): Ditto.
10170 2018-08-16 Tamar Christina <tamar.christina@arm.com>
10172 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
10174 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
10176 * doc/rtl.texi: Replace old RTX class names with new names.
10179 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
10181 * expmed.h (canonicalize_comparison): New declaration.
10182 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
10183 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
10184 * optabs.c (prepare_cmp_insn): Likewise.
10185 * rtl.h (unsigned_condition_p): New function which checks if a
10186 comparison operator is unsigned.
10188 2018-08-16 Nathan Sidwell <nathan@acm.org>
10190 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
10191 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
10193 2018-08-16 Tamar Christina <tamar.christina@arm.com>
10196 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
10197 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
10198 (mov<mov>): ..this and enable unconditionally.
10200 2018-08-16 Tamar Christina <tamar.christina@arm.com>
10202 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
10204 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
10206 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
10207 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
10208 "Common" with "Target".
10210 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
10212 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
10213 * doc/invoke.texi (mmitigate-rop): Remove.
10214 * config/i386/i386.c: Do not include "regrename.h".
10215 (ix86_rop_should_change_byte_p, reg_encoded_number)
10216 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
10218 (ix86_reorg): Remove call to ix86_mitigate_rop.
10219 * config/i386/i386.md (attr "modrm_class"): Remove.
10220 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
10221 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
10222 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
10224 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
10226 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
10227 allow folding of mergeh() and mergel() for the float and double types.
10228 (fold_mergehl_helper): Rework to handle building a permute tree
10231 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
10233 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
10236 2018-08-15 David Malcolm <dmalcolm@redhat.com>
10238 * common.opt (fdiagnostics-show-labels): New option.
10239 * diagnostic-show-locus.c (class layout_range): Add field
10241 (class layout): Add field "m_show_labels_p".
10242 (layout_range::layout_range): Add param "label" and use it to
10243 initialize m_label.
10244 (make_range): Pass in NULL for new "label" param of layout_range's
10246 (layout::layout): Initialize m_show_labels_p.
10247 (layout::maybe_add_location_range): Pass in loc_range->m_label
10248 when constructing layout_range instances.
10249 (struct line_label): New struct.
10250 (layout::print_any_labels): New member function.
10251 (layout::print_line): Call it if label-printing is enabled.
10252 (selftest::test_one_liner_labels): New test.
10253 (selftest::test_diagnostic_show_locus_one_liner): Call it.
10254 * diagnostic.c (diagnostic_initialize): Initialize
10255 context->show_labels_p.
10256 * diagnostic.h (struct diagnostic_context): Add field
10258 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
10259 -fno-diagnostics-show-labels.
10260 * dwarf2out.c (gen_producer_string): Add
10261 OPT_fdiagnostics_show_labels to the ignored options.
10262 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
10264 (gcc_rich_location::maybe_add_expr): Likewise.
10265 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
10266 label" param, defaulting to NULL.
10267 (gcc_rich_location::add_expr): Add "label" param.
10268 (gcc_rich_location::maybe_add_expr): Likewise.
10269 (class text_range_label): New class.
10270 (class range_label_for_type_mismatch): New class.
10271 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
10272 of format_warning_va.
10273 (fmtwarn_n): Likewise for new params of format_warning_n_va.
10274 * lto-wrapper.c (merge_and_complain): Add
10275 OPT_fdiagnostics_show_labels to the "pick one setting" options.
10276 (append_compiler_options): Likewise to the dropped options.
10277 (append_diag_options): Likewise to the passed-on options.
10278 * opts.c (common_handle_option): Handle the new option.
10279 * selftest-diagnostic.c
10280 (test_diagnostic_context::test_diagnostic_context): Enable
10282 * substring-locations.c: Include "gcc-rich-location.h".
10283 (format_warning_n_va): Add "fmt_label" and "param_label" params
10284 and use them as appropriate.
10285 (format_warning_va): Add "fmt_label" and "param_label" params,
10286 passing them on to format_warning_n_va.
10287 (format_warning_at_substring): Likewise.
10288 (format_warning_at_substring_n): Likewise.
10289 * substring-locations.h (format_warning_va): Add "fmt_label" and
10290 "param_label" params.
10291 (format_warning_n_va): Likewise.
10292 (format_warning_at_substring): Likewise.
10293 (format_warning_at_substring_n): Likewise.
10294 * toplev.c (general_init): Initialize global_dc->show_labels_p.
10296 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
10299 * builtins.c (expand_builtin_memcmp): Do not expand the call
10300 when overflow is detected.
10302 2018-08-15 Martin Sebor <msebor@redhat.com>
10304 PR tree-optimization/71625
10305 * config/aarch64/aarch64-builtins.c
10306 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
10308 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
10310 * config/s390/s390.c (s390_reorg): Remove loop.
10312 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
10315 (darwin_function_switched_text_sections): Delete.
10316 * gcc/config/darwin.h
10317 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
10319 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
10322 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
10323 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
10324 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
10326 2018-08-15 Martin Liska <mliska@suse.cz>
10328 PR tree-optimization/86925
10329 * predict.c (expr_expected_value_1): When taking
10330 later predictor, assign also probability.
10331 Use fold_build2_initializer_loc in order to fold
10332 the expression in -frounding-math.
10334 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
10336 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
10338 (expand_vec_perm_1): Try the new method.
10340 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
10343 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
10344 Check whether lra_live_max_point is 0 before dividing.
10346 2018-08-14 Martin Sebor <msebor@redhat.com>
10348 PR tree-optimization/86650
10349 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
10350 (vrp_prop::check_mem_ref): Same.
10352 2018-08-13 Liu Hao <lh_mouse@126.com>
10354 * pretty-print.c (eat_esc_sequence): Swap the foreground and
10355 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
10356 and clear it thereafter, as it only works for DBCS.
10358 2018-08-13 Liu Hao <lh_mouse@126.com>
10360 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
10361 handle returned by _get_osf_handle().
10363 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
10365 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
10366 for folding vec_perm.
10368 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
10370 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
10371 Add support for gimple-folding of vec_pack() and vec_unpack()
10374 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
10376 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
10377 vec_xst variants to the list.
10378 (rs6000_gimple_fold_builtin): Add support for folding unaligned
10379 vector loads and stores.
10381 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
10383 * config.gcc (rs6000-ibm-aix4.x): Delete.
10384 (rs6000-ibm-aix5.1): Delete.
10385 (rs6000-ibm-aix5.2): Delete.
10386 (rs6000-ibm-aix5.3): Delete.
10387 * config/rs6000/aix43.h: Delete.
10388 * config/rs6000/aix51.h: Delete.
10389 * config/rs6000/aix52.h: Delete.
10390 * config/rs6000/t-aix43: Delete.
10392 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
10394 * config/s390/s390.c (s390_decompose_constant_pool_ref):
10396 (s390_decompose_address): Factor out constant pool ref
10399 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
10401 * config/nds32/nds32-predicates.c
10402 (nds32_can_use_bclr_p): Change return type as bool.
10403 (nds32_can_use_bset_p): Ditto.
10404 (nds32_can_use_btgl_p): Ditto.
10405 (nds32_can_use_bitci_p): Ditto.
10406 * config/nds32/nds32-protos.h
10407 (nds32_can_use_bclr_p): Change declaration.
10408 (nds32_can_use_bset_p): Ditto.
10409 (nds32_can_use_btgl_p): Ditto.
10410 (nds32_can_use_bitci_p): Ditto.
10412 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
10414 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
10415 Support -msched-prolog-epilog option.
10416 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
10418 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
10420 * common/config/nds32/nds32-common.c
10421 (nds32_option_optimization_table): Enalbe -malways-align.
10423 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
10425 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
10427 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
10428 OPT_misr_secure_ case.
10429 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
10430 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
10431 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
10433 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
10434 (nds32_isr_info): New field security_level.
10435 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
10436 * config/nds32/nds32.md (return_internal): Consider critical attribute.
10437 * config/nds32/nds32.opt (misr-secure): New option.
10438 * config/nds32/nds32_init.inc: New file.
10439 * config/nds32/nds32_isr.h: New file.
10441 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
10443 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
10444 Update comment for atomic instructions.
10445 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
10446 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
10448 (atomic_loaddi): Revise fence expansion to only emit fence prior to
10449 load for __ATOMIC_SEQ_CST model.
10450 (atomic_loaddi_1): Remove float register target.
10451 (atomic_storedi): Handle CONST_INT values.
10452 (atomic_storedi_1): Remove float register source. Add special case
10454 (memory_barrier): New expander and insn.
10456 2018-08-11 Jakub Jelinek <jakub@redhat.com>
10458 PR tree-optimization/86835
10459 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
10460 new_stmt after def_gsi, make sure to insert new_square_stmt after
10461 that stmt, not 2 stmts before it.
10463 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
10466 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
10469 2018-08-10 Martin Liska <mliska@suse.cz>
10472 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
10474 * builtins.c (expand_builtin_expect_with_probability):
10476 (expand_builtin_expect_with_probability): New function.
10477 (build_builtin_expect_predicate): Add new argumnet probability
10478 for BUILT_IN_EXPECT_WITH_PROBABILITY.
10479 (fold_builtin_expect):
10482 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
10483 * builtins.h (fold_builtin_expect): Set new argument.
10484 * doc/extend.texi: Document __builtin_expect_with_probability.
10485 * doc/invoke.texi: Likewise.
10486 * gimple-fold.c (gimple_fold_call): Pass new argument.
10487 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
10488 also BUILT_IN_EXPECT_WITH_PROBABILITY.
10489 * predict.c (get_predictor_value): New function.
10490 (expr_expected_value): Add new argument probability. Assume
10491 that predictor and probability are always non-null.
10492 (expr_expected_value_1): Likewise. For __builtin_expect and
10493 __builtin_expect_with_probability set probability. Handle
10494 combination in binary expressions.
10495 (tree_predict_by_opcode): Simplify code by simply calling
10496 get_predictor_value.
10497 (pass_strip_predict_hints::execute): Add handling of
10498 BUILT_IN_EXPECT_WITH_PROBABILITY.
10499 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
10501 * tree.h (DECL_BUILT_IN_P): New function.
10503 2018-08-10 Martin Liska <mliska@suse.cz>
10505 PR tree-optimization/85799
10506 * passes.def: Add argument for pass_strip_predict_hints.
10507 * predict.c (class pass_strip_predict_hints): Add new argument
10509 (strip_predictor_early): New function.
10510 (pass_strip_predict_hints::execute): Call the function to
10512 (strip_predict_hints): New function.
10513 * predict.def: Fix comment.
10515 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
10517 * Makefile.in: Clarify which tm.texi to copy over to assert the
10518 right to grant a GFDL license for all.
10520 2018-08-09 Jeff Law <law@redhat.com>
10522 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
10525 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
10527 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
10530 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
10532 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
10533 reductions for variable-length vectors.
10535 2018-08-09 David Malcolm <dmalcolm@redhat.com>
10538 * common.opt (fdiagnostics-show-line-numbers): New option.
10539 * diagnostic-show-locus.c (class layout): Add fields
10540 "m_show_line_numbers_p" and "m_linenum_width";
10541 (num_digits): New function.
10542 (test_num_digits): New function.
10543 (layout::layout): Initialize new fields. Update m_x_offset
10544 logic to handle any left margin.
10545 (layout::print_source_line): Print line number when requested.
10546 (layout::start_annotation_line): New member function.
10547 (layout::print_annotation_line): Call it.
10548 (layout::print_leading_fixits): Likewise.
10549 (layout::print_trailing_fixits): Likewise. Update calls to
10550 move_to_column for new parameter.
10551 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
10552 it to potentially call start_annotation_line.
10553 (layout::show_ruler): Call start_annotation_line.
10554 (selftest::test_line_numbers_multiline_range): New selftest.
10555 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
10556 and selftest::test_line_numbers_multiline_range.
10557 * diagnostic.c (diagnostic_initialize): Initialize
10558 show_line_numbers_p.
10559 * diagnostic.h (struct diagnostic_context): Add field
10560 "show_line_numbers_p".
10561 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
10562 -fno-diagnostics-show-line-numbers.
10563 * dwarf2out.c (gen_producer_string): Add
10564 OPT_fdiagnostics_show_line_numbers to the ignored options.
10565 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
10566 one setting" options.
10567 (append_compiler_options): Likewise to the dropped options.
10568 (append_diag_options): Likewise to the passed-on options.
10569 * opts.c (common_handle_option): Handle the new option.
10570 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
10572 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
10574 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
10575 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
10576 third argument of type "const signed char" to descriptions of
10577 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
10578 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
10579 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
10580 __builtin_bcdsub_ov functions.
10582 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
10584 PR tree-optimization/86858
10585 * tree-vect-loop.c (vect_is_simple_reduction): Restore
10586 flow_bb_inside_loop_p calls.
10588 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
10590 PR tree-optimization/86871
10591 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
10592 instead of gimple_assign_lhs.
10594 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
10597 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
10598 register constraint to operand 0.
10599 (add<mode>3_carryinC): Likewise.
10600 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
10602 2018-08-09 Martin Liska <mliska@suse.cz>
10605 * common.opt: Remove extra line.
10607 2018-08-09 Martin Liska <mliska@suse.cz>
10609 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
10610 at the end of a line, make first letter capital and end up
10611 a sentence with a dot.
10612 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
10613 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
10614 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
10615 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
10616 (PARAM_MAX_ISL_OPERATIONS): Likewise.
10617 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
10618 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
10619 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
10620 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
10621 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
10622 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
10623 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
10624 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
10625 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
10626 (PARAM_TREE_REASSOC_WIDTH): Likewise.
10627 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
10628 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
10629 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
10631 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
10634 * config/s390/s390.c (s390_option_override_internal): Reduce the
10635 stack-clash-protection-probe-interval param if it would be too big
10638 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
10641 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
10642 * config/m68k/m68k.c (handle_move_double): Don't call
10643 m68k_final_prescan_insn.
10644 (m68k_adjust_decorated_operand): Renamed from
10645 m68k_final_prescan_insn, remove first and third operand and
10647 (print_operand): Call it.
10648 (print_operand_address): Call it.
10650 2018-08-08 Nathan Sidwell <nathan@acm.org>
10652 * diagnostic.c (diagnostic_report_current_module): Use
10653 linemap_included_from & linemap_included_from_linemap.
10655 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
10657 * config/aarch64/aarch64-cores.def: Add phecda core.
10658 * config/aarch64/aarch64-tune.md: Regenerate.
10659 * doc/invoke.texi: Add phecda core.
10661 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
10664 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
10666 * config/s390/s390.md ("movti"): Add more alternatives for
10667 constant to GPR copies.
10669 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
10671 * config/s390/s390.c: Fix whitespace damage throughout the file.
10672 * config/s390/s390.h: Likewise.
10673 * config/s390/tpf.h: Likewise.
10675 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
10677 * config/s390/s390.c (s390_loadrelative_operand_p):
10678 Remove TARGET_CPU_ZARCH usages.
10679 (s390_rtx_costs): Likewise.
10680 (s390_legitimate_constant_p): Likewise.
10681 (s390_cannot_force_const_mem): Likewise.
10682 (legitimate_reload_constant_p): Likewise.
10683 (s390_preferred_reload_class): Likewise.
10684 (legitimize_pic_address): Likewise.
10685 (legitimize_tls_address): Likewise.
10686 (s390_split_branches): Removed.
10687 (s390_add_execute): Removed.
10688 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
10689 (s390_mainpool_start): Likewise.
10690 (s390_mainpool_finish): Likewise.
10691 (s390_mainpool_cancel): Removed.
10692 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
10693 (s390_chunkify_cancel): Likewise.
10694 (s390_return_addr_rtx): Likewise.
10695 (s390_register_info): Remove split_branches_pending_p uages.
10696 (s390_optimize_register_info): Likewise.
10697 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
10698 split_branches_pending_p usages.
10699 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
10700 (s390_load_got): Likewise.
10701 (s390_expand_split_stack_prologue): Likewise.
10702 (output_asm_nops): Likewise.
10703 (s390_function_profiler): Likewise.
10704 (s390_emit_call): Likewise.
10705 (s390_conditional_register_usage): Likewise.
10706 (s390_optimize_prologue): Likewise.
10707 (s390_reorg): Remove TARGET_CPU_ZARCH and
10708 split_branches_pending_p usages.
10709 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
10711 (s390_output_indirect_thunk_function): Likewise.
10712 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
10713 (TARGET_CPU_ZARCH_P): Removed.
10714 (struct machine_function): Remove split_branches_pending_p.
10715 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
10717 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
10719 * common/config/s390/s390-common.c (processor_flags_table):
10721 * config.gcc: Remove with_arch/with_tune support.
10722 * config/s390/2064.md: Remove cpu attribute comparisons.
10723 * config/s390/driver-native.c (s390_host_detect_local_cpu):
10725 * config/s390/linux.h (ASM_SPEC):
10726 Remove -march support.
10727 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
10728 Use a table to get an arch level.
10729 * config/s390/s390-opts.h (enum processor_type):
10730 Remove enum values.
10731 * config/s390/s390.c
10732 (processor_table): Remove entries, add arch_level values.
10733 (s390_issue_rate): Remove cases.
10734 (s390_option_override): Adjust
10735 s390_option_override_internal() call.
10736 (s390_option_override_internal): Remove deprecation warning.
10737 (s390_valid_target_attribute_tree): Adjust
10738 s390_option_override_internal() call.
10739 * config/s390/s390.h (struct s390_processor):
10740 Share with s390-c.c, add arch_level field.
10741 * config/s390/s390.md:
10742 Remove occurrences in cpu attribute.
10743 * config/s390/s390.opt: Remove -march/-mtune support.
10744 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
10745 * doc/invoke.texi: Remove deprecation warning.
10747 2018-08-08 Luis Machado <luis.machado@linaro.org>
10749 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
10751 (qdf24xx_tunings): Set vector cost structure to
10752 qdf24xx_vector_cost.
10754 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
10755 <register_sextend>: Set to 3.
10757 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
10760 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
10761 * config/aarch64/aarch64-simd.md
10762 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
10763 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
10764 (aarch64_frecpx<mode>): ...this new pattern.
10765 * config/aarch64/aarch64-simd-builtins.def: Remove comment
10766 about aarch64_frecp<FRECP:frecp_suffix><mode>.
10768 2018-08-07 Martin Liska <mliska@suse.cz>
10770 PR middle-end/83023
10771 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
10772 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
10773 * predict.def (PRED_MALLOC_NONNULL): New predictor.
10774 * doc/extend.texi: Document that malloc attribute adds
10777 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
10780 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10781 Define to speculation_safe_value_not_needed.
10783 2018-08-06 Jeff Law <law@redhat.com>
10785 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
10786 the vr_values instance to cprop_into_stmt.
10787 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
10788 (cprop_operand): Also query EVRP to determine if OP is a constant.
10790 2018-08-06 Nathan Sidwell <nathan@acm.org>
10792 * diagnostic.c (diagnostic_report_current_module): Reroll
10793 included-at loop. Translate text.
10795 2018-08-06 David Malcolm <dmalcolm@redhat.com>
10797 * function-tests.c (selftest::test_expansion_to_rtl): Call
10798 free_after_compilation.
10800 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10802 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
10804 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
10806 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
10807 loops with memory block operations from getting unrolled.
10809 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
10812 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10813 Define to speculation_safe_value_not_needed.
10815 2018-08-06 Jeff Law <law@redhat.com>
10817 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
10820 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10823 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
10824 with all enabled __intN types.
10826 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
10828 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10830 * alias.c (record_set): Check for clobber high.
10831 * cfgexpand.c (expand_gimple_stmt): Likewise.
10832 * combine-stack-adj.c (single_set_for_csa): Likewise.
10833 * combine.c (find_single_use_1): Likewise.
10834 (set_nonzero_bits_and_sign_copies): Likewise.
10835 (get_combine_src_dest): Likewise.
10836 (is_parallel_of_n_reg_sets): Likewise.
10837 (try_combine): Likewise.
10838 (record_dead_and_set_regs_1): Likewise.
10839 (reg_dead_at_p_1): Likewise.
10840 (reg_dead_at_p): Likewise.
10841 * dce.c (deletable_insn_p): Likewise.
10842 (mark_nonreg_stores_1): Likewise.
10843 (mark_nonreg_stores_2): Likewise.
10844 * df-scan.c (df_find_hard_reg_defs): Likewise.
10845 (df_uses_record): Likewise.
10846 (df_get_call_refs): Likewise.
10847 * dwarf2out.c (mem_loc_descriptor): Likewise.
10848 * haifa-sched.c (haifa_classify_rtx): Likewise.
10849 * ira-build.c (create_insn_allocnos): Likewise.
10850 * ira-costs.c (scan_one_insn): Likewise.
10851 * ira.c (equiv_init_movable_p): Likewise.
10852 (rtx_moveable_p): Likewise.
10853 (interesting_dest_for_shprep): Likewise.
10854 * jump.c (mark_jump_label_1): Likewise.
10855 * postreload-gcse.c (record_opr_changes): Likewise.
10856 * postreload.c (reload_cse_simplify): Likewise.
10857 (struct reg_use): Add source expr.
10858 (reload_combine): Check for clobber high.
10859 (reload_combine_note_use): Likewise.
10860 (reload_cse_move2add): Likewise.
10861 (move2add_note_store): Likewise.
10862 * print-rtl.c (print_pattern): Likewise.
10863 * recog.c (decode_asm_operands): Likewise.
10864 (store_data_bypass_p): Likewise.
10865 (if_test_bypass_p): Likewise.
10866 * regcprop.c (kill_clobbered_value): Likewise.
10867 (kill_set_value): Likewise.
10868 * reginfo.c (reg_scan_mark_refs): Likewise.
10869 * reload1.c (maybe_fix_stack_asms): Likewise.
10870 (eliminate_regs_1): Likewise.
10871 (elimination_effects): Likewise.
10872 (mark_not_eliminable): Likewise.
10873 (scan_paradoxical_subregs): Likewise.
10874 (forget_old_reloads_1): Likewise.
10875 * reorg.c (find_end_label): Likewise.
10876 (try_merge_delay_insns): Likewise.
10877 (redundant_insn): Likewise.
10878 (own_thread_p): Likewise.
10879 (fill_simple_delay_slots): Likewise.
10880 (fill_slots_from_thread): Likewise.
10881 (dbr_schedule): Likewise.
10882 * resource.c (update_live_status): Likewise.
10883 (mark_referenced_resources): Likewise.
10884 (mark_set_resources): Likewise.
10885 * rtl.c (copy_rtx): Likewise.
10886 * rtlanal.c (reg_referenced_p): Likewise.
10887 (single_set_2): Likewise.
10888 (noop_move_p): Likewise.
10889 (note_stores): Likewise.
10890 * sched-deps.c (sched_analyze_reg): Likewise.
10891 (sched_analyze_insn): Likewise.
10893 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10895 * cse.c (invalidate_reg): New function extracted from...
10896 (invalidate): ...here.
10897 (canonicalize_insn): Check for clobber high.
10898 (invalidate_from_clobbers): invalidate clobber highs.
10899 (invalidate_from_sets_and_clobbers): Likewise.
10900 (count_reg_usage): Check for clobber high.
10901 (insn_live_p): Likewise.
10902 * cselib.c (cselib_expand_value_rtx_1):Likewise.
10903 (cselib_invalidate_regno): Check for clobber in setter.
10904 (cselib_invalidate_rtx): Pass through setter.
10905 (cselib_invalidate_rtx_note_stores):
10906 (cselib_process_insn): Check for clobber high.
10907 * cselib.h (cselib_invalidate_rtx): Add operand.
10909 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10911 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
10912 (mark_not_eliminable): Likewise.
10913 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
10914 * lra-lives.c (process_bb_lives): Check for clobber high.
10915 * lra.c (new_insn_reg): Remember clobber highs.
10916 (collect_non_operand_hard_regs): Check for clobber high.
10917 (lra_set_insn_recog_data): Likewise.
10918 (add_regs_to_insn_regno_info): Likewise.
10919 (lra_update_insn_regno_info): Likewise.
10921 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10923 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
10924 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
10926 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10928 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
10929 (copy_insn_1): Likewise.
10930 (gen_hard_reg_clobber_high): New gen function.
10931 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
10932 * genemit.c (gen_exp): Likewise.
10933 (gen_emit_seq): Pass through info.
10934 (gen_insn): Check for CLOBBER_HIGH.
10935 (gen_expand): Pass through info.
10936 (gen_split): Likewise.
10937 (output_add_clobbers): Likewise.
10938 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
10939 (remove_clobbers): Likewise.
10940 * rtl.h (gen_hard_reg_clobber_high): New declaration.
10942 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10944 * doc/rtl.texi (clobber_high): Add.
10945 (parallel): Add in clobber high
10946 * rtl.c (rtl_check_failed_code3): Add function.
10947 * rtl.def (CLOBBER_HIGH): Add expression.
10948 * rtl.h (RTL_CHECKC3): Add macro.
10949 (rtl_check_failed_code3): Add declaration.
10950 (XC3EXP): Add macro.
10952 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
10955 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
10956 cfun->machine->max_used_stack_alignment if needed.
10958 2018-08-04 Martin Sebor <msebor@redhat.com>
10960 PR tree-optimization/86571
10961 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
10964 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
10967 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10970 2018-08-03 Jeff Law <law@redhat.com>
10973 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10974 Define to speculation_safe_value_not_needed.
10976 2018-08-03 David Malcolm <dmalcolm@redhat.com>
10978 * doc/gcov.texi (-x): Remove duplicate "to".
10979 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
10980 (-Wif-not-aligned): Remove duplicate "is".
10981 (-flto): Remove duplicate "the".
10982 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
10983 duplicate "v5.00.b".
10984 (MSP430 Options): Remove duplicate "and" from the description
10985 of "-mgprel-sec=regexp".
10986 (x86 Options): Remove duplicate copies of "vmldLog102" and
10987 vmlsLog104 from description of "-mveclibabi=type".
10989 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
10991 * internal-fn.h (first_commutative_argument): Declare.
10992 * internal-fn.c (first_commutative_argument): New function.
10993 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
10994 restrictions for pattern statements. Use first_commutative_argument
10995 to look for commutative operands in calls to internal functions.
10997 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
10999 * Makefile.in (wide-int-range.o): New.
11000 * tree-vrp.c: Move all the wide_int_* functions to...
11001 * wide-int-range.cc: ...here.
11002 * tree-vrp.h: Move all the wide_int_* prototypes to...
11003 * wide-int-range.h: ...here.
11005 2018-08-03 Tom de Vries <tdevries@suse.de>
11007 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
11009 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
11010 * except.c (output_function_exception_table): Do early exit if
11011 targetm_common.except_unwind_info (&global_options) == UI_NONE.
11013 2018-08-03 Martin Liska <mliska@suse.cz>
11015 * predict.c (dump_prediction): Change to 2 digits
11018 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
11020 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
11022 (falkor_am_1_gtov_gtov): ... a new insn reservation.
11024 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
11026 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
11027 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
11028 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
11029 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
11030 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
11032 2018-08-02 David Malcolm <dmalcolm@redhat.com>
11034 * diagnostic-show-locus.c (diagnostic_show_locus): Use
11035 pp_take_prefix when saving the existing prefix.
11036 * diagnostic.c (diagnostic_append_note): Likewise.
11037 * langhooks.c (lhd_print_error_function): Likewise.
11038 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
11039 param's type. Free the existing prefix.
11040 (pp_take_prefix): New function.
11041 (pretty_printer::pretty_printer): Drop the prefix parameter.
11042 Rename the length parameter to match the comment.
11043 (pretty_printer::~pretty_printer): Free the prefix.
11044 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
11046 (struct pretty_printer): Drop the "const" from "prefix" field's
11047 type and clarify memory management.
11048 (pp_set_prefix): Drop the "const" from the 2nd param.
11049 (pp_take_prefix): New decl.
11051 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
11053 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
11054 for word_mode_ok here instead of passing as argument.
11055 (expand_block_compare): Change select_block_compare_mode() call.
11056 (expand_strncmp_gpr_sequence): New function.
11057 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
11059 2018-08-02 Jeff Law <law@redhat.com>
11062 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11063 Define to speculation_safe_value_not_needed.
11066 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11067 Define to speculation_safe_value_not_needed.
11069 2018-08-02 Tom de Vries <tdevries@suse.de>
11072 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
11073 function. Return UI_TARGET unconditionally.
11074 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
11075 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
11077 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
11079 * genemit.c (print_overload_test): Fix typo.
11081 2018-08-02 Richard Biener <rguenther@suse.de>
11083 PR tree-optimization/86816
11084 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
11085 which checks for value availability before querying it.
11086 (gvn_uses_equal): Use it.
11087 (same_succ_hash): Likewise.
11088 (gimple_equal_p): Likewise.
11090 2018-08-02 Nick Clifton <nickc@redhat.com>
11093 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11094 Define to speculation_safe_value_not_needed.
11097 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11098 Define to speculation_safe_value_not_needed.
11101 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11102 Define to speculation_safe_value_not_needed.
11105 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11106 Define to speculation_safe_value_not_needed.
11109 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11110 Define to speculation_safe_value_not_needed.
11113 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11114 Define to speculation_safe_value_not_needed.
11117 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11118 Define to speculation_safe_value_not_needed.
11121 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11122 Define to speculation_safe_value_not_needed.
11125 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
11126 speculation_safe_value_not_needed.
11129 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
11130 to speculation_safe_value_not_needed.
11132 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
11134 * doc/md.texi: Expand the documentation of instruction names
11135 to mention port-local uses. Document '@' in pattern names.
11136 * read-md.h (overloaded_instance, overloaded_name): New structs.
11137 (mapping): Declare.
11138 (md_reader::handle_overloaded_name): New member function.
11139 (md_reader::get_overloads): Likewise.
11140 (md_reader::m_first_overload): New member variable.
11141 (md_reader::m_next_overload_ptr): Likewise.
11142 (md_reader::m_overloads_htab): Likewise.
11143 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
11144 m_next_overload_ptr and m_overloads_htab.
11145 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
11146 (get_mode_token, get_code_token, get_int_token): New functions.
11147 (map_attr_string): Add an optional argument that passes back
11148 the associated iterator.
11149 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
11150 (md_reader::handle_overloaded_name, add_overload_instance): New
11152 (apply_iterators): Handle '@' names. Report an error if '@'
11153 is used without iterators.
11154 (initialize_iterators): Initialize the new iterator_group fields.
11155 * genopinit.c (handle_overloaded_code_for)
11156 (handle_overloaded_gen): New functions.
11157 (main): Use them to print declarations of maybe_code_for_* and
11158 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
11159 * genemit.c (print_overload_arguments, print_overload_test)
11160 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
11161 (main): Use it to print definitions of maybe_code_for_* and
11162 maybe_gen_* functions.
11163 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
11164 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
11165 instead of explicit mode checks.
11166 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
11167 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
11168 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
11169 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
11170 (aarch64_expand_compare_and_swap): Likewise
11171 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
11172 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
11173 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
11174 (aarch64_constant_pool_reload_icode): Delete.
11175 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
11176 instead of aarch64_constant_pool_reload_icode. Use
11177 code_for_aarch64_reload_mov instead of explicit mode checks.
11178 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
11179 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
11180 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
11181 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
11182 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
11183 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
11184 (aarch64_atomic_load_op_code): Delete.
11185 (aarch64_emit_atomic_load_op): Likewise.
11186 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
11187 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
11188 instead of aarch64_emit_atomic_load_op.
11189 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
11190 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
11191 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
11192 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
11193 character before the pattern name.
11194 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
11195 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
11196 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
11197 (aarch64_frecps<mode>): Likewise.
11198 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
11199 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
11200 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
11201 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
11202 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
11204 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
11206 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
11207 Allow HFmode constants if TARGET_FP_F16INST.
11209 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
11212 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
11213 No longer check last store for clobber of address register.
11215 2018-08-02 Martin Liska <mliska@suse.cz>
11217 PR gcov-profile/86817
11218 * gcov.c (process_all_functions): New function.
11220 (process_file): Move functions processing to
11221 process_all_functions.
11223 2018-08-02 David Malcolm <dmalcolm@redhat.com>
11225 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
11226 "const" to the "gimple *" and "rtx_insn *" parameters.
11227 * dumpfile.h (dump_user_location_t::dump_user_location_t):
11229 (dump_location_t::dump_location_t): Likewise.
11231 2018-08-01 Martin Sebor <msebor@redhat.com>
11233 PR tree-optimization/86650
11234 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
11235 rather than a "gcall *". Directly pass the data of interest
11236 to percent_K_format, rather than building a temporary CALL_EXPR
11238 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
11239 (gimple_fold_builtin_strncat): Adjust.
11240 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
11241 gcall* argument with gimple*.
11242 * gimple-ssa-warn-restrict.c (check_call): Same.
11243 (wrestrict_dom_walker::before_dom_children): Same.
11244 (builtin_access::builtin_access): Same.
11245 (check_bounds_or_overlap): Same
11246 (maybe_diag_overlap): Same.
11247 (maybe_diag_offset_bounds): Same.
11248 * tree-diagnostic.c (default_tree_printer): Move usage of
11249 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
11251 * tree-pretty-print.c (percent_K_format): Add argument.
11252 * tree-pretty-print.h: Add argument.
11253 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
11254 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
11255 (maybe_diag_stxncpy_trunc): Same.
11256 (handle_builtin_stxncpy): Same.
11257 (handle_builtin_strcat): Same.
11259 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11261 * match.pd: Optimise pointer range checks.
11263 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11265 PR tree-optimization/86758
11266 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
11267 to remove pattern statements.
11269 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11271 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
11272 result of dfs_enumerate_from when constructing stmt_vec_infos,
11273 instead of additionally calling get_loop_body.
11275 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11277 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
11279 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
11280 When creating an iv, assert that the step is not known to be zero.
11281 (vect_setup_realignment): Update call accordingly.
11282 * tree-vect-stmts.c (vectorizable_store): Likewise.
11283 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
11285 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11287 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
11288 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
11289 (vectorizable_reduction): Likewise.
11290 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
11291 (vect_detect_hybrid_slp_stmts): Likewise.
11292 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
11294 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
11296 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
11297 (wide_int_set_zero_nonzero_bits): ...this.
11298 (zero_nonzero_bits_from_vr): Rename to...
11299 (vrp_set_zero_nonzero_bits): ...this.
11300 (extract_range_from_multiplicative_op_1): Abstract wide int
11302 (wide_int_range_multiplicative_op): ...here.
11303 (extract_range_from_binary_expr_1): Extract wide int binary
11304 operations into their own functions.
11305 (wide_int_range_lshift): New.
11306 (wide_int_range_can_optimize_bit_op): New.
11307 (wide_int_range_shift_undefined_p): New.
11308 (wide_int_range_bit_xor): New.
11309 (wide_int_range_bit_ior): New.
11310 (wide_int_range_bit_and): New.
11311 (wide_int_range_trunc_mod): New.
11312 (extract_range_into_wide_ints): New.
11313 (vrp_shift_undefined_p): New.
11314 (extract_range_from_multiplicative_op): New.
11315 (vrp_can_optimize_bit_op): New.
11316 * tree-vrp.h (value_range::dump): New.
11317 (wide_int_range_multiplicative_op): New.
11318 (wide_int_range_lshift):New.
11319 (wide_int_range_shift_undefined_p): New.
11320 (wide_int_range_bit_xor): New.
11321 (wide_int_range_bit_ior): New.
11322 (wide_int_range_bit_and): New.
11323 (wide_int_range_trunc_mod): New.
11324 (zero_nonzero_bits_from_bounds): Rename to...
11325 (wide_int_set_zero_nonzero_bits): ...this.
11326 (zero_nonzero_bits_from_vr): Rename to...
11327 (vrp_set_zero_nonzero_bits): ...this.
11328 (range_easy_mask_min_max): Rename to...
11329 (wide_int_range_can_optimize_bit_op): this.
11330 * vr-values.c (simplify_bit_ops_using_ranges): Rename
11331 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
11333 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11335 * tree-vectorizer.h (vect_orig_stmt): New function.
11336 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
11337 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
11338 (vect_create_epilog_for_reduction): Likewise.
11339 (vectorizable_live_operation): Likewise.
11340 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
11341 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
11342 * tree-vect-stmts.c (vectorizable_call): Likewise.
11343 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
11345 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11347 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
11349 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
11350 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
11351 (vect_transform_loop): Likewise.
11352 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11354 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11356 * tree-vectorizer.h (vect_schedule_slp): Return void.
11357 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11358 (vect_schedule_slp): Likewise.
11360 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11362 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
11364 (vect_transform_loop): Update calls accordingly. Schedule SLP
11365 instances before the main loop, if any exist.
11367 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11369 PR tree-optimization/86749
11370 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
11371 If the lhs is used in a COND_EXPR, check that it is being used
11372 as the "then" or "else" value.
11374 2018-08-01 Tom de Vries <tdevries@suse.de>
11377 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
11378 speculation_safe_value_not_needed.
11380 2018-08-01 Richard Biener <rguenther@suse.de>
11382 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
11383 as base and offset.
11385 2018-08-01 Martin Liska <mliska@suse.cz>
11387 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
11388 format how successful transformation is dumped.
11389 (gimple_mod_pow2_value_transform): Likewise.
11390 (gimple_mod_subtract_transform): Likewise.
11391 (gimple_stringops_transform): Likewise.
11393 2018-08-01 Martin Liska <mliska@suse.cz>
11395 PR value-prof/35543
11396 * value-prof.c (interesting_stringop_to_profile_p):
11397 Simplify the code and add BUILT_IN_MEMMOVE.
11398 (gimple_stringops_transform): Likewise.
11400 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
11402 * config/aarch64/aarch64-simd.md
11403 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
11404 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
11405 use GPI iterator instead of SI mode.
11407 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
11409 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
11410 rs6000_speculation_barrier.
11411 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
11412 new barrier pattern name.
11414 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
11416 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
11417 (speculation_barrier): New insn.
11419 2018-08-01 Richard Biener <rguenther@suse.de>
11422 * graphite.h: Include isl/id.h and isl/space.h to allow build
11425 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
11428 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
11429 mode for COFF targets.
11430 * defaults.h (TARGET_COFF): Define.
11431 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
11432 TARGET_COFF): Define.
11433 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
11434 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
11435 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
11437 2018-07-31 Alexandre Oliva <oliva@adacore.com>
11439 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
11440 * gimple-streamer-out.c (output_bb): Save it.
11441 * lto-streamer-in.c (input_struct_function_base): Restore
11442 instance discriminator if available. Create map on demand.
11443 * lto-streamer-out.c (output_struct_function_base): Save it if
11445 * final.c (decl_to_instance_map): Document LTO strategy.
11447 2018-07-31 Alexandre Oliva <oliva@adacore.com>
11448 Olivier Hainque <hainque@adacore.com>
11450 * debug.h (decl_to_instance_map_t): New type.
11451 (decl_to_instance_map): Declare.
11452 (maybe_create_decl_to_instance_map): New inline function.
11453 * final.c (bb_discriminator, last_bb_discriminator): New statics,
11454 to track basic block discriminators.
11455 (final_start_function_1): Initialize them.
11456 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
11458 (decl_to_instance_map): New variable.
11459 (map_decl_to_instance, maybe_set_discriminator): New functions.
11460 (notice_source_line): Set discriminator.
11462 2018-07-31 Ian Lance Taylor <iant@golang.org>
11464 * targhooks.c (default_have_speculation_safe_value): Add
11467 2018-07-31 David Malcolm <dmalcolm@redhat.com>
11469 * dump-context.h: Include "pretty-print.h".
11470 (dump_context::refresh_dumps_are_enabled): New decl.
11471 (dump_context::emit_item): New decl.
11472 (class dump_context): Add fields "m_test_pp" and
11474 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
11475 (temp_dump_context::get_dumped_text): New decl.
11476 (class temp_dump_context): Add field "m_pp".
11477 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
11478 (dump_context::refresh_dumps_are_enabled): ...and add a test for
11480 (set_dump_file): Update for above change.
11481 (set_alt_dump_file): Likewise.
11482 (dump_loc): New overload, taking a pretty_printer *.
11483 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
11484 to any test pretty-printer.
11485 (make_item_for_dump_gimple_stmt): New function, adapted from
11486 optinfo::add_gimple_stmt.
11487 (dump_context::dump_gimple_stmt): Call it, and use the result,
11488 eliminating the direct usage of dump_file and alt_dump_file in
11489 favor of indirectly using them via emit_item.
11490 (make_item_for_dump_gimple_expr): New function, adapted from
11491 optinfo::add_gimple_expr.
11492 (dump_context::dump_gimple_expr): Call it, and use the result,
11493 eliminating the direct usage of dump_file and alt_dump_file in
11494 favor of indirectly using them via emit_item.
11495 (make_item_for_dump_generic_expr): New function, adapted from
11497 (dump_context::dump_generic_expr): Call it, and use the result,
11498 eliminating the direct usage of dump_file and alt_dump_file in
11499 favor of indirectly using them via emit_item.
11500 (make_item_for_dump_printf_va): New function, adapted from
11501 optinfo::add_printf_va.
11502 (make_item_for_dump_printf): New function.
11503 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
11504 and use the result, eliminating the direct usage of dump_file and
11505 alt_dump_file in favor of indirectly using them via emit_item.
11506 (make_item_for_dump_dec): New function.
11507 (dump_context::dump_dec): Call it, and use the result,
11508 eliminating the direct usage of dump_file and alt_dump_file in
11509 favor of indirectly using them via emit_item.
11510 (make_item_for_dump_symtab_node): New function, adapted from
11511 optinfo::add_symtab_node.
11512 (dump_context::dump_symtab_node): Call it, and use the result,
11513 eliminating the direct usage of dump_file and alt_dump_file in
11514 favor of indirectly using them via emit_item.
11515 (dump_context::begin_scope): Reimplement, avoiding direct usage
11516 of dump_file and alt_dump_file in favor of indirectly using them
11518 (dump_context::emit_item): New member function.
11519 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
11520 Set up test pretty-printer on the underlying context. Call
11521 refresh_dumps_are_enabled.
11522 (temp_dump_context::~temp_dump_context): Call
11523 refresh_dumps_are_enabled.
11524 (temp_dump_context::get_dumped_text): New member function.
11525 (selftest::verify_dumped_text): New function.
11526 (ASSERT_DUMPED_TEXT_EQ): New macro.
11527 (selftest::test_capture_of_dump_calls): Run all tests twice, with
11528 and then without optinfo enabled. Add uses of
11529 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
11530 * dumpfile.h: Update comment for the dump_* API.
11531 * optinfo-emit-json.cc
11532 (selftest::test_building_json_from_dump_calls): Update for new
11533 param for temp_dump_context ctor.
11534 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
11535 and "m_owned" field.
11536 (optinfo_item::~optinfo_item): Likewise.
11537 (optinfo::add_item): New member function.
11538 (optinfo::emit): Update comment.
11539 (optinfo::add_string): Delete.
11540 (optinfo::add_printf): Delete.
11541 (optinfo::add_printf_va): Delete.
11542 (optinfo::add_gimple_stmt): Delete.
11543 (optinfo::add_gimple_expr): Delete.
11544 (optinfo::add_tree): Delete.
11545 (optinfo::add_symtab_node): Delete.
11546 (optinfo::add_dec): Delete.
11547 * optinfo.h (class dump_context): New forward decl.
11548 (optinfo::add_item): New decl.
11549 (optinfo::add_string): Delete.
11550 (optinfo::add_printf): Delete.
11551 (optinfo::add_printf_va): Delete.
11552 (optinfo::add_gimple_stmt): Delete.
11553 (optinfo::add_gimple_expr): Delete.
11554 (optinfo::add_tree): Delete.
11555 (optinfo::add_symtab_node): Delete.
11556 (optinfo::add_dec): Delete.
11557 (optinfo::add_poly_int): Delete.
11558 (optinfo_item::optinfo_item): Remove "owned" param.
11559 (class optinfo_item): Remove field "m_owned".
11561 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11563 PR middle-end/86705
11564 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
11565 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
11566 requested variable alignment.
11567 (expand_one_ssa_partition): Likewise.
11568 (expand_one_var): Likewise.
11570 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
11572 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
11573 to speculation_safe_value_not_needed.
11575 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
11577 * targhooks.h (speculation_safe_value_not_needed): New prototype.
11578 * targhooks.c (speculation_safe_value_not_needed): New function.
11579 * target.def (have_speculation_safe_value): Update documentation.
11580 * doc/tm.texi: Regenerated.
11582 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
11584 * config/aarch64/iterators.md (ALLI_TI): New iterator.
11585 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
11587 (despeculate_copy<ALLI:mode>_insn): New insn.
11588 (despeculate_copyti_insn): New insn.
11589 (despeculate_simple<ALLI:mode>): New insn
11590 (despeculate_simpleti): New insn.
11591 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
11593 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
11594 aarch64_speculation_safe_value.
11595 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
11597 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
11599 * config/aarch64/aarch64-speculation.cc: New file.
11600 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
11601 before pass_reorder_blocks.
11602 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
11604 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
11605 X14 and X15 when tracking speculation.
11606 * config/aarch64/aarch64.md (register name constants): Add
11607 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
11608 (unspec): Add UNSPEC_SPECULATION_TRACKER.
11609 (speculation_barrier): New insn attribute.
11610 (cmp<mode>): Allow SP in comparisons.
11611 (speculation_tracker): New insn.
11612 (speculation_barrier): Add speculation_barrier attribute.
11613 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
11614 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
11615 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
11617 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
11619 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
11620 aarch64_track_speculation is true.
11621 (tb<optab><mode>1): Likewise.
11622 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
11623 generate CB[N]Z when tracking speculation.
11624 (aarch64_split_compare_and_swap): Likewise.
11625 (aarch64_split_atomic_op): Likewise.
11627 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
11629 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
11631 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
11633 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
11634 (speculation_barrier): New insn.
11636 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
11638 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
11639 * config/arm/arm.md (speculation_barrier): New expand.
11640 (speculation_barrier_insn): New pattern.
11642 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
11644 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
11645 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
11646 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
11647 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
11649 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
11650 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
11651 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
11652 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
11653 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
11654 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
11655 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
11656 * builtins.c (expand_speculation_safe_value): New function.
11657 (expand_builtin): Call it.
11658 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
11659 * doc/extend.texi: Document __builtin_speculation_safe_value.
11660 * doc/md.texi: Document "speculation_barrier" pattern.
11661 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
11662 TARGET_HAVE_SPECULATION_SAFE_VALUE.
11663 * doc/tm.texi: Regenerated.
11664 * target.def (have_speculation_safe_value, speculation_safe_value): New
11666 * targhooks.c (default_have_speculation_safe_value): New function.
11667 (default_speculation_safe_value): New function.
11668 * targhooks.h (default_have_speculation_safe_value): Add prototype.
11669 (default_speculation_safe_value): Add prototype.
11671 2018-07-31 David Malcolm <dmalcolm@redhat.com>
11673 * dump-context.h (dump_context::dump_loc): New decl.
11674 * dumpfile.c (dump_context::dump_loc): New member function.
11675 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
11676 and dump_gimple_stmt.
11677 (dump_context::dump_gimple_expr_loc): Likewise, using
11679 (dump_context::dump_generic_expr_loc): Likewise, using
11681 (dump_context::dump_printf_loc_va): Likewise, using
11683 (dump_context::begin_scope): Explicitly using the global function
11684 "dump_loc", rather than the member function.
11686 2018-07-31 Martin Sebor <msebor@redhat.com>
11688 PR tree-optimization/86741
11689 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
11691 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
11693 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
11694 depend on whether prefetch instructions will be emitted or not.
11695 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
11696 will be emitted or not.
11697 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
11698 (TARGET_SETMEM_PFD): New macros.
11700 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11702 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
11703 (NULL_STMT_VEC_INFO): Delete.
11704 (stmt_vec_info::operator*): Likewise.
11705 (stmt_vec_info::operator gimple *): Likewise.
11706 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
11707 of NULL_STMT_VEC_INFO.
11708 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
11709 (vect_reassociating_reduction_p): Likewise.
11710 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
11711 (vectorizable_store): Likewise.
11712 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
11713 (vec_info::free_stmt_vec_infos): Likewise.
11715 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11717 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
11718 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
11720 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11722 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
11723 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
11724 (vec_info::free_stmt_vec_info): New private member functions.
11725 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
11726 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
11727 * tree-parloops.c (gather_scalar_reductions): Remove calls to
11728 set_stmt_vec_info_vec and free_stmt_vec_infos.
11729 * tree-vect-loop.c (_loop_vec_info): Remove call to
11730 set_stmt_vec_info_vec.
11731 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
11732 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
11733 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
11734 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
11735 (vec_info::free_stmt_vec_info): ...these new functions. Remove
11736 assignments in {vec_info::,}new_stmt_vec_info that are redundant
11737 with the clearing in the xcalloc.
11738 (stmt_vec_info_vec): Delete.
11739 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
11740 (vectorize_loops): Likewise.
11741 (vec_info::~vec_info): Remove argument from call to
11742 free_stmt_vec_infos.
11743 (vec_info::add_stmt): Remove vinfo argument from call to
11746 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11748 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
11749 rather than a gimple stmt.
11750 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
11751 information for pattern statements when passed the original
11752 statement; instead wait to be passed the pattern statement itself.
11753 Don't call set_vinfo_for_stmt here.
11754 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
11755 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
11756 stmt_vec_infos here.
11757 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
11758 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
11759 stmt_vec_infos entry.
11761 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11763 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
11764 * tree-vectorizer.c (vec_info::replace_stmt): New function.
11765 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
11766 * tree-vect-stmts.c (vectorizable_call): Likewise.
11767 (vectorizable_simd_clone_call): Likewise.
11769 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11771 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
11772 * tree-vectorizer.c (vec_info::remove_stmt): New function.
11773 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
11774 * tree-vect-loop.c (vect_transform_loop): Likewise.
11775 * tree-vect-slp.c (vect_schedule_slp): Likewise.
11776 * tree-vect-stmts.c (vect_remove_stores): Likewise.
11778 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11780 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
11781 (vect_dr_stmt): Delete.
11782 * tree-vectorizer.c (vec_info::lookup_dr): New function.
11783 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
11785 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
11786 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
11787 (vect_verify_datarefs_alignment, vect_peeling_supportable)
11788 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
11789 (vect_analyze_data_refs): Likewise.
11790 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
11792 (vect_find_same_alignment_drs): Likewise.
11793 (vect_slp_analyze_node_dependences): Update calls accordingly.
11794 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
11795 instead of DR_VECT_AUX.
11796 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
11797 of a vector data references. Use vec_info::lookup_dr instead of
11799 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
11800 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
11801 instead of DR_VECT_AUX.
11803 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11805 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
11807 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
11809 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
11810 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
11811 (vect_gen_prolog_loop_niters): Likewise.
11813 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11815 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
11816 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
11817 (vect_known_alignment_in_bytes, vect_dr_behavior)
11818 (vect_get_scalar_dr_size): Take references as dr_vec_infos
11819 instead of data_references. Update calls to other routines for
11820 which the same change has been made.
11821 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
11822 dr_vec_infos instead of stmt_vec_infos.
11823 (vect_analyze_data_ref_dependence): Update call accordingly.
11824 (vect_slp_analyze_data_ref_dependence)
11825 (vect_record_base_alignments): Use DR_VECT_AUX.
11826 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
11827 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
11828 (vector_alignment_reachable_p, vect_get_data_access_cost)
11829 (vect_peeling_supportable, vect_analyze_group_access_1)
11830 (vect_analyze_group_access, vect_analyze_data_ref_access)
11831 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
11832 (vect_compile_time_alias, vect_small_gap_p)
11833 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
11834 (vect_supportable_dr_alignment): Take references as dr_vec_infos
11835 instead of data_references. Update calls to other routines for
11836 which the same change has been made.
11837 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
11838 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
11839 (vect_slp_analyze_and_verify_node_alignment)
11840 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
11841 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
11842 (vect_setup_realignment): Use dr_vec_infos. Update calls after
11844 (_vect_peel_info::dr): Replace with...
11845 (_vect_peel_info::dr_info): ...this new field.
11846 (vect_peeling_hash_get_most_frequent)
11847 (vect_peeling_hash_choose_best_peeling): Update accordingly.
11848 (vect_peeling_hash_get_lowest_cost):
11849 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
11850 routines for which the same change has been made.
11851 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
11853 * tree-vect-loop-manip.c (get_misalign_in_elems)
11854 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
11856 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
11857 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
11858 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
11859 (get_group_load_store_type, get_negative_load_store_type)
11860 (vect_get_data_ptr_increment, vectorizable_store)
11861 (vectorizable_load): Likewise.
11862 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
11863 Update calls to other routines for which the same change has been made.
11865 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11867 * tree-vectorizer.h (vec_info::move_dr): New member function.
11868 (dataref_aux): Rename to...
11869 (dr_vec_info): ...this and add "dr" and "stmt" fields.
11870 (_stmt_vec_info::dr_aux): Update accordingly.
11871 (_stmt_vec_info::data_ref_info): Delete.
11872 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
11873 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
11874 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
11875 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
11877 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
11878 (STMT_VINFO_DR_INFO): New macro.
11879 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
11880 (set_dr_misalignment): Update after rename of dataref_aux.
11881 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
11882 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
11883 initialization of STMT_VINFO_DATA_REF.
11884 * tree-vectorizer.c (vec_info::move_dr): New function.
11885 * tree-vect-patterns.c (vect_recog_bool_pattern)
11886 (vect_recog_mask_conversion_pattern)
11887 (vect_recog_gather_scatter_pattern): Use it.
11888 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
11889 the "dr" and "stmt" fields of dr_vec_info instead of
11890 STMT_VINFO_DATA_REF.
11892 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11894 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
11895 (is_pattern_stmt_p): Use it.
11896 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
11897 on pattern statements.
11899 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11901 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
11902 original stmt as a stmt_vec_info rather than a gimple stmt.
11903 (vect_pattern_recog_1): Take the statement directly as a
11904 stmt_vec_info, rather than via a gimple_stmt_iterator.
11905 Update call to vect_mark_pattern_stmts.
11906 (vect_pattern_recog): Update calls accordingly.
11908 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11910 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
11911 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
11912 a vect_def_type for the first argument.
11913 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
11914 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
11915 operand if it isn't defined by a vectorized statement.
11916 (vect_build_gather_load_calls): Remove the mask_dt argument and
11917 update calls to vect_get_vec_def_for_stmt_copy.
11918 (vectorizable_bswap): Likewise the dt argument.
11919 (vectorizable_call): Update calls to vectorizable_bswap and
11920 vect_get_vec_def_for_stmt_copy.
11921 (vectorizable_simd_clone_call, vectorizable_assignment)
11922 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
11923 (vectorizable_comparison): Update calls to
11924 vect_get_vec_def_for_stmt_copy.
11925 (vectorizable_store): Likewise. Remove now-unnecessary calls to
11926 vect_is_simple_use.
11927 (vect_get_loop_based_defs): Remove dt argument and update call
11928 to vect_get_vec_def_for_stmt_copy.
11929 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
11930 and vect_get_vec_def_for_stmt_copy.
11931 (vectorizable_load): Update calls to vect_build_gather_load_calls
11932 and vect_get_vec_def_for_stmt_copy.
11933 * tree-vect-loop.c (vect_create_epilog_for_reduction)
11934 (vectorizable_reduction, vectorizable_live_operation): Update calls
11935 to vect_get_vec_def_for_stmt_copy.
11937 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11939 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
11940 and gimple stmt arguments with a stmt_vec_info.
11941 (vect_record_base_alignments): Update calls accordingly.
11942 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
11943 and gimple stmt arguments with a stmt_vec_info.
11944 (vect_build_slp_tree_1): Remove vinfo argument and update call
11945 to vect_record_max_nunits.
11946 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
11947 and vect_record_max_nunits.
11949 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11951 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
11952 file and take a stmt_vec_info instead of a gimple stmt.
11953 (supportable_widening_operation, vect_finish_replace_stmt)
11954 (vect_finish_stmt_generation, vect_get_store_rhs)
11955 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
11956 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
11957 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
11958 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
11959 (vect_create_data_ref_ptr, bump_vector_ptr)
11960 (vect_permute_store_chain, vect_setup_realignment)
11961 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
11962 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
11963 (vectorizable_reduction, vectorizable_induction)
11964 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
11965 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
11966 than gimple stmts as arguments.
11967 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
11968 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
11969 (can_group_stmts_p, vect_check_gather_scatter)
11970 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
11971 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
11972 (vect_permute_load_chain, vect_shift_permute_load_chain)
11973 (vect_transform_grouped_load)
11974 (vect_record_grouped_load_vectors): Likewise.
11975 * tree-vect-loop.c (vect_fixup_reduc_chain)
11976 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
11977 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
11978 (vectorizable_reduction, vectorizable_induction)
11979 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
11980 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
11981 (vect_get_load_store_mask): Likewise.
11982 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
11983 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
11984 * tree-vect-stmts.c (vect_mark_relevant)
11985 (is_simple_and_all_uses_invariant)
11986 (exist_non_indexing_operands_for_use_p, process_use)
11987 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
11988 (vect_get_vec_def_for_operand, vect_get_vec_defs)
11989 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
11990 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
11991 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
11992 (get_negative_load_store_type, get_load_store_type)
11993 (vect_check_load_store_mask, vect_check_store_rhs)
11994 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
11995 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
11996 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
11997 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
11998 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
11999 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
12000 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
12001 (supportable_widening_operation): Likewise.
12003 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12005 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
12006 a stmt_vec_info instead of a gcall.
12007 (vect_check_gather_scatter): Update call accordingly.
12008 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
12010 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
12011 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
12012 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
12013 instead of a gimple stmt.
12014 (vect_transform_loop): Update calls accordingly.
12015 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
12016 stmt_vec_infos instead of gimple stmts.
12017 (vect_analyze_slp_instance): Update use accordingly.
12018 * tree-vect-stmts.c (read_vector_array, write_vector_array)
12019 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
12020 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
12021 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
12022 (vect_gen_widened_results_half, vect_get_loop_based_defs)
12023 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
12024 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
12025 down to subroutines.
12027 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12029 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
12030 of the worklist from a vector of gimple stmts to a vector of
12032 * tree-vect-stmts.c (vect_mark_relevant, process_use)
12033 (vect_mark_stmts_to_be_vectorized): Likewise
12035 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12037 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
12038 statement before passing it to vect_analyze_stmt.
12039 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
12040 the chain of phi vector definitions. Track the exit phi via its
12042 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
12043 STMT_VINFO_REDUC_DEF.
12044 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
12045 stmt_vec_infos to handle the statement chains.
12046 (vect_get_slp_defs): Record the first statement in the node
12047 using a stmt_vec_info.
12048 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
12049 statements here and pass their stmt_vec_info down to subroutines.
12050 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
12051 down to vect_finish_stmt_generation.
12052 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
12053 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
12054 stmt_vec_infos to subroutines.
12055 (vect_remove_stores): Use stmt_vec_infos to handle the statement
12058 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12060 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
12061 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
12062 (vect_permute_store_chain, vect_setup_realignment)
12063 (vect_permute_load_chain, vect_shift_permute_load_chain)
12064 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
12065 stmts internally, and when passing values to other vectorizer routines.
12066 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
12067 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
12068 (vect_analyze_loop_operations, get_initial_def_for_reduction)
12069 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
12070 (vectorizable_reduction, vectorizable_induction)
12071 (vectorizable_live_operation, vect_transform_loop_stmt)
12072 (vect_transform_loop): Likewise.
12073 * tree-vect-patterns.c (vect_reassociating_reduction_p)
12074 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
12075 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
12076 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
12077 (vect_slp_analyze_node_operations_1): Likewise.
12078 * tree-vect-stmts.c (vect_mark_relevant, process_use)
12079 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
12080 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
12081 (vect_finish_stmt_generation_1, get_group_load_store_type)
12082 (get_load_store_type, vect_build_gather_load_calls)
12083 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
12084 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
12085 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
12086 (vectorizable_store, vectorizable_load, vectorizable_condition)
12087 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
12088 (supportable_widening_operation): Likewise.
12089 (vect_get_vector_types_for_stmt): Likewise.
12090 * tree-vectorizer.h (vect_dr_behavior): Likewise.
12092 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12094 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
12095 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
12096 (vect_permute_store_chain, vect_permute_load_chain)
12097 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
12098 repeated stmt_vec_info lookups.
12099 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
12100 (vect_update_ivs_after_vectorizer): Likewise.
12101 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
12102 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
12103 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
12104 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
12105 (vect_bb_slp_scalar_cost): Likewise.
12106 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
12108 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12110 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
12111 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
12112 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
12113 to get gassigns and gcalls, rather than operating on generc gimple
12115 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
12116 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
12117 (vectorizable_load, vect_analyze_stmt): Likewise.
12118 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
12120 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12122 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
12123 return stmt_vec_infos rather than gimple stmts. Do not accept
12125 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
12127 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
12128 Update use of get_later_stmt.
12129 (vect_get_constant_vectors): Update call accordingly.
12130 (vect_schedule_slp_instance): Likewise
12131 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
12132 (vect_slp_analyze_instance_dependence): Likewise.
12133 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
12135 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12137 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
12138 (stmt_info_for_cost::stmt_info): ...this new field.
12139 (add_stmt_costs): Update accordingly.
12140 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
12141 (vect_get_known_peeling_cost): Likewise.
12142 (vect_estimate_min_profitable_iters): Likewise.
12143 * tree-vect-stmts.c (record_stmt_cost): Likewise.
12145 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12147 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
12148 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
12149 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
12151 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
12153 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12155 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
12156 a gimple stmt to a stmt_vec_info.
12157 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
12159 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12161 * tree-vectorizer.h (vec_info::grouped_stores): Change from
12162 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
12163 (_loop_vec_info::reduction_chains): Likewise.
12164 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
12166 * tree-vect-slp.c (vect_analyze_slp): Likewise.
12168 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12170 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
12171 a gimple stmt to a stmt_vec_info.
12172 (_stmt_vec_info::next_element): Likewise.
12173 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
12174 (vect_slp_analyze_and_verify_node_alignment)
12175 (vect_analyze_group_access_1, vect_analyze_group_access)
12176 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
12177 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
12178 (vect_supportable_dr_alignment): Update accordingly.
12179 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
12180 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
12181 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
12182 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
12183 * tree-vect-slp.c (vect_build_slp_tree_1)
12184 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
12185 (vect_split_slp_store_group, vect_analyze_slp_instance)
12186 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
12187 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
12188 (get_group_load_store_type, get_load_store_type)
12189 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
12190 (vect_transform_stmt, vect_remove_stores): Likewise.
12192 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12194 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
12195 than a gimple stmt.
12196 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
12197 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
12198 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
12199 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
12200 (vector_alignment_reachable_p, vect_get_data_access_cost)
12201 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
12202 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
12203 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
12204 (vect_analyze_group_access_1, vect_analyze_group_access)
12205 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
12206 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
12207 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
12208 result of vect_dr_stmt and use the stmt_vec_info instead of
12209 the associated gimple stmt.
12210 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
12211 (vect_gen_prolog_loop_niters): Likewise.
12212 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
12214 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12216 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
12217 to a vec<stmt_vec_info>.
12218 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
12219 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
12220 vec<stmt_vec_info>.
12221 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
12222 to a vec<stmt_vec_info>.
12223 (bst_traits::value_type, bst_traits::value_type): Likewise.
12224 (bst_traits::hash): Update accordingly.
12225 (vect_get_and_check_slp_defs): Change the stmts parameter from
12226 a vec<gimple *> to a vec<stmt_vec_info>.
12227 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
12228 (vect_build_slp_tree): Likewise.
12229 (vect_build_slp_tree_2): Likewise. Update uses of
12230 SLP_TREE_SCALAR_STMTS.
12231 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
12232 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
12233 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
12234 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
12235 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
12236 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
12237 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
12238 (vect_get_constant_vectors, vect_get_slp_defs)
12239 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
12240 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
12241 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
12242 instead of gimple stmts.
12243 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
12244 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
12245 (vect_slp_analyze_instance_dependence): Update uses of
12246 SLP_TREE_SCALAR_STMTS.
12247 (vect_slp_analyze_and_verify_node_alignment): Likewise.
12248 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
12249 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
12250 (get_initial_defs_for_reduction): Likewise.
12251 (vect_create_epilog_for_reduction): Likewise.
12252 (vectorize_fold_left_reduction): Likewise.
12253 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
12254 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
12255 (can_vectorize_live_stmts): Likewise.
12257 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12259 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
12260 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
12261 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
12263 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
12265 (gather_scalar_reductions): Update after above interface changes.
12266 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
12267 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
12269 (vect_force_simple_reduction): Likewise.
12270 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
12271 LOOP_VINFO_REDUCTIONS.
12272 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
12274 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12276 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
12277 a gimple stmt to a stmt_vec_info.
12278 * tree-vect-loop.c (vect_active_double_reduction_p)
12279 (vect_force_simple_reduction, vectorizable_reduction): Update
12282 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12284 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
12285 vec<gimple *> to a vec<stmt_vec_info>.
12286 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
12287 the reduction_phis argument from a vec<gimple *> to a
12288 vec<stmt_vec_info>.
12289 (vectorizable_reduction): Likewise the phis local variable that
12290 is passed to vect_create_epilog_for_reduction. Update for new type
12291 of SLP_TREE_VEC_STMTS.
12292 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
12293 (vectorizable_live_operation): Likewise.
12294 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
12295 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
12297 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12299 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
12300 a gimple stmt to a stmt_vec_info.
12301 (vectorizable_condition, vectorizable_live_operation)
12302 (vectorizable_reduction, vectorizable_induction): Pass back the
12303 vectorized statement as a stmt_vec_info.
12304 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
12305 use of STMT_VINFO_VEC_STMT.
12306 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
12307 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
12308 as stmt_vec_infos rather than gimple stmts.
12309 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
12310 to a stmt_vec_info.
12311 (vectorizable_live_operation): Likewise.
12312 (vectorizable_reduction, vectorizable_induction): Likewise,
12313 updating use of STMT_VINFO_VEC_STMT.
12314 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
12315 of STMT_VINFO_VEC_STMT.
12316 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
12317 (vectorizable_simd_clone_call, vectorizable_conversion)
12318 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
12319 (vectorizable_store, vectorizable_load, vectorizable_condition)
12320 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
12321 from a gimple stmt to a stmt_vec_info.
12322 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
12323 pointer to a stmt_vec_info to the vectorizable_* routines.
12325 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12327 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
12328 a gimple stmt to a stmt_vec_info.
12329 (is_pattern_stmt_p): Update accordingly.
12330 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
12331 (vect_record_grouped_load_vectors): Likewise.
12332 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
12333 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
12334 (vect_model_reduction_cost): Likewise.
12335 (vect_create_epilog_for_reduction): Likewise.
12336 (vectorizable_reduction, vectorizable_induction): Likewise.
12337 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
12338 Return the stmt_vec_info for the pattern statement.
12339 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
12340 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
12341 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
12342 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
12343 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
12344 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
12345 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
12346 (free_stmt_vec_info, vect_is_simple_use): Likewise.
12348 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12350 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
12351 (vect_finish_stmt_generation): Likewise.
12352 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
12353 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
12354 (vect_build_gather_load_calls): Use the return value of the above
12355 functions instead of a separate call to vinfo_for_stmt. Use narrow
12356 scopes for the input gimple stmt and wider scopes for the associated
12357 stmt_vec_info. Use vec_info::lookup_def when setting these
12358 stmt_vec_infos from an SSA_NAME definition.
12359 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
12360 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
12361 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
12362 (vectorizable_store, vectorizable_load, vectorizable_condition)
12363 (vectorizable_comparison): Likewise.
12364 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
12365 (vectorizable_reduction): Likewise.
12367 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12369 * tree-vectorizer.h (vect_is_simple_use): Add an optional
12370 stmt_vec_info * parameter before the optional gimple **.
12371 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
12372 (process_use, vect_get_vec_def_for_operand_1): Update callers.
12373 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
12374 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12375 (vectorizable_live_operation): Likewise.
12376 * tree-vect-patterns.c (type_conversion_p): Likewise.
12377 (vect_look_through_possible_promotion): Likewise.
12378 (vect_recog_rotate_pattern): Likewise.
12379 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
12381 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12383 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
12384 a typedef to a wrapper class.
12385 (NULL_STMT_VEC_INFO): New macro.
12386 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
12387 (stmt_vec_info::operator*): New function.
12388 (stmt_vec_info::operator gimple *): Likewise.
12389 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
12390 (add_stmt_costs): Likewise.
12391 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
12392 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
12393 (vect_get_known_peeling_cost): Likewise.
12394 (vect_estimate_min_profitable_iters): Likewise.
12395 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
12396 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
12397 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
12398 (vectorizable_store, free_stmt_vec_infos): Likewise.
12399 (new_stmt_vec_info): Change return type of xcalloc to
12402 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12404 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
12405 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
12406 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
12407 a single_imm_use-based sequence.
12408 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
12410 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12412 * tree-vectorizer.h (vec_info::lookup_def): Declare.
12413 * tree-vectorizer.c (vec_info::lookup_def): New function.
12414 * tree-vect-patterns.c (vect_get_internal_def): Use it.
12415 (vect_widened_op_tree): Likewise.
12416 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
12417 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
12418 (vectorizable_reduction): Likewise.
12419 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
12421 (vect_is_slp_reduction): Update calls accordingly. Use
12422 vec_info::lookup_def.
12423 (vect_is_simple_reduction): Likewise
12424 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
12426 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12428 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
12429 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
12430 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
12432 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
12433 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
12434 (vect_update_vf_for_slp, vect_analyze_loop_operations)
12435 (vect_is_slp_reduction, vectorizable_induction)
12436 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
12437 * tree-vect-patterns.c (vect_init_pattern_stmt):
12438 (vect_determine_min_output_precision_1, vect_determine_precisions)
12439 (vect_pattern_recog): Likewise.
12440 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
12441 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
12442 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
12443 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
12444 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
12445 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
12446 info field from a loop to a loop_vec_info.
12448 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12450 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
12451 (vec_info::add_stmt): Declare.
12452 * tree-vectorizer.c (vec_info::add_stmt): New function.
12453 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
12454 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
12455 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
12456 (vectorizable_induction): Likewise.
12457 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
12458 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
12459 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
12460 (vectorizable_load): Likewise.
12461 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
12462 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
12463 (vect_recog_gather_scatter_pattern): Likewise.
12464 (append_pattern_def_seq): Likewise. Remove a check that is
12465 performed by add_stmt itself.
12467 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12469 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
12470 which make_ssa_name was called with new_stmt before new_stmt
12473 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12475 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
12477 (vect_is_slp_reduction): ...here...
12478 (vect_is_simple_reduction): ...and here. Remove repetition of tests
12479 that are already known to be false.
12481 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12483 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
12484 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
12485 STMT_VINFO_NUM_SLP_USES when it's true.
12486 (vect_free_slp_instance): Add a final_p parameter and pass it to
12487 vect_free_slp_tree.
12488 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
12489 (vect_analyze_slp_instance): Likewise.
12490 (vect_slp_analyze_operations): Likewise.
12491 (vect_slp_analyze_bb_1): Likewise.
12492 * tree-vectorizer.c (vec_info): Likewise.
12493 * tree-vect-loop.c (vect_transform_loop): Likewise.
12495 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12497 * tree-vect-loop.c (vectorizable_reduction): Assert that the
12498 function is not called for second and subsequent members of
12501 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12503 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
12504 cases for nested loops from here to ...
12505 (vect_create_epilog_for_reduction): ...here. Only call
12506 vect_is_simple_use for inner-loop reductions.
12508 2018-07-31 Martin Liska <mliska@suse.cz>
12510 PR gcov-profile/85338
12511 PR gcov-profile/85350
12512 PR gcov-profile/85372
12513 * profile.c (struct location_triplet): New.
12514 (struct location_triplet_hash): Likewise.
12515 (output_location): Do not output a BB that
12516 is already recorded for a line.
12517 (branch_prob): Use streamed_locations.
12519 2018-07-31 Martin Liska <mliska@suse.cz>
12521 PR gcov-profile/85370
12522 * coverage.c (coverage_begin_function): Do not mark target
12523 clones as artificial functions.
12525 2018-07-31 Martin Liska <mliska@suse.cz>
12527 PR gcov-profile/83813
12528 PR gcov-profile/84758
12529 PR gcov-profile/85217
12530 PR gcov-profile/85332
12531 * profile.c (branch_prob): Do not record GOTO expressions
12532 for GIMPLE statements which locations are already streamed.
12534 2018-07-31 Olivier Hainque <hainque@adacore.com>
12536 * gcc.c (handle_spec_function): Accept a soft_matched_part
12537 argument, as do_spec_1. Pass it down to ...
12538 (eval_spec_function): Accept a soft_matched_part argument,
12539 and pass it down to ...
12540 (do_spec_2): Accept a soft_matched_part argument, and pass
12541 it down to do_spec_1.
12542 (do_spec_1): Pass soft_matched_part to handle_spec_function.
12543 (handle_braces): Update call to handle_spec_function.
12544 (driver::set_up_specs): Update calls to do_spec_2.
12545 (compare_debug_dump_opt_spec_function): Likewise.
12546 (compare_debug_self_opt_spec_function): Likewise.
12548 2018-07-31 Olivier Hainque <hainque@adacore.com>
12550 * common.opt (nolibc): New option.
12551 * doc/invoke.texi (Link Options): Document it.
12552 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
12553 * config/alpha/linux.h: Likewise.
12554 * config/arc/elf.h: Likewise.
12555 * config/arm/uclinux-elf.h: Likewise.
12556 * config/arm/unknown-elf.h: Likewise.
12557 * config/avr/avrlibc.h: Likewise.
12558 * config/bfin/bfin.h: Likewise.
12559 * config/bfin/linux.h: Likewise.
12560 * config/bfin/uclinux.h: Likewise.
12561 * config/darwin.h: Likewise.
12562 * config/darwin10.h: Likewise.
12563 * config/darwin12.h: Likewise.
12564 * config/gnu-user.h: Likewise.
12565 * config/lm32/uclinux-elf.h: Likewise.
12566 * config/pa/pa-hpux11.h: Likewise.
12567 * config/pa/pa64-hpux.h: Likewise.
12568 * config/sparc/sparc.h: Likewise.
12570 2018-07-31 Olivier Hainque <hainque@adacore.com>
12572 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
12573 undefined variables.
12575 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
12578 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
12579 instead of GEN_INT.
12581 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
12583 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
12584 terminated string literal.
12586 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
12588 PR rtl-optimization/85160
12589 * combine.c (is_just_move): New function.
12590 (try_combine): Allow combining two instructions into two if neither of
12591 the original instructions was a move.
12593 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
12596 * doc/extend.texi (Global Register Variables): Discourage use of type
12598 (Local Register Variables): Likewise.
12600 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
12602 PR tree-optimization/86506
12603 * hwint.h (ceil_log2): Resync with hwint.c implementation.
12605 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
12608 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
12609 hard_regno, make sure no insn between `from` and `to` clobbers it.
12611 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
12612 Tom de Vries <tdevries@suse.de>
12614 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
12615 (PTX_DEFAULT_RUNTIME_DIM): ... this.
12616 (nvptx_goacc_validate_dims): Set default worker and gang dims to
12617 PTX_DEFAULT_RUNTIME_DIM.
12618 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
12620 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
12622 * config/pa/pa.c (pa_output_addr_vec): Align address table.
12623 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
12624 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
12626 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
12628 * config/rs6000/constraints.md (wG constraint): Delete, no longer
12630 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
12631 predicate to reflect toc fusion has been deleted.
12632 (toc_fusion_mem_raw): Delete, no longer used.
12633 (toc_fusion_mem_wrapped): Likewise.
12634 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
12636 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
12637 Delete, no longer used.
12638 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
12639 meant to be used for toc fusion.
12640 (rs6000_debug_print_mode): Delete toc fusion debugging.
12641 (rs6000_debug_reg_global): Likewise.
12642 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
12643 fusion and secondary reload support that were never used.
12644 (rs6000_option_override_internal): Delete TOC fusion, that was only
12645 partially defined, and it did not work unless you also used the
12647 (rs6000_legitimate_address_p): Delete TOC fusion support.
12648 (rs6000_opt_masks): Likewise.
12649 (fusion_wrap_memory_address): Delete function, no longer used.
12650 (fusion_split_address); Delete TOC fusion support.
12651 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
12652 longer used with toc fusion being deleted.
12653 (TARGET_TOC_FUSION_FP): Likewise.
12654 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
12656 (toc fusion spliter): Delete TOC fusion support.
12657 (toc_fusionload_<mode>): Likewise.
12658 (toc_fusionload_di): Likewise.
12659 (fusion_gpr_load_<mode>): Delete generator function, this insn no
12660 longer needs to be named. Rename predicate to delete TOC fusion.
12661 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
12662 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
12663 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
12664 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
12665 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
12667 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12669 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
12670 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
12671 __int128 in built-in function prototypes.
12672 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
12673 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
12675 2018-07-27 Martin Sebor <msebor@redhat.com>
12677 PR tree-optimization/86696
12678 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
12679 types, including enums.
12680 (handle_char_store): Be prepared for the above function to fail.
12682 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
12684 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
12685 when optimization level is lower than 2 or optimize for size.
12687 2018-07-26 Martin Sebor <msebor@redhat.com>
12689 PR tree-optimization/86043
12690 PR tree-optimization/86042
12691 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
12692 (get_string_cst_length): Rename...
12693 (get_min_string_length): ...to this. Add argument.
12694 (handle_char_store): Extend to handle multi-character stores by
12696 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
12697 * tree.h (initializer_zerop): Add argument.
12699 2018-07-26 Jakub Jelinek <jakub@redhat.com>
12701 PR middle-end/86660
12702 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
12703 declare target to variables if they have always,{to,from,tofrom} map
12706 2018-07-26 Martin Liska <mliska@suse.cz>
12709 * lto-wrapper.c: Add linker_output as prefix
12710 for ltrans_output_file.
12712 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
12714 PR rtl-optimization/85805
12715 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
12716 value for hard registers if that was written in the same mode.
12718 2018-07-26 Martin Liska <mliska@suse.cz>
12720 PR gcov-profile/86536
12721 * gcov.c (format_gcov): Use printf format %.*f directly
12722 and do not handle special values.
12724 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
12726 * common/config/arc/arc-common.c (arc_option_optimization_table):
12727 Update default optimizations for size.
12729 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
12731 * config/arc/arc.md (movsf_insn): Add short instruction selection.
12732 * config/arc/constraints.md (CfZ): New constraint.
12733 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
12734 (subsf3_fpu): Likewise.
12735 (cmpsf_fpu): Likewise.
12736 (cmpsf_fpu_uneq): Likewise.
12738 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
12740 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
12742 (arc_is_uncached_mem_p): uncached applies to both the variable and
12745 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
12747 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
12750 2018-07-25 David Malcolm <dmalcolm@redhat.com>
12752 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
12753 field "m_scopes" from vec to auto_vec.
12755 2018-07-25 Martin Liska <mliska@suse.cz>
12757 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
12760 2018-07-25 Richard Biener <rguenther@suse.de>
12763 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
12764 special wrt context_die late.
12765 (gen_subprogram_die): Re-use DIEs in local scope.
12767 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
12769 PR tree-optimization/86644
12770 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
12772 2018-07-25 Martin Liska <mliska@suse.cz>
12774 PR middle-end/86645
12775 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
12776 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
12778 2018-07-25 Martin Liska <mliska@suse.cz>
12781 * params.def: Explain ASan abbreviation and provide
12782 a documentation link.
12784 2018-07-24 Martin Sebor <msebor@redhat.com>
12786 PR tree-optimization/86622
12787 PR tree-optimization/86532
12788 * builtins.h (string_length): Declare.
12789 * builtins.c (c_strlen): Correct handling of non-constant offsets.
12790 (check_access): Be prepared for non-constant length ranges.
12791 (string_length): Make extern.
12792 * expr.c (string_constant): Only handle the minor non-constant
12793 array index. Use string_constant to compute the length of
12794 a generic string constant.
12796 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
12798 PR tree-optimization/86618
12799 * tree-vect-stmts.c (vectorizable_call): Don't take the address
12800 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
12802 2018-07-24 David Malcolm <dmalcolm@redhat.com>
12804 PR tree-optimization/86636
12805 * json.cc (json::object::set): Fix comment. Add assertions.
12806 (json::array::append): Move here from json.h. Add comment and an
12808 (json::string::string): Likewise.
12809 * json.h (json::array::append): Move to json.cc.
12810 (json::string::string): Likewise.
12811 * optinfo-emit-json.cc
12812 (optrecord_json_writer::impl_location_to_json): Assert that we
12813 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
12814 wrapper around it. Expand the location once, rather than three
12816 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
12817 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
12819 (optrecord_json_writer::optinfo_to_json): Likewise, in four
12820 places. Fix some overlong lines.
12822 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
12824 * config/aarch64/aarch64-simd.md
12825 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
12826 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
12827 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
12828 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
12829 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
12830 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
12831 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
12832 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
12833 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
12835 2018-07-24 Jakub Jelinek <jakub@redhat.com>
12837 PR middle-end/86627
12838 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
12839 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
12840 and abs_d == d, do the power of two handling if profitable.
12842 2018-07-24 Richard Biener <rguenther@suse.de>
12844 * match.pd: Add BIT_FIELD_REF canonicalizations.
12846 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12849 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
12851 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12853 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
12854 terminated STRING_CST object.
12856 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12858 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
12860 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
12862 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
12863 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
12864 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
12865 the elements into a register.
12866 (rs6000_split_v4si_init_di_reg): Delete.
12867 (rs6000_split_v4si_init): Delete.
12868 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
12869 (vsx_init_v4si): Rewrite as a define_expand.
12871 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
12873 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
12874 zero_extend argument from memory): New.
12876 2018-07-22 Martin Sebor <msebor@redhat.com>
12879 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
12880 diagnosing calls with unknown arguments unless -Walloca-larger-than
12881 is restricted to less than PTRDIFF_MAX bytes.
12883 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
12885 * doc/gcov.texi (Invoking Gcov): Editorial changes.
12887 2018-07-20 David Malcolm <dmalcolm@redhat.com>
12889 * pretty-print.c (text_info::set_location): Remove redundant
12890 "line_table" parameter from call to rich_location::set_range.
12892 2018-07-20 Martin Sebor <msebor@redhat.com>
12894 PR middle-end/82063
12895 * builtins.c (expand_builtin_alloca): Adjust.
12896 * calls.c (alloc_max_size): Simplify.
12897 * cgraphunit.c (cgraph_node::expand): Adjust.
12898 * common.opt (larger_than_size, warn_frame_larger_than): Remove
12900 (frame_larger_than_size): Same.
12901 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
12902 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
12904 * doc/invoke.texi (GCC Command Options): Document option arguments.
12905 Explain byte-size arguments and suffixes.
12906 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
12907 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
12908 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
12909 * doc/options.texi (UInteger): Expand.
12910 (Host_Wide_Int, ByteSize): Document new properties.
12911 * final.c (final_start_function_1): Include sizes in an error message.
12912 * function.c (frame_offset_overflow): Same.
12913 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
12914 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
12915 Diagnose unbounded alloca calls only for limits of less than
12917 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
12918 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
12920 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
12921 only for limits of less than PTRDIFF_MAX.
12922 * langhooks-def.h (lhd_handle_option): Change function argument
12924 * langhooks.c (lhd_handle_option): Same.
12925 * langhooks.h (handle_option): Same.
12926 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
12928 (var_type, var_type_struct): Same.
12929 (var_set): Handle ByteSize flag.
12930 * optc-gen.awk: Add comments to output to ease debugging. Make
12931 use of HOST_WIDE_INT where appropriate.
12932 * opts-gen-save.awk: Use %lx to format unsigned long.
12933 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
12934 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
12935 arguments. Parse bytes-size suffixes.
12936 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
12937 (enum_value_to_arg): Same.
12938 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
12939 (handle_option): Adjust.
12940 (generate_option): Change function argument to HOST_WIDE_INT.
12941 (cmdline_handle_error): Adjust.
12942 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
12943 (set_option): Change function argument to HOST_WIDE_INT.
12944 (option_enabled): Handle cl_host_wide_int.
12945 (get_option_state): Handle CLVC_SIZE.
12946 (control_warning_option): Same.
12947 * opts.c (common_handle_option): Change function argument to
12948 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
12949 OPT_Wvla_larger_than_.
12950 * opts.h (enum cl_var_type): Add an enumerator.
12951 * stor-layout.c (layout_decl): Print a more meaningful warning.
12952 * toplev.c (output_stack_usage): Adjust.
12954 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
12956 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
12957 call to inline_expand_builtin_string_cmp.
12958 (expand_builtin_strcmp): Likewise.
12959 (expand_builtin_strncmp): Likewise.
12960 (inline_string_cmp): Delete the last parameter, change char_type_node
12961 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
12963 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
12964 the inlining expansion on target where the type of the call has same or
12965 narrower precision than unsigned char.
12967 2018-07-20 David Malcolm <dmalcolm@redhat.com>
12969 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
12970 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
12971 * common.opt (fsave-optimization-record): New option.
12972 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
12973 * doc/invoke.texi (-fsave-optimization-record): New option.
12974 * dumpfile.c: Include "optinfo-emit-json.h".
12975 (struct kv_pair): Move to coretypes.h.
12976 (optgroup_options): Make non-static.
12977 (dump_context::end_scope): Call
12978 optimization_records_maybe_pop_dump_scope.
12979 * dumpfile.h (optgroup_options): New decl.
12980 * json.cc: New file.
12981 * json.h: New file.
12982 * optinfo-emit-json.cc: New file.
12983 * optinfo-emit-json.h: New file.
12984 * optinfo.cc: Include "optinfo-emit-json.h".
12985 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
12986 (optinfo_enabled_p): Check optimization_records_enabled_p.
12987 (optinfo_wants_inlining_info_p): Likewise.
12988 * optinfo.h: Update comment.
12989 * profile-count.c (profile_quality_as_string): New function.
12990 * profile-count.h (profile_quality_as_string): New decl.
12991 (profile_count::quality): New accessor.
12992 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
12993 and optinfo_emit_json_cc_tests.
12994 * selftest.h (selftest::json_cc_tests): New decl.
12995 (selftest::optinfo_emit_json_cc_tests): New decl.
12996 * toplev.c: Include "optinfo-emit-json.h".
12997 (compile_file): Call optimization_records_finish.
12998 (do_compile): Call optimization_records_start.
12999 * tree-ssa-live.c: Include optinfo.h.
13000 (remove_unused_scope_block_p): Retain inlining information if
13001 optinfo_wants_inlining_info_p returns true.
13003 2018-07-20 Richard Biener <rguenther@suse.de>
13006 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
13007 to cover -flto-partition=none.
13009 2018-07-20 Martin Liska <mliska@suse.cz>
13011 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
13012 (get_decl_source_range): Remove unused function.
13014 2018-07-20 Richard Biener <rguenther@suse.de>
13016 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
13017 (struct vn_phi_s): Likewise.
13018 (struct vn_reference_s): Likewise.
13019 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
13020 for searching the slot of an entry known to be in the hash itself.
13021 (vn_phi_hasher::equal): Likewise.
13022 (vn_reference_hasher::equal): Likewise.
13023 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
13025 (optimistic_info, current_info): Remove, keeping only valid_info.
13026 (vn_reference_lookup_1): Remove fallback lookup.
13027 (vn_reference_lookup_2): Likewise.
13028 (vn_nary_op_lookup_1): Likewise.
13029 (vn_phi_lookup): Likewise.
13030 (vn_nary_build_or_lookup_1): Make sure to not chain the built
13032 (vn_reference_insert): Adjust, chain the inserted hash element
13033 at last_inserted_ref.
13034 (vn_reference_insert_pieces): Likewise.
13035 (visit_reference_op_call): Likewise.
13036 (vn_nary_op_insert_into): Chain the inserted hash element at
13037 last_inserted_nary.
13038 (vn_nary_op_insert_pieces): Adjust.
13039 (vn_nary_op_insert): Likewise.
13040 (vn_nary_op_insert_stmt): Likewise.
13041 (vn_phi_insert): Adjust, chain the inserted hash element at
13043 (process_scc): Remove clearing and copying the optimistic
13044 table. Instead remove elements inserted during an optimistic
13045 iteration from the single table we maintain.
13046 (init_scc_vn): Adjust.
13047 (free_scc_vn): Likewise.
13048 (sccvn_dom_walker::record_cond): Likewise.
13049 (sccvn_dom_walker::after_dom_children): Likewise.
13051 2018-07-19 Martin Sebor <msebor@redhat.com>
13053 PR tree-optimization/84047
13054 PR tree-optimization/83776
13055 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
13056 (check_array_bounds): Call it.
13058 2018-07-19 Martin Sebor <msebor@redhat.com>
13060 * align.h (align_flags): Use member initialization.
13062 2018-07-19 David Malcolm <dmalcolm@redhat.com>
13064 * Makefile.in (OBJS): Add optinfo.o.
13065 * coretypes.h (class symtab_node): New forward decl.
13066 (struct cgraph_node): New forward decl.
13067 (class varpool_node): New forward decl.
13068 * dump-context.h: New file.
13069 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
13071 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
13072 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
13073 (set_alt_dump_file): Likewise.
13074 (dump_context::~dump_context): New dtor.
13075 (dump_gimple_stmt): Move implementation to...
13076 (dump_context::dump_gimple_stmt): ...this new member function.
13077 Add the stmt to any pending optinfo, creating one if need be.
13078 (dump_gimple_stmt_loc): Move implementation to...
13079 (dump_context::dump_gimple_stmt_loc): ...this new member function.
13080 Start a new optinfo and add the stmt to it.
13081 (dump_gimple_expr): Move implementation to...
13082 (dump_context::dump_gimple_expr): ...this new member function.
13083 Add the stmt to any pending optinfo, creating one if need be.
13084 (dump_gimple_expr_loc): Move implementation to...
13085 (dump_context::dump_gimple_expr_loc): ...this new member function.
13086 Start a new optinfo and add the stmt to it.
13087 (dump_generic_expr): Move implementation to...
13088 (dump_context::dump_generic_expr): ...this new member function.
13089 Add the tree to any pending optinfo, creating one if need be.
13090 (dump_generic_expr_loc): Move implementation to...
13091 (dump_context::dump_generic_expr_loc): ...this new member
13092 function. Add the tree to any pending optinfo, creating one if
13094 (dump_printf): Move implementation to...
13095 (dump_context::dump_printf_va): ...this new member function. Add
13096 the text to any pending optinfo, creating one if need be.
13097 (dump_printf_loc): Move implementation to...
13098 (dump_context::dump_printf_loc_va): ...this new member function.
13099 Start a new optinfo and add the stmt to it.
13100 (dump_dec): Move implementation to...
13101 (dump_context::dump_dec): ...this new member function. Add the
13102 value to any pending optinfo, creating one if need be.
13103 (dump_context::dump_symtab_node): New member function.
13104 (dump_context::get_scope_depth): New member function.
13105 (dump_context::begin_scope): New member function.
13106 (dump_context::end_scope): New member function.
13107 (dump_context::ensure_pending_optinfo): New member function.
13108 (dump_context::begin_next_optinfo): New member function.
13109 (dump_context::end_any_optinfo): New member function.
13110 (dump_context::s_current): New global.
13111 (dump_context::s_default): New global.
13112 (dump_scope_depth): Delete global.
13113 (dumpfile_ensure_any_optinfo_are_flushed): New function.
13114 (dump_symtab_node): New function.
13115 (get_dump_scope_depth): Reimplement in terms of dump_context.
13116 (dump_begin_scope): Likewise.
13117 (dump_end_scope): Likewise.
13118 (selftest::temp_dump_context::temp_dump_context): New ctor.
13119 (selftest::temp_dump_context::~temp_dump_context): New dtor.
13120 (selftest::verify_item): New function.
13121 (ASSERT_IS_TEXT): New macro.
13122 (ASSERT_IS_TREE): New macro.
13123 (ASSERT_IS_GIMPLE): New macro.
13124 (selftest::test_capture_of_dump_calls): New test.
13125 (selftest::dumpfile_c_tests): Call it.
13126 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
13127 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
13128 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
13129 descriptive comment.
13130 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
13131 (dump_node, dump_bb): Move these unrelated decls.
13132 (class dump_manager): Add leading comment.
13133 * optinfo.cc: New file.
13134 * optinfo.h: New file.
13136 2018-07-19 Michael Collison <michael.collison@arm.com>
13137 Richard Henderson <rth@redhat.com>
13139 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
13140 (subti): Handle op1 zero.
13141 (subvti4, usub4ti4): New.
13142 (*sub<GPI>3_compare1_imm): New.
13143 (sub<GPI>3_carryinCV): New.
13144 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
13145 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
13147 2018-07-19 Michael Collison <michael.collison@arm.com>
13148 Richard Henderson <rth@redhat.com>
13150 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
13151 (addti3): Create simpler code if low part is already known to be 0.
13152 (addvti4, uaddvti4): New.
13153 (*add<GPI>3_compareC_cconly_imm): New.
13154 (*add<GPI>3_compareC_cconly): New.
13155 (*add<GPI>3_compareC_imm): New.
13156 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
13157 handle constants within this pattern..
13158 (*add<GPI>3_compareV_cconly_imm): New.
13159 (*add<GPI>3_compareV_cconly): New.
13160 (*add<GPI>3_compareV_imm): New.
13161 (add<GPI>3_compareV): New.
13162 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
13163 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
13164 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
13165 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
13166 with 'comparison' operator.
13167 (*add<GPI>3_compareV_cconly_imm): Ditto.
13168 (*add<GPI>3_compareV_cconly): Ditto.
13169 (*add<GPI>3_compareV_imm): Ditto.
13170 (add<GPI>3_compareV): Ditto.
13171 (add<mode>3_carryinC): Ditto.
13172 (*add<mode>3_carryinC_zero): Ditto.
13173 (*add<mode>3_carryinC): Ditto.
13174 (add<mode>3_carryinV): Ditto.
13175 (*add<mode>3_carryinV_zero): Ditto.
13176 (*add<mode>3_carryinV): Ditto.
13178 2018-07-19 Michael Collison <michael.collison@arm.com>
13179 Richard Henderson <rth@redhat.com>
13181 * config/aarch64/aarch64-modes.def (CC_V): New.
13182 * config/aarch64/aarch64-protos.h
13183 (aarch64_addti_scratch_regs): Declare
13184 (aarch64_subvti_scratch_regs): Declare.
13185 (aarch64_expand_subvti): Declare.
13186 (aarch64_gen_unlikely_cbranch): Declare
13187 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
13188 for signed overflow using CC_Vmode.
13189 (aarch64_get_condition_code_1): Handle CC_Vmode.
13190 (aarch64_gen_unlikely_cbranch): New function.
13191 (aarch64_addti_scratch_regs): New function.
13192 (aarch64_subvti_scratch_regs): New function.
13193 (aarch64_expand_subvti): New function.
13195 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
13197 * config/aarch64/aarch64-option-extensions.def: New entry for profile
13199 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
13200 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
13203 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
13206 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
13207 address check not strict.
13209 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
13211 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
13213 (store_pair_lanes<mode>): Likewise.
13214 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
13215 enum value 'ADDR_QUERY_LDP_STP_N'.
13216 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
13217 (aarch64_print_address_internal): Add declaration.
13218 (aarch64_print_ldpstp_address): Remove.
13219 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
13220 (aarch64_print_operand): Change printing of 'y'.
13221 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
13222 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
13223 'true' rather than '1'.
13224 * config/aarch64/constraints.md (Uml): Likewise.
13225 (Uml): Rename to Umn.
13228 2018-07-19 Richard Biener <rguenther@suse.de>
13230 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
13232 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
13233 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
13234 (vn_reference_hasher): Likewise.
13235 (struct vn_tables_s): Remove obstack and alloc-pool members.
13236 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
13237 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
13238 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
13239 (vn_reference_insert_pieces): Likewise.
13240 (alloc_vn_nary_op_noinit): Adjust.
13241 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
13242 (vn_phi_eq): Adjust.
13243 (shared_lookup_phiargs): Remove.
13244 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
13245 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
13246 (visit_reference_op_call): Likewise.
13247 (copy_nary, copy_phi, copy_reference): Remove.
13248 (process_scc): Rewind the obstack when iterating. Do not
13249 copy the elements to valid_info but just move them from one
13250 hashtable to the other.
13251 (allocate_vn_table): Adjust.
13252 (free_vn_table): Likewise.
13253 (init_scc_vn): Likewise.
13254 (free_scc_vn): Likewise.
13256 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
13259 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
13260 indirect_return as function type attribute.
13261 (ix86_attribute_table): Change indirect_return to function
13263 * doc/extend.texi: Update indirect_return attribute.
13265 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
13267 * wide-int.h (widest2_int): New.
13268 * gimple-fold.c (arith_overflowed_p): Use it.
13269 * tree.h (widest2_int_cst): New.
13270 * tree-vrp.c (wide_int_binop_overflow): Rename from
13271 vrp_int_const_binop.
13272 Rewrite to work on trees.
13273 (extract_range_from_multiplicative_op_1): Abstract code to...
13274 (wide_int_range_min_max): ...here.
13275 (wide_int_range_cross_product): ...and here.
13276 (extract_range_from_binary_expr_1): Abstract overflow code to...
13277 (wide_int_range_mult_wrapping): ...here.
13278 * tree-vrp.h (wide_int_range_cross_product): New.
13279 (wide_int_range_mult_wrapping): New.
13281 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
13282 Julia Koval <julia.koval@intel.com>
13284 * config/i386/x86-tune-costs.h (skylake_memcpy,
13285 skylake_memset): Replace rep_prefix with unrolling for size 512.
13287 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
13289 PR middle-end/86544
13290 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
13291 comparision with EQ_EXPR in last stmt.
13293 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
13295 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
13296 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
13297 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
13298 previously known as "PowerPC AltiVec Built-in Functions". Move
13299 some material to new subsubsections "PowerPC AltiVec Built-in
13300 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
13302 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
13303 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
13304 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
13305 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
13307 2018-07-18 Richard Biener <rguenther@suse.de>
13309 PR tree-optimization/86557
13310 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
13313 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
13315 * config/s390/s390.c (s390_function_profiler): Generate CFI.
13317 2018-07-17 Jeff Law <law@redhat.com>
13319 * config/arm/arm.c (get_label_padding): Update for recent
13320 changes to label_to_alignment.
13322 PR tree-optimization/86010
13323 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
13325 * config/mips/mips.c (vr4130_align_insns): Update for recent
13326 changes to label_to_alignment.
13328 * config/frv/frv.c (frv_label_align): Update for recent changes
13329 to label_to_alignment.
13331 * config/nios2/nios2.c (nios2_label_align): Update for recent
13332 changes which dropped ALIGN_LABELS_LOG.
13334 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
13336 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
13337 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
13339 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13341 * config/arc/arc.c (arc_label_align): Use align_labels instead of
13342 deprecated align_labels_log.
13344 2018-07-17 Richard Biener <rguenther@suse.de>
13347 * dwarf2out.c (init_sections_and_labels): Always generate
13348 a debug_line_str_section for early LTO debug.
13349 (dwarf2out_finish): Reset debug_line_str_hash output early.
13350 Bump counter for extra dwarf5 .debug_loc labels to not conflict
13352 (dwarf2out_early_finish): Output debug_line_str.
13354 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
13356 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
13357 index register on z196 or later.
13359 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
13361 * config/s390/s390.c (s390_default_align): Set default function
13363 (s390_override_options_after_change): Call s390_default align.
13364 (s390_option_override_internal): Call s390_default_align.
13365 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
13367 2018-07-17 Jakub Jelinek <jakub@redhat.com>
13369 PR middle-end/86542
13370 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
13371 to _looptemp_ clauses, other than the first two.
13373 2018-07-17 Martin Liska <mliska@suse.cz>
13375 * opts.c: Do not enable OPT_falign_* for -Os.
13377 2018-07-17 Martin Liska <mliska@suse.cz>
13379 * align.h (MAX_CODE_ALIGN): New.
13380 (MAX_CODE_ALIGN_VALUE): New.
13381 * common/config/i386/i386-common.c (ix86_handle_option):
13382 (MAX_CODE_ALIGN): Moved to align.h.
13383 * final.c (MAX_CODE_ALIGN): Likewise.
13384 * opts.c (parse_and_check_align_values):
13385 (MAX_CODE_ALIGN): Likewise.
13386 (MAX_CODE_ALIGN_VALUE): Likewise.
13388 2018-07-17 Martin Liska <mliska@suse.cz>
13390 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
13391 in order to fulfil coding style.
13392 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
13393 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13394 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
13395 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
13396 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
13397 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
13398 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
13399 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
13400 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13402 2018-07-17 Martin Liska <mliska@suse.cz>
13404 * align.h: New file.
13405 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
13407 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
13408 align_flags of label_to_alignment.
13409 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
13411 * config/m68k/m68k.c: Do not use removed align_labels_value and
13413 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
13414 (LOOP_ALIGN): Likewise.
13415 (LABEL_ALIGN): Likewise.
13416 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
13417 Remove not used macro.
13418 (rs6000_loop_align): Change return type to align_flags.
13419 (rs6000_loop_align_max_skip): Remove.
13420 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
13421 Change return type to align_flags.
13422 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
13423 Remove not used macro.
13424 (rs6000_loop_align): Change return type to align_flags.
13425 (rs6000_loop_align_max_skip): Remove.
13426 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
13427 * config/rx/rx-protos.h (rx_align_for_label): Make it
13429 * config/rx/rx.c (rx_align_for_label): Change return type
13431 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
13433 into align_flags class.
13434 (LABEL_ALIGN): Likewise.
13435 (LOOP_ALIGN): Likewise.
13436 * config/s390/s390.c (s390_label_align): Use align_flags
13438 (s390_asm_output_function_label): Likewise.
13439 * config/sh/sh.c (sh_override_options_after_change):
13440 Use align_flags class directly without macros.
13441 (find_barrier): Likewise.
13442 (barrier_align): Likewise.
13443 (sh_loop_align): Likewise.
13444 * config/spu/spu.c (spu_option_override):
13445 Use align_flags_tuple::get_value instead of removed macros.
13446 (spu_sched_init): Likewise.
13447 * config/spu/spu.h (GTY): Likewise.
13448 * config/visium/visium.c (visium_option_override):
13449 Set "8" as default secondary alignment.
13450 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
13451 in order to guarantee secondary alignment of 8.
13452 * coretypes.h: Include align.h header file.
13453 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
13454 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
13455 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
13456 * doc/tm.texi.in: Likewise.
13457 * final.c (struct label_alignment): Remove not used structure.
13458 (LABEL_ALIGN): Change type to align_flags.
13459 (LOOP_ALIGN): Likewise.
13460 (JUMP_ALIGN): Likewise.
13461 (default_loop_align_max_skip): Remove.
13462 (default_label_align_max_skip): Likewise.
13463 (default_jump_align_max_skip): Likewise.
13464 (default_label_align_after_barrier_max_skip):
13465 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
13466 (LABEL_TO_MAX_SKIP): Remove.
13467 (label_to_alignment): Return align_flags type instead of integer.
13468 (label_to_max_skip): Remove.
13469 (align_fuzz): Use align_flags type.
13470 (compute_alignments): Use align_flags type and use align_flags::max
13471 to combine multiple alignments.
13472 (grow_label_align): Grow vec instead of C array.
13473 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
13474 (shorten_branches): Use align_flags type and use align_flags::max
13475 to combine multiple alignments.
13476 (final_scan_insn_1): Remove usage of secondary alignment that comes
13477 from label alignment, but instead use proper secondary alignment
13478 which is computed in grow_label_align.
13479 * flags.h (struct align_flags_tuple): Move to align.h.
13480 (struct align_flags): Likewise.
13481 (state_align_loops): Rename to align_loops.
13482 (state_align_jumps): Rename to align_jumps.
13483 (state_align_labels): Rename to align_labels.
13484 (state_align_functions): Rename to align_functions.
13485 (align_loops_log): Remove.
13486 (align_jumps_log): Remove.
13487 (align_labels_log): Remove.
13488 (align_functions_log): Remove.
13489 (align_loops_max_skip): Remove.
13490 (align_jumps_max_skip): Remove.
13491 (align_labels_max_skip): Remove.
13492 (align_functions_max_skip): Remove.
13493 (align_loops_value): Remove.
13494 (align_jumps_value): Remove.
13495 (align_labels_value): Remove.
13496 (align_functions_value): Remove.
13497 * output.h (label_to_alignment): Change return type to align_flags.
13498 (label_to_max_skip): Remove.
13499 * target.def: Remove loop_align_max_skip, label_align_max_skip,
13500 jump_align_max_skip macros.
13501 * targhooks.h (default_loop_align_max_skip): Remove.
13502 (default_label_align_max_skip): Likewise.
13503 (default_jump_align_max_skip): Likewise.
13504 (default_label_align_after_barrier_max_skip): Remove.
13505 * toplev.c (read_log_maxskip): Use ::normalize function.
13506 (parse_N_M): Remove not used argument and also call ::normalize.
13507 (parse_alignment_opts): Do not pass unused arguments.
13508 * varasm.c (assemble_start_function): Use directly align_functions
13509 instead of removed macros.
13510 * system.h: Do not poison removed macros.
13512 2018-07-17 Jakub Jelinek <jakub@redhat.com>
13514 PR middle-end/86539
13515 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
13516 and cond temporaries don't have reference type if iterator has
13517 pointer type. For init use &for_pre_body instead of pre_p if
13518 for_pre_body is non-empty.
13520 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
13522 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
13523 double-double modes to SFmode directly directly.
13524 (trunc<mode>sf2_fprs): Delete.
13526 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
13528 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
13529 for conversions between IFmode and the decimal floating point modes.
13530 (init_float128_ieee): Use the correct names for conversions between
13531 KFmode and the decimal floating point modes.
13533 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
13535 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
13536 for the conversions between TDmode and IFmode.
13537 (init_float128_ieee): Use more correct names for the conversions
13538 between TDmode and KFmode.
13540 2018-07-16 Jakub Jelinek <jakub@redhat.com>
13542 PR tree-optimization/86526
13543 * builtins.c (expand_builtin_memcmp): Formatting fixes.
13544 (inline_expand_builtin_string_cmp): Likewise.
13545 (inline_string_cmp): Likewise. Use c_readstr instead of
13546 builtin_memcpy_read_str. Add unit_mode temporary.
13548 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
13550 PR middle-end/86528
13551 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
13552 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
13554 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
13556 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
13557 Alphabetize prototypes of built-in functions, separating out
13558 built-in functions that are listed in this section but should be
13559 described elsewhere.
13561 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
13564 * expmed.c (emit_store_flag): Do not emit setcc followed by a
13565 conditional move when trapping comparison was split to a
13566 non-trapping one (and vice versa).
13568 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
13570 * config/s390/s390.c (s390_function_profiler): Generate nops
13571 instead of profiler call sequences.
13572 * config/s390/s390.opt: Add the new option.
13574 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
13576 * config/s390/s390.c (s390_function_profiler): Generate
13577 __mcount_loc section.
13578 * config/s390/s390.opt: Add the new option.
13580 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
13582 * common.opt: Add the new warning.
13583 * config/s390/s390.c (s390_function_profiler): Emit "brasl
13584 %r0,__fentry__" when -mfentry is specified.
13585 (s390_option_override_internal): Disallow -mfentry for 31-bit
13587 * config/s390/s390.opt: Add the new option.
13589 2018-07-16 Richard Biener <rguenther@suse.de>
13592 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
13593 for function-local FUNCTION_DECL and RESULT_DECL immediately.
13595 2018-07-16 Martin Liska <mliska@suse.cz>
13598 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
13601 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
13603 * config/arc/arcHS.md: Update ARCHS scheduling rules.
13605 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
13607 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
13609 * config/arc/arc-cpus.def (hs4x): New cpu.
13611 * config/arc/arc-tables.opt: Regenerate.
13612 * config/arc/arc.c (arc_sched_issue_rate): New function.
13613 (TARGET_SCHED_ISSUE_RATE): Define.
13614 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
13615 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
13617 (attr tune): Add ARCHS4x tune values.
13618 (attr tune_dspmpy): Define.
13619 (*tst): Correct instruction type.
13620 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
13621 * config/arc/arcHS4x.md: New file.
13622 * config/arc/fpu.md: Update instruction type attributes.
13623 * config/arc/t-multilib: Regenerate.
13625 2018-07-16 Tom de Vries <tdevries@suse.de>
13628 * var-tracking.c (vt_initialize): Fix pre_dec handling.
13630 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13632 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
13635 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
13637 PR tree-optimization/86514
13638 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
13639 conversion to a boolean type from a type with greater precision.
13641 2018-07-16 Tom de Vries <tdevries@suse.de>
13643 * var-tracking.c (vt_initialize): Print adjusted insn slim if
13644 dump_flags request TDF_SLIM.
13646 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
13648 * fold-const.c (int_const_binop_1): Abstract...
13649 (wide_int_binop): ...wide int code here.
13650 (poly_int_binop): ...poly int code here.
13651 Abstract the rest of int_const_binop_1 into int_const_binop.
13652 * fold-const.h (wide_int_binop): New.
13653 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
13654 Remove useless PLUS/MINUS_EXPR case.
13655 (zero_nonzero_bits_from_vr): Move wide int code...
13656 (zero_nonzero_bits_from_bounds): ...here.
13657 (extract_range_from_binary_expr_1): Move mask optimization code...
13658 (range_easy_mask_min_max): ...here.
13659 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
13660 (range_easy_mask_min_max): New.
13662 2018-07-15 Jeff Law <law@redhat.com>
13665 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
13668 2018-07-14 Jim Wilson <jimw@sifive.com>
13670 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
13672 2018-07-14 Paul Koning <ni1d@arrl.net>
13674 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
13676 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
13678 * lto-streamer-out.c (copy_function_or_variable): Dump info about
13681 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
13682 Steve Munroe <munroesj52@gmail.com>
13684 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
13685 (_mm_andnot_si128): Likewise.
13686 (_mm_or_si128): Likewise.
13687 (_mm_xor_si128): Likewise.
13689 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
13691 PR middle-end/78809
13692 * builtins.c (expand_builtin_memcmp): Inline the calls first
13693 when result_eq is false.
13694 (expand_builtin_strcmp): Inline the calls first.
13695 (expand_builtin_strncmp): Likewise.
13696 (inline_string_cmp): New routine. Expand a string compare
13697 call by using a sequence of char comparison.
13698 (inline_expand_builtin_string_cmp): New routine. Inline expansion
13699 a call to str(n)cmp/memcmp.
13700 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
13702 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
13704 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
13706 * config/arm/driver-arm.c: Include arm-native.h.
13707 (host_detect_local_cpu): Use auto-generated data tables.
13708 (vendors, arm_cpu_table): Delete. Move part information to ...
13709 * config/arm/arm-cpus.in: ... here.
13710 * config/arm/parsecpu.awk (gen_native): New function.
13711 (vendor, part): New CPU fields.
13712 (END): Add support for building the native CPU detection tables.
13713 * config/arm/t-arm (arm-native.h): Add build rule.
13714 (driver-arm.o): Add dependency on arm-native.h.
13716 2018-07-13 Richard Biener <rguenther@suse.de>
13718 PR middle-end/85974
13719 * match.pd (addr1 - addr2): Allow either of the operand to
13722 2018-07-13 Tom de Vries <tdevries@suse.de>
13724 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
13727 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
13729 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
13730 arrays instead of numbered variables.
13732 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
13734 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
13735 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
13737 2018-07-13 Richard Biener <rguenther@suse.de>
13740 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
13741 instead of get_context_die.
13743 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
13744 Richard Biener <rguenther@suse.de>
13746 PR middle-end/86489
13747 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
13748 that the loop latch destination where phi is defined.
13750 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
13752 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
13753 (riscv_expand_epilogue): Add assertion to check interrupt mode.
13754 (riscv_set_current_function): Extract getting interrupt type to new
13756 (riscv_get_interrupt_type): New function.
13757 (riscv_merge_decl_attributes): New function, checking interrupt type is
13759 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
13761 2018-07-12 Paul Koning <ni1d@arrl.net>
13763 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
13766 2018-07-12 Paul Koning <ni1d@arrl.net>
13768 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
13769 zero reference, add doloop_end instead.
13770 * doc/md.texi (decrement_and_branch_until_zero): Remove.
13771 (Looping patterns): Remove decrement_and_branch_until_zero. Add
13772 detail for doloop_end.
13774 2018-07-12 Martin Sebor <msebor@redhat.com>
13777 * attribs.c (decl_attributes): Reject conflicting attributes before
13778 calling attribute handlers.
13780 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
13782 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
13784 (gcc::dump_manager::get_dump_file_name): likewise.
13785 (dump_begin): Likewise.
13786 * dumpfile.h (dump_begin): Update prototype.
13787 (gcc::dump_manager::get_dump_file_name,
13788 gcc::dump_manager::get_dump_file_name): Update prototype.
13790 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
13792 * internal-fn.h (vectorizable_internal_fn_p): New function.
13793 * tree-vect-slp.c (compatible_calls_p): Likewise.
13794 (vect_build_slp_tree_1): Remove nops argument. Handle calls
13795 to internal functions.
13796 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
13798 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
13800 * fold-const.h (inverse_conditions_p): Declare.
13801 * fold-const.c (inverse_conditions_p): New function.
13802 * match.pd: Use inverse_conditions_p. Add folds of view_converts
13803 that test the inverse condition of a conditional internal function.
13804 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
13805 * internal-fn.c (internal_fn_mask_index): Handle conditional
13806 internal functions.
13807 (vectorized_internal_fn_supported_p): New function.
13808 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
13809 (any_pred_load_store): Replace with...
13810 (need_to_predicate): ...this new variable.
13811 (redundant_ssa_names): New variable.
13812 (ifcvt_can_use_mask_load_store): Move initial checks to...
13813 (ifcvt_can_predicate): ...this new function. Handle tree codes
13814 for which a conditional internal function exists.
13815 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
13816 instead of ifcvt_can_use_mask_load_store. Update after variable
13818 (predicate_load_or_store): New function, split out from
13819 predicate_mem_writes.
13820 (check_redundant_cond_expr): New function.
13821 (value_available_p): Likewise.
13822 (predicate_rhs_code): Likewise.
13823 (predicate_mem_writes): Rename to...
13824 (predicate_statements): ...this. Use predicate_load_or_store
13825 and predicate_rhs_code.
13826 (combine_blocks, tree_if_conversion): Update after above name changes.
13827 (ifcvt_local_dce): Handle redundant_ssa_names.
13828 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
13829 general conditional functions.
13830 * tree-vect-stmts.c (vectorizable_call): Likewise.
13832 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
13833 Alan Hayward <alan.hayward@arm.com>
13834 David Sherwood <david.sherwood@arm.com>
13836 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
13837 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
13838 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
13839 plus and minus and convert them into IFN_COND_FMA-based sequences.
13840 (convert_mult_to_fma): Handle conditional plus and minus.
13842 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
13844 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
13845 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
13846 (cond_fnms_optab): New optabs.
13847 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
13848 internal functions.
13849 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
13850 * internal-fn.h (get_conditional_internal_fn): Declare.
13851 (get_unconditional_internal_fn): Likewise.
13852 * internal-fn.c (cond_ternary_direct): New macro.
13853 (expand_cond_ternary_optab_fn): Likewise.
13854 (direct_cond_ternary_optab_supported_p): Likewise.
13855 (FOR_EACH_COND_FN_PAIR): Likewise.
13856 (get_conditional_internal_fn): New function.
13857 (get_unconditional_internal_fn): Likewise.
13858 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
13859 (gimple_match_op::gimple_match_op): Add a new overload for 5
13861 (gimple_match_op::set_op): Likewise.
13862 (gimple_resimplify5): Declare.
13863 * genmatch.c (decision_tree::gen): Generate simplifications for
13865 * gimple-match-head.c (gimple_simplify): Define an overload for
13866 5 operands. Handle calls with 5 arguments in the top-level overload.
13867 (convert_conditional_op): Handle conversions from unconditional
13868 internal functions to conditional ones.
13869 (gimple_resimplify5): New function.
13870 (build_call_internal): Pass a fifth operand.
13871 (maybe_push_res_to_seq): Likewise.
13872 (try_conditional_simplification): Try converting conditional
13873 internal functions to unconditional internal functions.
13874 Handle 3-operand unconditional forms.
13875 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
13876 Define ternary equivalents of the current rules for binary conditional
13877 internal functions.
13878 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
13879 ternary operations.
13880 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
13881 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
13882 (optab): Handle them.
13883 (SVE_COND_FP_TERNARY): New int iterator.
13884 (sve_fmla_op, sve_fmad_op): New int attributes.
13885 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
13886 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
13887 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
13889 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
13891 * target.def (preferred_else_value): New target hook.
13892 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
13893 * doc/tm.texi: Regenerate.
13894 * targhooks.h (default_preferred_else_value): Declare.
13895 * targhooks.c (default_preferred_else_value): New function.
13896 * internal-fn.h (conditional_internal_fn_code): Declare.
13897 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
13898 (get_conditional_internal_fn): Use it.
13899 (conditional_internal_fn_code): New function.
13900 * gimple-match.h (gimple_match_cond): New struct.
13901 (gimple_match_op): Add a cond member function.
13902 (gimple_match_op::gimple_match_op): Update all forms to take a
13904 * genmatch.c (expr::gen_transform): Use the same condition as res_op
13905 for the suboperation, but don't specify a particular else_value.
13906 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
13907 (visit_nary_op, visit_reference_op_load): Pass
13908 gimple_match_cond::UNCOND to the gimple_match_op constructor.
13909 * gimple-match-head.c: Include tree-eh.h
13910 (convert_conditional_op): New function.
13911 (maybe_resimplify_conditional_op): Likewise.
13912 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
13913 (gimple_resimplify2): Likewise.
13914 (gimple_resimplify3): Likewise.
13915 (gimple_resimplify4): Likewise.
13916 (maybe_push_res_to_seq): Return null for conditional operations.
13917 (try_conditional_simplification): New function.
13918 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
13920 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
13922 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
13924 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
13926 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
13928 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
13930 (hash_tree): Do not hash DECL_FCONTEXT
13931 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
13932 Do not stream DECL_FCONTEXT.
13933 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
13934 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
13936 2018-07-12 Richard Biener <rguenther@suse.de>
13939 * dwarf2out.c (gen_block_die): Only output blocks when they have
13940 at least one !DECL_IGNORED_P variable.
13942 2018-07-12 Richard Biener <rguenther@suse.de>
13945 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
13946 Remove -mieee-fp handling.
13948 2018-07-12 Richard Biener <rguenther@suse.de>
13950 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
13951 left-over from last patch.
13953 2018-07-12 Jakub Jelinek <jakub@redhat.com>
13955 PR tree-optimization/86492
13956 * gimple-ssa-store-merging.c
13957 (imm_store_chain_info::coalesce_immediate_stores): Call
13958 check_no_overlap even for the merge_overlapping case. Formatting fix.
13960 2018-07-12 Richard Biener <rguenther@suse.de>
13962 PR middle-end/86479
13963 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
13964 move possibly trapping operations into the conditional.
13966 2018-07-12 Richard Biener <rguenther@suse.de>
13968 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
13969 (vn_lookup_simplify_result): Remove recursion limit applied
13971 (vn_nary_build_or_lookup_1): Adjust.
13972 (try_to_simplify): Likewise.
13973 * gimple-match-head.c (gimple_resimplify1): Instead apply one
13975 (gimple_resimplify2): Likewise.
13976 (gimple_resimplify3): Likewise.
13977 (gimple_resimplify4): Likewise.
13979 2018-07-11 Jakub Jelinek <jakub@redhat.com>
13981 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
13982 Use __mmask64 type instead of __mmask8 for __M argument.
13983 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
13984 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
13986 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
13987 __mmask16 for __M argument.
13988 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
13989 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
13990 to __mmask16 instead of __mmask8.
13991 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
13992 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
13993 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
13994 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
13995 instead of __mmask16 for __U argument.
13996 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
13997 __mmask16 instead of __mmask8 for __U argument.
13998 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
14000 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
14002 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
14004 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
14006 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
14008 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
14010 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
14012 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
14014 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
14016 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
14018 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
14019 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
14020 return type as well as __M argument type and all casts from __mmask8
14022 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
14023 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
14024 return type as well as __M argument type and all casts from __mmask8
14026 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
14027 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
14028 return type as well as __M argument type and all casts from __mmask8
14030 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
14031 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
14032 return type as well as __M argument type and all casts from __mmask8
14034 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
14035 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
14038 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
14040 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
14041 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
14044 2018-07-11 Paul Koning <ni1d@arrl.net>
14046 * doc/md.texi (define_subst): Document how multiple occurrences of
14047 the same argument in the replacement pattern are handled.
14049 2018-07-11 Paul Koning <ni1d@arrl.net>
14051 * doc/extend.texi (Common Variable Attributes): Move "mode" into
14052 alphabetical order.
14053 (Common Type Attributes): Add "mode" attribute.
14055 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
14057 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
14058 stream DECL_ORIGINAL_TYPE.
14059 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
14060 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
14061 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
14062 Do not walk original type.
14063 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
14065 (write_ts_decl_non_common_tree_pointers): Do not stream
14067 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
14068 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
14070 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
14072 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
14073 thread twice from the same starting edge.
14075 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
14077 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
14078 * gimple.c (gimple_call_nonnull_result_p): ...here...
14079 (gimple_call_nonnull_arg): ...and here.
14080 * gimple.h (gimple_call_nonnull_result_p): New.
14081 (gimple_call_nonnull_arg): New.
14083 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
14085 * config/arm/arm-cpus.in: Move information from fpu field of each
14086 cpu definition to the isa field.
14087 * config/arm/parsecpu.awk (fpu): Delete match rule.
14088 (gen_comm_data): Don't add bits from the CPU's FPU entry.
14090 2018-07-11 Richard Biener <rguenther@suse.de>
14093 * dwarf2out.c (init_sections_and_labels): Use
14094 output_asm_line_debug_info consistently.
14095 (dwarf2out_early_finish): Likewise.
14096 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
14099 2018-07-11 Richard Biener <rguenther@suse.de>
14101 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
14102 Rework father_bb setting in a way to avoid propagating constants
14103 multiple times on a loop body.
14105 2018-07-10 Mark Wielaard <mark@klomp.org>
14108 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
14110 2018-07-10 Richard Biener <rguenther@suse.de>
14112 * hash-map.h (hash_map::iterator::operator*): Return
14113 references to key and value.
14115 2018-07-10 Jakub Jelinek <jakub@redhat.com>
14118 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
14119 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
14120 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
14121 (gimplify_omp_for): For composite loops, move outer
14122 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
14123 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
14124 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
14125 TREE_LIST for both the original class iterator and the "last" helper
14126 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
14127 loop, remember has_decl_expr from outer composite loops for the
14128 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
14130 2018-07-09 Martin Sebor <msebor@redhat.com>
14132 PR middle-end/77357
14133 PR middle-end/86428
14134 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
14135 accessing implicitly initialized array elements.
14136 * expr.c (string_constant): Handle string initializers of
14137 character arrays within aggregates.
14138 * gimple-fold.c (fold_array_ctor_reference): Add argument.
14139 Store element offset. As a special case, handle zero size.
14140 (fold_nonarray_ctor_reference): Same.
14141 (fold_ctor_reference): Add argument. Store subobject offset.
14142 * gimple-fold.h (fold_ctor_reference): Add argument.
14144 2018-07-09 Paul Koning <ni1d@arrl.net>
14146 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
14147 (pdp11_insn_cost): New function.
14148 (pdp11_md_asm_adjust): New function.
14149 (TARGET_INVALID_WITHIN_DOLOOP): Define.
14150 (pdp11_rtx_costs): Update to match machine better.
14151 (output_addr_const_pdp11): Correct format mismatch warnings.
14152 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
14153 * config/pdp11/pdp11.md: General change to add base_cost and/or
14154 length attributes for use by new pdp11_insn_cost function.
14155 (MIN_BRANCH): Correct definition.
14157 (doloop_end): Use standard pattern name for looping pattern.
14158 (doloop_end_nocc): New.
14159 (movsf): Add another constraint alternative.
14160 (zero_extendqihi2): Add constraint alternatives for not in place
14162 (zero_extendhisi2): Remove.
14163 (shift patterns): Add CC handling variants.
14167 (define_peephole2): New peephole to recognize mov that sets CC for
14170 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14172 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
14173 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
14176 2018-07-09 Jakub Jelinek <jakub@redhat.com>
14179 * real.c (real_nextafter): Return true if result is denormal.
14181 2018-07-09 Martin Liska <mliska@suse.cz>
14183 * common.opt: Add back wrongly removed attribute.
14185 2018-07-09 Richard Biener <rguenther@suse.de>
14188 * dwarf2out.c (gen_block_die): For an early generated DIE
14189 always output high/low PC attributes.
14191 2018-07-09 Tom de Vries <tdevries@suse.de>
14193 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
14194 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
14195 onto VAR_DECL with abstract origin.
14197 2018-07-07 Jim Wilson <jimw@sifive.com>
14199 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
14201 2018-07-07 Tom de Vries <tdevries@suse.de>
14203 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
14205 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
14207 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
14209 (combine_bound): Use wide-int overflow calculation instead of
14211 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
14213 * fold-const.c (int_const_binop_2): Same.
14214 (extract_muldiv_1): Same.
14215 (fold_div_compare): Same.
14216 (fold_abs_const): Same.
14218 * poly-int.h (add): Same.
14222 * predict.c (predict_iv_comparison): Same.
14223 * profile-count.c (slow_safe_scale_64bit): Same.
14224 * simplify-rtx.c (simplify_const_binary_operation): Same.
14225 * tree-chrec.c (tree_fold_binomial): Same.
14226 * tree-data-ref.c (split_constant_offset_1): Same.
14227 * tree-if-conv.c (idx_within_array_bound): Same.
14228 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
14229 * tree-ssa-phiopt.c (minmax_replacement): Same.
14230 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
14231 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
14232 * vr-values.c (vr_values::adjust_range_with_scev): Same.
14233 * wide-int.cc (wi::add_large): Same.
14234 (wi::mul_internal): Same.
14235 (wi::sub_large): Same.
14236 (wi::divmod_internal): Same.
14237 * wide-int.h: Change overflow type to overflow_type for neg, add,
14238 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
14239 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
14240 mul_internal, divmod_internal.
14241 (overflow_type): New enum.
14242 (accumulate_overflow): New.
14244 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
14246 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
14247 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
14249 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
14251 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
14252 argument is checked for zero before entering loop, avoid checking again.
14254 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
14256 * gimplify.h (generic_expr_could_trap_p): Set as global function.
14257 * gimplify.h (generic_expr_could_trap_p): Likwise.
14258 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
14260 2018-07-06 Jakub Jelinek <jakub@redhat.com>
14262 PR tree-optimization/86401
14263 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
14264 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
14265 (fold_bit_and_mask): ... here. New helper function for match.pd.
14266 * fold-const.h (fold_bit_and_mask): Declare.
14267 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
14269 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
14272 * target.def (translate_mode_attribute): New hook.
14273 * targhooks.h (default_translate_mode_attribute): Declare.
14274 * targhooks.c (default_translate_mode_attribute): New function.
14275 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
14276 * doc/tm.texi: Regenerate.
14277 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
14278 (rs6000_translate_mode_attribute): New function.
14280 2018-07-06 Paul Koning <ni1d@arrl.net>
14282 * doc/md.texi (define_split): Document DONE and FAIL.
14283 (define_peephole2): Ditto.
14285 2018-07-05 Jeff Law <law@redhat.com>
14287 PR tree-optimization/86010
14288 * tree-ssa-dse.c (compute_trims): More aggressively trim at
14289 both the head and tail of mem* and str* calls.
14291 2018-07-05 Jim Wilson <jimw@sifive.com>
14293 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
14295 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
14297 * config/aarch64/aarch64-simd.md: correct flags text for
14298 MIN_EXPR replacement.
14300 2018-07-05 James Clarke <jrtc27@jrtc27.com>
14302 * configure: Regenerated.
14304 2018-07-05 Carl Love <cel@us.ibm.com>
14306 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
14307 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
14308 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
14309 VSX_BUILTIN_DOUBLEL_V4SF.
14311 2018-07-05 Martin Sebor <msebor@redhat.com>
14314 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
14315 than its domain to compute its the upper bound of a char array.
14317 2018-07-05 Nathan Sidwell <nathan@acm.org>
14319 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
14320 * doc/cpp.texi: Update comment.
14321 * doc/tm.texi: Rebuilt.
14322 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
14323 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
14324 * doc/extend.texi (Backwards Compatibility): Clarify it is system
14325 headers affected by extern "C".
14326 * system.h: Poison NO_IMPLICIT_EXTERN_C.
14327 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
14328 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
14329 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
14330 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
14331 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
14332 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
14333 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
14334 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
14335 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
14336 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
14337 config/sparc/sp64-elf.h, config/spu/spu.h,
14338 config/stormy16/stormy16.h, config/v850/v850.h,
14339 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
14340 define NO_IMPLICIT_EXTERN_C.
14341 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
14343 2018-07-05 Tamar Christina <tamar.christina@arm.com>
14346 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
14347 instead of GET_MODE_SIZE when comparing Units.
14349 2018-07-05 Tamar Christina <tamar.christina@arm.com>
14352 * rtlanal.c (set_noop_p): Constrain on mode change,
14353 include hard-reg-set.h
14355 2018-07-05 Tamar Christina <tamar.christina@arm.com>
14357 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
14359 2018-07-05 Jakub Jelinek <jakub@redhat.com>
14362 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
14365 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
14367 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
14369 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
14372 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
14374 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
14376 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
14378 PR middle-end/86380
14379 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
14381 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
14383 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
14384 neg_*_op* variables.
14386 2018-07-04 Martin Liska <mliska@suse.cz>
14388 * tree-switch-conversion.c: Define
14389 max_ratio_for_speed and max_ratio_for_size constants.
14391 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
14392 Martin Liska <mliska@suse.cz>
14394 PR middle-end/66240
14397 * common.opt: Rename align options with 'str_' prefix.
14398 * common/config/i386/i386-common.c (set_malign_value): New
14400 (ix86_handle_option): Use it to set -falign-* options/
14401 * config/aarch64/aarch64-protos.h (struct tune_params): Change
14402 type from int to string.
14403 * config/aarch64/aarch64.c: Update default values from int
14405 * config/alpha/alpha.c (alpha_override_options_after_change):
14407 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
14408 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
14409 max skip conditionally.
14410 * config/i386/freebsd.h (SUBALIGN_LOG): New.
14411 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
14412 max skip conditionally.
14413 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
14414 max skip conditionally.
14415 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
14416 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
14417 max skip conditionally.
14418 * config/i386/i386.c (struct ptt): Change type from int to
14420 (ix86_default_align): Set default values.
14421 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
14422 max skip conditionally.
14423 * config/i386/iamcu.h (SUBALIGN_LOG): New.
14424 (ASM_OUTPUT_MAX_SKIP_ALIGN):
14425 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
14426 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
14427 max skip conditionally.
14428 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
14429 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
14430 * config/i386/x86-64.h (SUBALIGN_LOG): New.
14431 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
14432 max skip conditionally.
14433 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
14434 * config/ia64/ia64.c (ia64_option_override): Set default values
14435 for alignment options.
14436 * config/m68k/m68k.c: Handle new str_align_* options.
14437 * config/mips/mips.c (mips_set_compression_mode): Change
14439 (mips_option_override): Set default values for options.
14440 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
14442 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14444 * config/rx/rx.c (rx_option_override): Likewise.
14445 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
14446 (LABEL_ALIGN): Use align_labels_log.
14447 (LOOP_ALIGN): Use align_loops_align.
14448 * config/s390/s390.c (s390_asm_output_function_label): Use new
14450 * config/sh/sh.c (sh_override_options_after_change):
14451 Change type of constants.
14452 * config/spu/spu.c (spu_sched_init): Likewise.
14453 * config/sparc/sparc.c (sparc_option_override): Set default
14454 values for options.
14455 * config/visium/visium.c (visium_option_override): Likewise.
14456 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
14457 emit p2align format with last argument if it's not needed.
14458 * doc/invoke.texi: Document extended format of -falign-*.
14459 * final.c: Use align_labels alignment.
14460 * flags.h (struct target_flag_state): Change type to use
14462 (struct align_flags_tuple): New.
14463 (struct align_flags): Likewise.
14464 (align_loops_log): Redefine macro to use new types.
14465 (align_loops_max_skip): Redefine macro to use new types.
14466 (align_jumps_log): Redefine macro to use new types.
14467 (align_jumps_max_skip): Redefine macro to use new types.
14468 (align_labels_log): Redefine macro to use new types.
14469 (align_labels_max_skip): Redefine macro to use new types.
14470 (align_functions_log): Redefine macro to use new types.
14471 (align_loops): Redefine macro to use new types.
14472 (align_jumps): Redefine macro to use new types.
14473 (align_labels): Redefine macro to use new types.
14474 (align_functions): Redefine macro to use new types.
14475 (align_functions_max_skip): Redefine macro to use new types.
14476 (align_loops_value): New macro.
14477 (align_jumps_value): New macro.
14478 (align_labels_value): New macro.
14479 (align_functions_value): New macro.
14480 * function.c (invoke_set_current_function_hook): Propagate
14481 alignment values from flags to global variables default in
14483 * ipa-icf.c (sem_function::equals_wpa): Use
14484 cl_optimization_option_eq instead of memcmp.
14485 * lto-streamer.h (cl_optimization_stream_out): Support streaming
14487 (cl_optimization_stream_in): Likewise.
14488 * optc-save-gen.awk: Support strings in cl_optimization.
14489 * opth-gen.awk: Likewise.
14490 * opts.c (finish_options): Remove error checking of invalid
14492 (MAX_CODE_ALIGN): Remove.
14493 (MAX_CODE_ALIGN_VALUE): Likewise.
14494 (parse_and_check_align_values): New function.
14495 (check_alignment_argument): Likewise.
14496 (common_handle_option): Use check_alignment_argument.
14497 * opts.h (parse_and_check_align_values): Declare.
14498 * toplev.c (init_alignments): Remove.
14499 (read_log_maxskip): New.
14500 (parse_N_M): Likewise.
14501 (parse_alignment_opts): Likewise.
14502 (backend_init_target): Remove usage of init_alignments.
14503 * toplev.h (parse_alignment_opts): Declare.
14504 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
14506 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
14507 * tree.c (cl_option_hasher::equal): New.
14508 * varasm.c: Use new global macros.
14510 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
14512 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
14513 Use a simpler align directive also if MAXSKIP = ALIGN-1.
14514 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14515 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14516 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14517 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
14518 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
14519 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
14520 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
14521 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14522 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14523 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14524 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14526 2018-07-04 Martin Liska <mliska@suse.cz>
14527 Jonathan Wakely <jwakely@redhat.com>
14529 * coverage.c: Use correct type.
14530 * doc/invoke.texi: Language correction.
14532 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
14535 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
14536 ENDBRANCH for non-tail call which may return via indirect branch.
14537 * doc/extend.texi: Document indirect_return attribute.
14539 2018-07-03 Martin Sebor <msebor@redhat.com>
14541 PR tree-optimization/86274
14542 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
14544 (format_floating): Correct handling of infinities and NaNs.
14546 2018-07-03 Martin Sebor <msebor@redhat.com>
14548 * print-tree.c (print_real_cst): New function.
14549 (print_node_brief): Call it.
14550 (print_node): Ditto.
14552 2018-07-03 Jeff Law <law@redhat.com>
14554 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
14555 into a single pattern.
14557 * config/h8300/h8300.md (ors code_iterator): New.
14558 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
14559 a single pattern and single splitter.
14560 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
14561 (iorqi3_1, xorqi3_1): Likewise.
14562 (iorqi3, xorqi3 expanders): Similarly.
14564 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
14565 (movmd_internal) into a single pattern using the P mode iterator.
14566 (movmd splitters): Similarly.
14567 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
14568 (movsd splitters): Similarly.
14570 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
14571 ADDB, ADDW and ADDL into a single ADD attribute which selects the
14572 right table based on the size of the operand.
14573 * config/h8300/h8300.md (length_table): Corresponding changes. All
14574 references to "addb", "addw" and "addl" changed to "add".
14575 (btst patterns): Merge two variants into a single pattern.
14576 (tstqi, tsthi): Likewise.
14577 (addhi3_incdec, addsi3_incdec): Likewise.
14578 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
14579 (mulhi3, mulsi3): Likewise.
14580 (udivhi3, udivsi3): Likewise.
14581 (divhi3, divsi3): Likewise.
14582 (andorqi3, andorhi3, andorsi3): Likewise.
14584 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
14587 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
14588 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
14590 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
14592 PR tree-optimization/85694
14593 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
14594 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
14596 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
14597 (<u>avg<mode>3_ceil): New patterns.
14599 2018-07-03 David Malcolm <dmalcolm@redhat.com>
14601 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
14602 scan-tree-dump directive.
14603 * gcc.dg/vect/slp-perm-2.c: Likewise.
14604 * gcc.dg/vect/slp-perm-3.c: Likewise.
14605 * gcc.dg/vect/slp-perm-5.c: Likewise.
14606 * gcc.dg/vect/slp-perm-6.c: Likewise.
14607 * gcc.dg/vect/slp-perm-7.c: Likewise.
14608 * gcc.dg/vect/slp-perm-8.c: Likewise.
14610 2018-07-03 Marek Polacek <polacek@redhat.com>
14612 PR middle-end/86202
14613 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
14615 2018-07-03 Richard Biener <rguenther@suse.de>
14618 * tree-ssa-structalias.c (find_func_clobbers): Properly
14619 handle indirect calls.
14621 2018-07-03 Jeff Law <law@redhat.com>
14623 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
14624 (shifts): New code iterator.
14625 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
14626 expander. Fix HImode handling on H8/SX.
14627 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
14628 (subqi3, subhi3, subsi3 expanders): Likewise.
14629 (andqi3, andhi3, andsi3 expanders): Likewise.
14630 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
14631 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
14632 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
14633 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
14634 (zero_extendqihi2, zero_extendqisi2): Likewise.
14635 (extendqihi2, extendqisi2): Likewise.
14636 (rotlqi3, rotlhi3, rotlsi3): Likewise.
14637 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
14638 (rotlqi3_1, rotlhi3_1): Likewise.
14639 (logicalhi3_sn, logicalsi3_sn): Likewise.
14640 (logicalhi3, logicalsi3): Likewise.
14642 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
14644 * tree-vect-patterns.c (vect_recog_rotate_pattern)
14645 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
14646 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
14647 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
14648 type to append_pattern_def_seq instead of creating a stmt_vec_info
14650 (build_mask_conversion): Likewise. Remove vinfo argument.
14651 (vect_add_conversion_to_patterm): Likewise, renaming to...
14652 (vect_add_conversion_to_pattern): ...this.
14653 (vect_recog_mask_conversion_pattern): Update call to
14654 build_mask_conversion. Pass the vector type to
14655 append_pattern_def_seq here too.
14656 (vect_recog_gather_scatter_pattern): Update call to
14657 vect_add_conversion_to_pattern.
14659 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
14661 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
14662 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
14663 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
14664 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
14665 STMT_VINFO_PATTERN_DEF_SEQ to null here.
14666 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
14667 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
14668 append_pattern_def_seq instead of new_pattern_def_seq.
14669 (vect_recog_divmod_pattern): Do both of the above.
14670 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
14673 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
14675 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
14676 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
14677 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
14678 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
14679 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
14680 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
14681 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
14682 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
14683 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
14684 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
14685 parameter with a single stmt_vec_info.
14686 (vect_recog_func_ptr): Likewise.
14687 (vect_recog_gather_scatter_pattern): Likewise, folding in...
14688 (vect_try_gather_scatter_pattern): ...this.
14689 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
14690 the stmt_vec_info of the statement to be matched. Don't clear
14691 STMT_VINFO_RELATED_STMT.
14692 (vect_pattern_recog): Update call accordingly.
14694 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
14696 PR tree-optimization/85694
14697 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
14698 (uavgM3_ceil): Document new optabs.
14699 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
14700 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
14702 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
14703 (savg_ceil_optab): New optabs.
14704 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
14705 (vect_vect_recog_func_ptrs): Add it.
14706 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
14707 constant directly from the associated lhs.
14709 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
14711 * tree-vect-patterns.c (vect_split_statement): New function.
14712 (vect_convert_input): Use it to try to split an existing cast.
14714 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
14716 * poly-int.h (print_hex): New function.
14717 * dumpfile.h (dump_dec, dump_hex): Declare.
14718 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
14719 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
14720 min_input_precision, operation_precision and operation_sign.
14721 * tree-vect-patterns.c (vect_get_range_info): New function.
14722 (vect_same_loop_or_bb_p, vect_single_imm_use)
14723 (vect_operation_fits_smaller_type): Delete.
14724 (vect_look_through_possible_promotion): Add an optional
14725 single_use_p parameter.
14726 (vect_recog_over_widening_pattern): Rewrite to use new
14727 stmt_vec_info infomration. Handle one operation at a time.
14728 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
14729 (vect_truncatable_operation_p, vect_set_operation_type)
14730 (vect_set_min_input_precision): New functions.
14731 (vect_determine_min_output_precision_1): Likewise.
14732 (vect_determine_min_output_precision): Likewise.
14733 (vect_determine_precisions_from_range): Likewise.
14734 (vect_determine_precisions_from_users): Likewise.
14735 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
14736 (vect_vect_recog_func_ptrs): Put over_widening first.
14738 (vect_pattern_recog): Call vect_determine_precisions.
14740 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
14742 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
14743 statements that have been replaced by further pattern statements.
14744 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
14746 2018-07-03 Richard Biener <rguenther@suse.de>
14748 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
14749 always set *dt. Dump vectype in vectype overload.
14750 * dumpfile.h (dump_gimple_expr): New function.
14751 (dump_gimple_expr_loc): Likewise.
14752 * dumpfile.c (dump_gimple_expr): New function.
14753 (dump_gimple_expr_loc): Likewise.
14755 2018-07-02 Jeff Law <law@redhat.com>
14757 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
14758 the H8/300, H8/300H and H8/S variants into a single pattern.
14759 (movhi_h8300, movqi_h8300hs): Similarly.
14760 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
14761 (QHI mode iterator): New.
14763 * config/h8300/h8300.md: Remove trailing whitespace.
14765 2018-07-02 Jim Wilson <jimw@sifive.com>
14767 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
14768 instead of emit_insn for interrupt returns.
14769 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
14770 (riscv_sret, riscv_uret): Likewise.
14772 2018-07-02 David Malcolm <dmalcolm@redhat.com>
14774 * pretty-print.c (selftest::test_pp_format): Move save and restore
14775 of quotes to class auto_fix_quotes, and add an instance.
14776 * selftest.c: Include "intl.h".
14777 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
14778 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
14779 * selftest.h (selftest::auto_fix_quotes): New class.
14781 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
14783 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
14784 (aarch64_sve_prepare_conditional_op): Remove.
14785 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
14786 Allow aarch64_simd_reg_or_zero as select operand; remove
14787 the aarch64_sve_prepare_conditional_op call.
14788 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
14789 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
14790 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
14791 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
14792 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
14793 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
14794 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
14795 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
14796 and a splitters to match all of the *_any patterns.
14797 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
14799 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
14800 (SVE_COND_FP_BINARY_REV): Remove.
14801 (sve_int_op_rev, sve_fp_op_rev): New.
14802 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
14803 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
14804 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
14805 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
14806 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
14807 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
14808 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
14809 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
14810 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
14812 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
14813 Remove match_dup 1 from the inner unspec.
14814 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
14816 * config/aarch64/aarch64.md (movprfx): New attr.
14817 (length): Default movprfx to 8.
14818 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
14819 (*madd<SVE_I>, *msub<SVE_I): Likewise.
14820 (*<su>mul<SVE_I>3_highpart): Likewise.
14821 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
14822 (*v<ASHIFT><SVE_I>3): Likewise.
14823 (*<su><MAXMIN><SVE_I>3): Likewise.
14824 (*<su><MAXMIN><SVE_F>3): Likewise.
14825 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
14826 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
14827 (*div<SVE_F>4): Likewise.
14829 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
14831 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
14834 2018-07-02 Richard Biener <rguenther@suse.de>
14836 PR tree-optimization/86363
14837 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
14838 memset argument refers to a non-variable address.
14840 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
14842 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
14843 {PLUS,MINUS}_EXPR code to...
14844 (adjust_symbolic_bound): ...here,
14845 (combine_bound): ...here,
14846 (set_value_range_with_overflow): ...and here.
14848 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
14850 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
14852 (extract_range_from_abs_expr): ...here.
14854 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
14856 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
14857 -fno-omit-frame-pointer when not optimizing.
14859 2018-07-02 Martin Liska <mliska@suse.cz>
14862 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
14863 (propagate_nothrow): Likewise.
14865 2018-07-02 Martin Liska <mliska@suse.cz>
14868 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
14870 2018-07-02 David Malcolm <dmalcolm@redhat.com>
14872 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
14873 function in r262149, changing "loc" param from source_location to
14874 const dump_location_t &.
14875 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
14876 declaration, as above.
14878 2018-07-01 Paul Koning <ni1d@arrl.net>
14880 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
14881 -munit-asm, -mgnu-asm, -mdec-asm.
14882 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
14883 (pdp11_output_labelref): New.
14884 (pdp11_output_def): New.
14885 (pdp11_output_addr_vec_elt): New.
14886 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
14887 %# and %@ format codes.
14888 (pdp11_option_override): New.
14889 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
14890 (pdp11_output_ident): New.
14891 (pdp11_asm_named_section): New.
14892 (pdp11_asm_init_sections): New.
14893 (pdp11_file_start): New.
14894 (pdp11_file_end): New.
14895 (output_ascii): Use .ascii/.asciz for -mdec-asm.
14896 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
14897 %o, like %c but octal.
14898 (pdp11_option_override): New.
14899 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
14901 (DATA_SECTION_ASM_OP): Ditto.
14902 (READONLY_DATA_SECTION_ASM_OP): New.
14903 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
14904 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
14905 (ASM_OUTPUT_LABELREF): Ditto.
14906 (ASM_OUTPUT_DEF): Ditto.
14907 (ASM_OUTPUT_EXTERNAL): New.
14908 (ASM_OUTPUT_SOURCE_FILENAME): New.
14909 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
14910 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
14911 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
14912 %# and %@ format codes.
14913 * config/pdp11/pdp11.opt (mgnu-asm): New.
14914 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
14915 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
14916 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
14918 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
14920 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
14921 dereferencing path[] beyond its length.
14923 (debug_all_paths): New.
14924 (rewire_first_differing_edge): New.
14925 (adjust_paths_after_duplication): New.
14926 (duplicate_thread_path): Call adjust_paths_after_duplication.
14928 (thread_through_all_blocks): Add new argument to
14929 duplicate_thread_path.
14931 2018-06-30 Jim Wilson <jimw@sifive.com>
14933 * config/riscv/predicates.md (p2m1_shift_operand): New.
14934 (high_mask_shift_operand): New.
14935 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
14936 pattern using p2m1_shift_operand.
14937 (lshsi3_zero_extend_3+2): New combiner pattern using
14938 high_mask_shift_operand.
14940 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
14942 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
14944 (vect_recog_rotate_pattern): ...here.
14945 (vect_convert_input): Try to insert casts of invariants in the
14947 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
14948 preheader to be empty.
14950 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
14952 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
14953 vector type. If given, install it in the new statement's
14954 STMT_VINFO_VECTYPE.
14955 (vect_element_precision): New function.
14956 (vect_unpromoted_value): New struct.
14957 (vect_unpromoted_value::vect_unpromoted_value): New function.
14958 (vect_unpromoted_value::set_op): Likewise.
14959 (vect_look_through_possible_promotion): Likewise.
14960 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
14961 (vect_widened_op_tree, vect_convert_input): Likewise.
14962 (vect_convert_inputs, vect_convert_output): Likewise.
14963 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
14964 to handle the optional cast of the multiplication result and
14965 vect_widened_op_tree to detect the widened multiplication itself.
14966 Do not require the input and output of promotion casts to have
14967 the same sign, but base the signedness of the operation on the
14968 input rather than the result. If the pattern includes two
14969 promotions, check that those promotions have the same sign.
14970 Do not restrict the MULT_EXPR handling to a double-width result;
14971 handle quadruple-width results and wider. Use vect_convert_inputs
14972 to convert the inputs to the common type.
14973 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
14974 to handle the optional cast of the ABS result. Also allow a sign
14975 change or a sign extension between the ABS and MINUS.
14976 Use vect_widened_op_tree to detect the widened subtraction and use
14977 vect_convert_inputs to convert the inputs to the common type.
14978 (vect_handle_widen_op_by_const): Delete.
14979 (vect_recog_widen_op_pattern): New function.
14980 (vect_recog_widen_mult_pattern): Use it.
14981 (vect_recog_widen_shift_pattern): Likewise.
14982 (vect_recog_widen_sum_pattern): Use
14983 vect_look_through_possible_promotion to handle the promoted
14986 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
14988 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
14989 the containing gimple_seq *.
14990 * gimple-iterator.h (gsi_for_stmt): Declare it.
14991 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
14992 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
14993 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
14994 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
14995 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
14997 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
14999 (vect_mark_pattern_stmts): ...here. Handle cases in which the
15000 statement being replaced is part of an existing pattern
15001 definition sequence, inserting the new pattern statements before
15003 (vect_pattern_recog_1): Don't return a bool. If the statement
15004 is already part of a pattern, instead apply pattern matching
15005 to the pattern definition statements. Don't clear the
15006 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
15007 (vect_pattern_recog): Don't break after the first match;
15008 continue processing the pattern definition statements instead.
15009 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
15011 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
15013 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
15014 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
15015 (vect_recog_widen_sum_pattern): Use it.
15017 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
15019 * tree-vect-loop.c (vectorizable_reduction): Assert that the
15020 phi is not a pattern statement and has not been replaced by
15021 a pattern statement.
15022 * tree-vect-patterns.c (type_conversion_p): Don't check
15023 STMT_VINFO_IN_PATTERN_P.
15024 (vect_recog_vector_vector_shift_pattern): Likewise.
15025 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
15026 the pattern statement rather than the original statement; check
15027 directly for a WIDEN_MULT_EXPR here.
15028 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
15029 vect_is_simple_use to return the pattern statement rather
15030 than the original statement; use is_pattern_stmt_p to check
15031 for such a pattern statement.
15032 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
15033 to return the pattern statement rather than the original statement;
15034 don't do the same transformation here.
15035 (vect_is_simple_use): If the defining statement has been replaced
15036 by a pattern statement, return the pattern statement instead.
15037 Remove the corresponding (local) transformation from the vectype
15040 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
15042 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
15043 end and default to null.
15044 * tree-vect-loop.c (vect_create_epilog_for_reduction)
15045 (vectorizable_reduction): Update calls accordingly, dropping the
15046 gimple ** argument if the passed-back statement isn't needed.
15047 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
15048 (vect_recog_rotate_pattern): Likewise.
15049 (vect_recog_mask_conversion_pattern): Likewise.
15050 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
15051 (vect_mask_constant_operand_p): Likewise.
15052 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
15053 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
15054 (get_group_load_store_type, get_load_store_type): Likewise.
15055 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
15056 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
15057 (vectorizable_conversion, vectorizable_assignment): Likewise.
15058 (vectorizable_shift, vectorizable_operation): Likewise.
15059 (vectorizable_store, vect_is_simple_cond): Likewise.
15060 (vectorizable_condition, vectorizable_comparison): Likewise.
15061 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
15062 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
15063 and move it to the end. Cope with null def_stmt_outs.
15065 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
15067 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
15069 2018-06-29 Jeff Law <law@redhat.com>
15071 * config/v850/v850.c (v850_legitimate_address_p): Handle large
15072 displacements for TARGET_V850E2V3 and newer.
15073 (TARGET_LRA_P): Remove. Defaults to LRA now.
15074 * config/v850/v850.md (sign23byte_load): Remove.
15075 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
15076 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
15078 2018-06-29 Martin Liska <mliska@suse.cz>
15081 * coverage.c (coverage_init): Mangle full path name.
15082 * doc/invoke.texi: Document the change.
15083 * gcov-io.c (mangle_path): New.
15084 * gcov-io.h (mangle_path): Likewise.
15085 * gcov.c (mangle_name): Use mangle_path for path mangling.
15087 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15089 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
15090 if starting source register is not even.
15092 2018-06-29 Martin Liska <mliska@suse.cz>
15094 PR tree-optimization/86263
15095 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
15096 Make edge redirection.
15098 2018-06-29 David Malcolm <dmalcolm@redhat.com>
15100 * dumpfile.c (dump_loc): Add indentation based on scope depth.
15101 (dump_scope_depth): New variable.
15102 (get_dump_scope_depth): New function.
15103 (dump_begin_scope): New function.
15104 (dump_end_scope): New function.
15105 * dumpfile.h (get_dump_scope_depth): New declaration.
15106 (dump_begin_scope): New declaration.
15107 (dump_end_scope): New declaration.
15108 (class auto_dump_scope): New class.
15109 (AUTO_DUMP_SCOPE): New macro.
15110 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
15113 2018-06-29 Richard Biener <rguenther@suse.de>
15115 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
15116 compute_all_dependences succeeds.
15117 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
15118 exceed --param loop-max-datarefs-for-datadeps.
15120 2018-06-29 Jakub Jelinek <jakub@redhat.com>
15122 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
15124 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
15127 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
15128 alternative 0 in preferred_for_speed attribute.
15130 2018-06-28 Paul Koning <ni1d@arrl.net>
15132 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
15133 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
15134 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
15135 * config/pdp11/pdp11.md: Correct "length" attribute calculation
15136 for shift insn patterns.
15138 2018-06-28 David Malcolm <dmalcolm@redhat.com>
15140 * cgraph.c (cgraph_node::get_body): Replace assignments to
15141 "dump_file" with calls to set_dump_file.
15142 * dumpfile.c (alt_dump_file): Make static, and group with...
15143 (alt_flags): ...this definition.
15144 (dumps_are_enabled): New variable.
15145 (refresh_dumps_are_enabled): New function.
15146 (set_dump_file): New function.
15147 (set_alt_dump_file): New function.
15148 (gcc::dump_manager::dump_start): Replace assignments to
15149 "dump_file" and "alt_dump_file" with calls to set_dump_file and
15151 (gcc::dump_manager::dump_finish): Likewise.
15152 * dumpfile.h (alt_dump_file): Delete decl.
15153 (dumps_are_enabled): New variable decl.
15154 (set_dump_file): New function decl.
15155 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
15157 * tree-nested.c (lower_nested_functions): Replace assignments to
15158 "dump_file" with calls to set_dump_file.
15160 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
15162 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
15163 goto_locus of each outgoing edge of each basic block.
15165 2018-06-28 Richard Biener <rguenther@suse.de>
15167 * dwarf2out.c (decl_scope_table): Remove.
15168 (push_decl_scope): Likewise.
15169 (pop_decl_scope): Likewise.
15170 (gen_type_die_for_member): Do not call push/pop_decl_scope.
15171 (gen_struct_or_union_type_die): Likewise.
15172 (gen_tagged_type_die): Likewise.
15173 (dwarf2out_init): Do not initialize decl_scope_table.
15174 (dwarf2out_c_finalize): Do not free it.
15176 2018-06-28 Richard Biener <rguenther@suse.de>
15178 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
15179 deciding whether to not re-use a DIE.
15181 2018-06-28 Richard Biener <rguenther@suse.de>
15183 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
15184 DW_AT_abstract_origin attribute.
15186 2018-06-28 Martin Liska <mliska@suse.cz>
15188 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
15189 Use newly introduced constants.
15190 * tree-switch-conversion.h (struct jump_table_cluster):
15191 Define max_ratio_for_size and max_ratio_for_speed.
15193 2018-06-28 Martin Liska <mliska@suse.cz>
15195 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
15196 Add new checking assert to catch invalid state.
15197 (jump_table_cluster::can_be_handled): Handle single case
15199 (jump_table_cluster::is_beneficial): Bail out for such case.
15200 (bit_test_cluster::find_bit_tests):
15201 Add new checking assert to catch invalid state.
15202 (bit_test_cluster::can_be_handled): Handle single case
15204 (bit_test_cluster::is_beneficial): Bail out for such case.
15205 (switch_decision_tree::analyze_switch_statement):
15208 2018-06-28 Martin Liska <mliska@suse.cz>
15210 * common.opt: Introduce -completion option.
15211 * gcc.c (driver_handle_option): Handle it.
15212 (driver::main): Print completions if completion
15214 * opt-suggestions.c (option_proposer::get_completions):
15216 (option_proposer::suggest_completion): Likewise.
15217 (option_proposer::find_param_completions): Likewise.
15218 (verify_autocompletions): Likewise.
15219 (test_completion_valid_options): Likewise.
15220 (test_completion_valid_params): Likewise.
15221 (in_completion_p): Likewise.
15222 (empty_completion_p): Likewise.
15223 (test_completion_partial_match): Likewise.
15224 (test_completion_garbage): Likewise.
15225 (opt_proposer_c_tests): Likewise.
15226 * opt-suggestions.h: Declare new functions.
15227 * opts.c (common_handle_option): Handle OPT__completion_.
15228 * selftest-run-tests.c (selftest::run_tests): Add
15229 opt_proposer_c_tests.
15230 * selftest.c (assert_str_startswith): New.
15231 * selftest.h (assert_str_startswith): Likewise.
15232 (opt_proposer_c_tests): New.
15233 (ASSERT_STR_STARTSWITH): Likewise.
15235 2018-06-28 Martin Liska <mliska@suse.cz>
15237 * Makefile.in: Add opt-suggestions.o.
15238 * gcc-main.c: Include opt-suggestions.h.
15239 * gcc.c (driver::driver): Likewise.
15240 (driver::~driver): Remove m_option_suggestions.
15241 (driver::build_option_suggestions): Moved to option_proposer.
15242 (driver::suggest_option): Likewise.
15243 (driver::handle_unrecognized_options): Use option_proposer.
15244 * gcc.h (class driver): Add new memver m_option_proposer.
15245 * opt-suggestions.c: New file.
15246 * opt-suggestions.h: New file.
15248 2018-06-28 Martin Liska <mliska@suse.cz>
15250 * vec.h (class auto_string_vec): New (moved from auto_argvec).
15251 (auto_string_vec::~auto_string_vec): Likewise.
15253 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
15255 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
15256 prevent_decl_creation_for_types fields up and add reset_location field.
15257 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
15258 statement if id->reset_location is true.
15259 (copy_edges_for_bb): Do not set goto_locus on the new edges if
15260 id->reset_location is true.
15261 (copy_phis_for_bb): Force input_location on the arguments if
15262 id->reset_location is true.
15263 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
15264 is set on the function to be inlined.
15266 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
15268 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
15270 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
15272 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
15273 registers for Pmode.
15274 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
15275 hard registers for the clobbered pseudo.
15277 2018-06-27 Paul Koning <ni1d@arrl.net>
15279 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
15280 mutually exclusive options.
15281 * config/pdp11/constraints.md (h): New constraint.
15282 (O): Update definition to match shift code generation.
15283 (D): New constraint.
15284 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
15286 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
15288 (output_jump): Change arguments.
15289 (pdp11_fixed_cc_regs): New function.
15290 (pdp11_cc_mode): Ditto.
15291 (pdp11_expand_shift): Ditto.
15292 (pdp11_assemble_shift): Ditto.
15293 (pdp11_small_shift): Ditto.
15294 (pdp11_branch_cost): Remove.
15295 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
15297 (pdp11_register_move_cost): Update for CC registers.
15298 (pdp11_rtx_costs): Add case for LSHIFTRT.
15299 (pdp11_output_jump): Add CCNZ mode conditional branches.
15300 (notice_update_cc_on_set): Remove.
15301 (pdp11_cc_mode): New function.
15302 (simple_memory_operand): Correct pre/post decrement case.
15303 (no_side_effect_operand): New function.
15304 (pdp11_regno_reg_class): Add CC_REGS class.
15305 (pdp11_fixed_cc_regs): New function.
15306 (pdp11_small_shift): New function.
15307 (pdp11_expand_shift): New function to expand shift insns.
15308 (pdp11_assemble_shift): New function to output shifts.
15309 (pdp11_branch_cost): Remove.
15310 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
15311 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
15312 (WCHAR_TYPE): Ditto.
15313 (PTRDIFF_TYPE): Ditto.
15314 (ADJUST_INSN_LENGTH): New macro.
15315 (FIXED_REGISTERS): Add CC registers.
15316 (CALL_USED_REGISTERS): Ditto.
15317 (reg_class): Ditto.
15318 (REG_CLASS_NAMES): Ditto.
15319 (REG_CLASS_CONTENTS): Ditto.
15320 (SELECT_CC_MODE): Use new function.
15321 (TARGET_FLAGS_REGNUM): New macro.
15322 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
15323 (cc0_reg_rtx): Remove.
15324 (CC_STATUS_MDEP): Remove.
15325 (CC_STATUS_MDEFP_INIT): Remove.
15326 (CC_IN_FPU): Remove.
15327 (NOTICE_UPDATE_CC): Remove.
15328 (REGISTER_NAMES): Add CC registers.
15329 (BRANCH_COST): Change to constant 1.
15330 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
15332 * config/pdp11/pdp11.opt (mbcopy): Remove.
15333 (mbcopy-builtin): Remove.
15334 (mbranch-cheap): Remove.
15335 (mbranch-expensive): Remove.
15336 * config/pdp11/predicates.md (expand_shift_operand): Update to
15337 match shift code generation.
15338 (ccnz_operator): New predicate.
15339 * doc/invoke.texi (PDP-11 Options): Remove deleted options
15340 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
15341 Remove non-existent option -mabshi, -mno-abshi. Document mutually
15343 * doc/md.texi (PDP-11): Document new D and h constraints. Update
15344 description of O constraint.
15346 2018-06-27 Jeff Law <law@redhat.com>
15347 Austin Law <austinklaw@gmail.com>
15349 * config/v850/v850.md (addsi3_set_flags): New pattern.
15350 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
15351 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
15352 (zero_extendhisi2_v850_set_flags): Likewise.
15353 (zero_extendqisi2_v850_set_flags): Likewise.
15354 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
15355 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
15356 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
15358 * config/v850/v850-protos.h (notice_update_cc): Remove.
15359 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
15360 (v850_print_operand): Handle 'D' and "d".
15361 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
15362 Add handling of arithmetic/logical operations compared against zero.
15363 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
15364 Do not look at v850_compare_op, instead get mode from last argument.
15365 (v850_gen_compare): Remove
15366 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
15367 after reload for prologue insns.
15368 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
15370 (construct_save_jarl): Likewise.
15371 (TARGET_FLAGS_REGNUM): Define.
15372 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
15373 (NOTICE_UPDATE_CC): Remove.
15374 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
15375 than cc0. Conditionalize on reload_completed.
15376 (cmpsi_insn, setfcc_insn): Likewise.
15377 (tst1 splitter): Turn into define_and_split which sets the flags
15379 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
15380 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
15381 (cstoresf4, cstoredf4): Clobber the flags.
15382 (cmpsi, cmpsf, cmpdf): Remove expanders.
15383 (setf_insn): Remove pattern.
15384 (addsi3): Turn into define_and_split which clobbers the flags after
15385 reload and a suitable pattern (addsi3_clobber_flags) for use after
15387 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
15388 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
15389 (ashrsi3, ashrsi3_v850e2): Likewise.
15390 (bins): Clobber the flags.
15391 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
15392 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
15393 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
15394 (call_value_internal_short, call_value_internal_long): Likewise.
15395 (callt_save_interrupt, callt_return_interrupt): Likewise.
15396 (save_interrupt, return_interrupt): Likewise.
15397 (callt_save_all_interrupt, save_all_interrupt): Likewise.
15398 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
15399 (restore_all_interrupt, _restore_all_interrupt): Likewise.
15400 (All FP comparisons): Only allow after reload has completed.
15402 (divh, divhu): Tweak output template.
15403 (branch_z_normal, branch_z_invert): Remove
15404 (branch_nz_normal, branch_nz_invert): Likewise.
15405 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
15407 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
15408 * config/v850/v850.c (notice_update_cc): Remove.
15409 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
15410 (CC_NO_CARRY): Likewise.
15411 (NOTICE_UPDATE_CC): Define to nothing.
15412 * config/v850/v850.md: Remove block comment on cc0 handling
15413 Remove "cc" attribute from all patterns. Remove cc_status handling
15414 from all patterns. Minor formatting fixes.
15416 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15418 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
15419 (cortex-a76.cortex-a55): Likewise.
15420 * config/aarch64/aarch64-tune.md: Regenerate.
15421 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
15422 cortex-a76.cortex-a55.
15424 2018-06-27 Jeff Law <law@redhat.com>
15426 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
15427 (MULTILIB_DIRNAMES): Similarly.
15429 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
15431 * gimple.h (gimple_return_retbnd): Delete.
15432 (gimple_return_set_retbnd): Likewise.
15433 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
15434 gimple_return_set_retbnd.
15435 * gimple-pretty-print.c (dump_gimple_return): Remove call to
15436 gimple_return_retbnd and adjust.
15437 * tree-inline.h (struct copy_body_data): Remove retbnd field.
15438 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
15439 Explicitly return NULL in a couple more cases. Move assertion
15440 on debug statements and remove unreachable code.
15441 (reset_debug_binding): Do not test id->retbnd.
15442 (expand_call_inline): Do not set it.
15444 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
15446 * configure.ac: Add --disable-gcov option.
15447 * configure: Regenerate.
15448 * Makefile.in: Honour @enable_gcov@.
15449 * doc/install.texi: Document --disable-gcov.
15451 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15453 * config/arm/arm-cpus.in (cortex-a76): New entry.
15454 (cortex-a76.cortex-a55): Likewise.
15455 * config/arm/arm-tables.opt: Regenerate.
15456 * config/arm/arm-tune.md: Likewise.
15457 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
15458 * doc/invoke.texi (ARM Options): Document cortex-a76 and
15459 cortex-a76.cortex-a55.
15461 2018-06-27 Tamar Christina <tamar.christina@arm.com>
15464 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
15466 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
15468 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
15470 (EPILOGUE_USES): Likewise.
15472 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
15474 * tree-inline.c (remap_location): New function extracted from...
15475 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
15476 (copy_phis_for_bb): ...here. Call remap_location.
15477 (copy_cfg_body): Adjust call to copy_edges_for_bb.
15479 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
15481 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
15482 unaligned vsx for 16B memset.
15484 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
15487 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
15488 ieee128_float_type_node to long_double_type_node unless
15489 TARGET_LONG_DOUBLE_128 is set.
15491 2018-06-26 David Malcolm <dmalcolm@redhat.com>
15493 * cfgloop.c (get_loop_location): Convert return type from
15494 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
15495 by implicit construction from rtx_insn *, and using
15496 dump_user_location_t::from_function_decl for the fallback case.
15497 * cfgloop.h (get_loop_location): Convert return type from
15498 location_t to dump_user_location_t.
15499 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
15500 dump_printf_loc to pass in a dump_location_t rather than a
15501 location_t, via the gimple stmt.
15502 * coverage.c (get_coverage_counts): Update calls to
15503 dump_printf_loc to pass in dump_location_t rather than a
15505 * doc/optinfo.texi (Dump types): Convert example of
15506 dump_printf_loc from taking "locus" to taking "insn". Update
15507 description of the "_loc" calls to cover dump_location_t.
15508 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
15510 (dump_user_location_t::dump_user_location_t): New constructors,
15511 from gimple *stmt and rtx_insn *.
15512 (dump_user_location_t::from_function_decl): New function.
15513 (dump_loc): Make static.
15514 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
15515 const dump_location_t &.
15516 (dump_generic_expr_loc): Delete.
15517 (dump_printf_loc): Convert param "loc" from location_t to
15518 const dump_location_t &.
15519 (selftest::test_impl_location): New function.
15520 (selftest::dumpfile_c_tests): New function.
15521 * dumpfile.h: Include "profile-count.h".
15522 (class dump_user_location_t): New class.
15523 (struct dump_impl_location_t): New struct.
15524 (class dump_location_t): New class.
15525 (dump_printf_loc): Convert 2nd param from source_location to
15526 const dump_location_t &.
15527 (dump_generic_expr_loc): Delete.
15528 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
15529 const dump_location_t &.
15530 * gimple-fold.c (fold_gimple_assign): Update call to
15531 dump_printf_loc to pass in a dump_location_t rather than a
15532 location_t, via the gimple stmt.
15533 (gimple_fold_call): Likewise.
15534 * gimple-loop-interchange.cc
15535 (loop_cand::analyze_iloop_reduction_var): Update for change to
15536 check_reduction_path.
15537 (tree_loop_interchange::interchange): Update for change to
15538 find_loop_location.
15539 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
15540 change in return-type of find_loop_location.
15541 (graphite_regenerate_ast_isl): Likewise.
15542 * graphite-optimize-isl.c (optimize_isl): Likewise.
15543 * graphite.c (graphite_transform_loops): Likewise.
15544 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
15545 pass in a dump_location_t rather than a location_t, via the
15547 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
15548 * ipa.c (walk_polymorphic_call_targets): Likewise.
15549 * loop-unroll.c (report_unroll): Convert "locus" param from
15550 location_t to dump_location_t.
15551 (decide_unrolling): Update for change to get_loop_location's
15553 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
15554 location_t to dump_user_location_t.
15555 (grid_find_single_omp_among_assignments_1): Updates calls to
15556 dump_printf_loc to pass in a dump_location_t rather than a
15557 location_t, via the gimple stmt.
15558 (grid_parallel_clauses_gridifiable): Convert "tloc" from
15559 location_t to dump_location_t. Updates calls to dump_printf_loc
15560 to pass in a dump_location_t rather than a location_t, via the
15562 (grid_inner_loop_gridifiable_p): Likewise.
15563 (grid_dist_follows_simple_pattern): Likewise.
15564 (grid_gfor_follows_tiling_pattern): Likewise.
15565 (grid_target_follows_gridifiable_pattern): Likewise.
15566 (grid_attempt_target_gridification): Convert initialization
15567 of local "grid" from memset to zero-initialization; FIXME: does
15568 this require C++11? Update call to dump_printf_loc to pass in a
15569 optinfo_location rather than a location_t, via the gimple stmt.
15570 * profile.c (read_profile_edge_counts): Updates call to
15571 dump_printf_loc to pass in a dump_location_t rather than a
15573 (compute_branch_probabilities): Likewise.
15574 * selftest-run-tests.c (selftest::run_tests): Call
15576 * selftest.h (dumpfile_c_tests): New decl.
15577 * tree-loop-distribution.c (pass_loop_distribution::execute):
15578 Update for change in return type of find_loop_location.
15579 * tree-parloops.c (parallelize_loops): Likewise.
15580 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
15581 "locus" from location_t to dump_user_location_t.
15582 (canonicalize_loop_induction_variables): Likewise.
15583 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
15584 for change in return type of find_loop_location.
15585 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
15586 to dump_printf_loc to pass in a dump_location_t rather than a
15587 location_t, via the stmt.
15588 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
15590 * tree-vect-loop-manip.c (find_loop_location): Convert return
15591 type from source_location to dump_user_location_t.
15592 (vect_do_peeling): Update for above change.
15593 (vect_loop_versioning): Update for change in type of
15595 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
15596 from location_t to dump_user_location_t.
15597 (vect_estimate_min_profitable_iters): Update for change in type
15599 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
15600 location_t to dump_location_t.
15601 (vect_slp_bb): Update for change in type of vect_location.
15602 * tree-vectorizer.c (vect_location): Convert from source_location
15603 to dump_user_location_t.
15604 (try_vectorize_loop_1): Update for change in vect_location's type.
15605 (vectorize_loops): Likewise.
15606 (increase_alignment): Likewise.
15607 * tree-vectorizer.h (vect_location): Convert from source_location
15608 to dump_user_location_t.
15609 (find_loop_location): Convert return type from source_location to
15610 dump_user_location_t.
15611 (check_reduction_path): Convert 1st param from location_t to
15612 dump_user_location_t.
15613 * value-prof.c (check_counter): Update call to dump_printf_loc to
15614 pass in a dump_user_location_t rather than a location_t; update
15615 call to error_at for change in type of "locus".
15616 (check_ic_target): Update call to dump_printf_loc to
15617 pass in a dump_user_location_t rather than a location_t, via the
15620 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
15622 * config/s390/s390.h (enum processor_flags): Do not use
15623 default tune parameter when -march was specified.
15625 2018-06-26 Jakub Jelinek <jakub@redhat.com>
15628 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
15629 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
15632 2018-06-26 Richard Biener <rguenther@suse.de>
15634 PR tree-optimization/86287
15636 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
15637 (vect_analyze_loop): Initialize n_stmts.
15639 2018-06-26 Richard Biener <rguenther@suse.de>
15641 PR middle-end/86271
15642 * fold-const.c (fold_convertible_p): Pointer extension
15645 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
15648 * dwarf2out.c (loc_list_has_views): Adjust comments.
15649 (dw_loc_list): Split single cross-partition range with
15652 2018-06-25 Jeff Law <law@redhat.com>
15654 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
15655 on -mbig-switch by default.
15657 * config/v850/predicates.md (const_float_1_operand): Fix match_code
15659 (const_float_0_operand): Remove unused predicate.
15660 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
15661 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
15662 (recipsf2): New expander. Original pattern now called
15664 (recipdf2, recipdf2_insn): Similarly.
15665 (rsqrtsf2, rsqrtsf2_insn): Similarly
15666 (rsqrtdf2, rsqrtdf2_insn): Similarly
15668 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
15670 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
15671 Simplify logic for FreeBSD (twice).
15673 2018-06-25 Martin Sebor <msebor@redhat.com>
15675 PR tree-optimization/86204
15676 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
15677 a strnlen result if it's less than the length of the string.
15679 2018-06-25 Martin Sebor <msebor@redhat.com>
15681 PR tree-optimization/85700
15682 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
15683 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
15684 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
15686 2018-06-25 Martin Sebor <msebor@redhat.com>
15688 * doc/extend.texi (Zero-length arrays): Update and clarify.
15690 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
15692 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
15693 added IEEE/IBM long double multilib support on PowerPC little
15694 endian Linux systems.
15695 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
15696 (MULTILIB_DEFAULTS): Likewise.
15697 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15699 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
15700 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
15701 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
15703 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
15705 PR middle-end/86311
15706 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
15707 (REORDER_45): Likewise.
15709 2018-06-25 Jeff Law <law@redhat.com>
15711 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
15712 dividend to 32 bits. Adjust length.
15713 (udivmodhi4): Cleanup output template. Fix length.
15715 2018-06-25 Carl Love <cel@us.ibm.com>
15717 * config/rs6000/vsx.md: Change word selector to prefered location.
15719 2018-06-25 Richard Biener <rguenther@suse.de>
15721 PR tree-optimization/86304
15722 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
15723 epilogue-if-converted loops as well.
15725 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
15727 * lto-section-out.c (lto_begin_section): Do not print section
15728 name for noaddr and unnumbered dumps.
15730 2018-06-25 Richard Biener <rguenther@suse.de>
15732 * tree-vectorizer.h (struct vec_info_shared): New structure
15733 with parts split out from struct vec_info and loop_nest from
15734 struct _loop_vec_info.
15735 (struct vec_info): Adjust accordingly.
15736 (struct _loop_vec_info): Likewise.
15737 (LOOP_VINFO_LOOP_NEST): Adjust.
15738 (LOOP_VINFO_DATAREFS): Likewise.
15739 (LOOP_VINFO_DDRS): Likewise.
15740 (struct _bb_vec_info): Likewise.
15741 (BB_VINFO_DATAREFS): Likewise.
15742 (BB_VINFO_DDRS): Likewise.
15743 (struct _stmt_vec_info): Add dr_aux member.
15744 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
15745 (DR_MISALIGNMENT_UNINITIALIZED): New.
15746 (set_dr_misalignment): Adjust.
15747 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
15748 (vect_analyze_loop): Adjust prototype.
15749 (vect_analyze_loop_form): Likewise.
15750 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
15751 Compute dependences lazily.
15752 (vect_record_base_alignments): Use shared datarefs/ddrs.
15753 (vect_verify_datarefs_alignment): Likewise.
15754 (vect_analyze_data_refs_alignment): Likewise.
15755 (vect_analyze_data_ref_accesses): Likewise.
15756 (vect_analyze_data_refs): Likewise.
15757 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
15758 constructor parameter for shared part.
15759 (vect_analyze_loop_form): Pass in shared part and adjust.
15760 (vect_analyze_loop_2): Pass in storage for the number of
15761 stmts. Move loop nest finding to the caller. Compute
15763 (vect_analyze_loop): Pass in shared part.
15764 (vect_transform_loop): Verify shared datarefs are unchanged.
15765 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
15766 constructor parameter for shared part.
15767 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
15768 (vect_slp_bb): Verify shared datarefs are unchanged before
15770 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
15772 (new_stmt_vec_info): Initialize DR_AUX misalignment to
15773 DR_MISALIGNMENT_UNINITIALIZED.
15774 * tree-vectorizer.c (vec_info::vec_info): Add constructor
15775 parameter for shared part.
15776 (vec_info::~vec_info): Adjust.
15777 (vec_info_shared::vec_info_shared): New.
15778 (vec_info_shared::~vec_info_shared): Likewise.
15779 (vec_info_shared::save_datarefs): Likewise.
15780 (vec_info_shared::check_datarefs): Likewise.
15781 (try_vectorize_loop_1): Construct shared part live for analyses
15782 of a single loop for multiple vector sizes.
15783 * tree-parloops.c (gather_scalar_reductions): Adjust.
15785 2018-06-25 Richard Biener <rguenther@suse.de>
15787 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
15788 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
15789 (vect_analyze_data_refs): Remove similar code from here and
15790 simplify accordingly.
15792 2018-06-25 Richard Biener <rguenther@suse.de>
15794 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
15795 for reverse storage order accesses rather than asserting
15796 they cannot happen here.
15798 2018-06-25 Tom de Vries <tdevries@suse.de>
15801 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
15802 Use data16 instead of .byte for insn prefix.
15804 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
15807 * parser.c (make_char_string_pack): Pass this literal chars
15808 through cpp_interpret_string.
15809 (cp_parser_userdef_numeric_literal): Check the result of
15810 make_char_string_pack.
15812 2018-06-24 Maya Rashish <coypu@sdf.org>
15814 * ginclude/stddef.h: Simplify conditions around avoiding
15815 re-definition of __size_t.
15817 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
15819 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
15820 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
15822 2018-06-22 Maya Rashish <coypu@sdf.org>
15824 * doc/invoke.texi (mno-fancy-math-387): Update for changes
15825 made to OpenBSD and NetBSD through the years.
15827 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
15829 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
15830 behavior of vec_pack (vector double, vector double) to match
15831 behavior of vec_float2 (vector double, vector double).
15833 2018-06-22 Olivier Hainque <hainque@adacore.com>
15835 * gimplify.c (gimplify_function_tree): Prevent creation
15836 of a trampoline for the address of the current function
15837 passed to entry/exit instrumentation hooks.
15839 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
15842 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
15845 2018-06-22 Martin Liska <mliska@suse.cz>
15847 PR tree-optimization/86263
15848 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
15849 Bail out if is_enabled is false.
15850 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
15852 (jump_table_cluster::is_enabled): New function.
15854 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
15856 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
15857 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
15858 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
15859 (lto_input_ts_binfo_tree_pointers): Likewise.
15860 * tree-streamer-out.c (streamer_write_tree_bitfields,
15861 write_ts_binfo_tree_pointers): Likewise.
15862 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
15864 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
15866 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
15868 2018-06-22 Martin Liska <mliska@suse.cz>
15870 * symbol-summary.h (get): Make it pure and inline move
15871 functionality from ::get function.
15872 (get): Remove and inline into ::get and ::get_create.
15873 (get_create): Move code from ::get function.
15875 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15878 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
15879 -x assembler-with-cpp.
15881 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15883 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
15884 _FILE_OFFSET_BITS=64 for C++.
15886 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
15888 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
15889 conversion insn that shows up when pr85657-3.c is compiled using
15890 IEEE 128-bit long double.
15891 (neg<mode>2_internal): Use the correct mode to check whether the
15892 mode is IBM extended.
15893 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
15894 multiply and divide external functions from being created more
15897 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
15899 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
15901 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
15902 the edge can be forwarded.
15903 (cfg_layout_merge_blocks): Likewise.
15905 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
15907 * except.c (finish_eh_generation): Commit edge insertions only after
15908 the EH edges have been redirected from post-landing to landing pads.
15910 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
15912 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
15913 create_tmp_var_for to create the FRAME decl.
15914 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
15916 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
15918 * tree-inline.c (copy_edges_for_bb): Minor tweak.
15919 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
15920 debug statement when resetting its value.
15921 (expand_call_inline): Copy the locus of the call onto the assignment
15922 of the return value, if any. Use local variable in more cases.
15924 2018-06-21 Martin Liska <mliska@suse.cz>
15926 * ipa-pure-const.c (propagate_nothrow): Use
15927 funct_state_summaries->get.
15928 (dump_malloc_lattice): Likewise.
15929 (propagate_malloc): Likewise.
15931 2018-06-21 Richard Biener <rguenther@suse.de>
15933 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
15934 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
15935 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
15937 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
15938 BLOCK_ABSTRACT_ORIGIN unconditionally.
15940 2018-06-21 David Malcolm <dmalcolm@redhat.com>
15942 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
15944 * ipa-reference.c (ipa_reference_c_finalize): Delete
15945 ipa_ref_opt_sum_summaries and set it to NULL.
15947 2018-06-21 Tom de Vries <tdevries@suse.de>
15949 PR tree-optimization/85859
15950 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
15951 test with comment from bb_no_side_effects_p.
15953 2018-06-21 Richard Biener <rguenther@suse.de>
15955 PR tree-optimization/86232
15956 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
15957 max for constant niter.
15959 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
15961 * config/aarch64/aarch64-simd.md
15962 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
15964 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
15966 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
15967 Make opernads of the unspec commutative.
15969 2018-06-21 Richard Biener <rguenther@suse.de>
15971 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
15972 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15973 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
15974 (vect_analyze_data_ref_dependence): Re-order checks to deal with
15976 (vect_record_base_alignments): Do not record base alignment
15977 for gathers or scatters.
15978 (vect_compute_data_ref_alignment): Drop return value that is always
15979 true. Bail out early for gathers or scatters.
15980 (vect_enhance_data_refs_alignment): Bail out early for gathers
15982 (vect_find_same_alignment_drs): Likewise.
15983 (vect_analyze_data_refs_alignment): Remove dead code.
15984 (vect_slp_analyze_and_verify_node_alignment): Likewise.
15985 (vect_analyze_data_refs): For possible gathers or scatters do
15986 not create an alternate DR, just check their possible validity
15987 and mark them. Adjust DECL_NONALIASED handling to not rely
15988 on DR_BASE_ADDRESS.
15989 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
15990 update inits of gathers or scatters.
15991 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
15992 Also copy gather/scatter flag to pattern vinfo.
15994 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
15996 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
15997 behavior of vec_packsu (vector unsigned long long, vector unsigned
15998 long long) to match behavior of vec_packs with same signature.
16000 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
16001 Thomas Schwinge <thomas@codesourcery.com>
16002 Cesar Philippidis <cesar@codesourcery.com>
16004 * gimplify.c (gimplify_scan_omp_clauses): Add support for
16005 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
16006 (gimplify_adjust_omp_clauses): Likewise.
16007 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
16008 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
16009 (gimplify_omp_target_update): Update handling of acc update and
16011 * omp-low.c (install_var_field): Remove unused parameter
16012 base_pointers_restrict.
16013 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
16014 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
16016 (omp_target_base_pointers_restrict_p): Delete.
16017 (scan_omp_target): Update call to scan_sharing_clauses.
16018 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
16020 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
16021 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
16022 (convert_local_omp_clauses): Likewise.
16023 * tree-pretty-print.c (dump_omp_clause): Likewise.
16024 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
16026 (omp_clause_code_name): Likewise.
16028 2018-06-20 Jakub Jelinek <jakub@redhat.com>
16031 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
16034 PR tree-optimization/86231
16035 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
16036 anti-range don't overwrite *vr0min before using it to compute *vr0max.
16038 2018-06-20 Tom de Vries <tdevries@suse.de>
16040 PR tree-optimization/86097
16041 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
16042 iv type if signedness of iv type is not the same as that of *nit.
16044 2018-06-20 Jakub Jelinek <jakub@redhat.com>
16046 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
16047 EDGE_EH edges, verify they are all EDGE_EH.
16049 2018-06-20 Maya Rashish <coypu@sdf.org>
16051 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
16053 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16055 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
16056 * config/aarch64/aarch64.c (xgene1_tunings): Add
16057 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
16058 (aarch64_mode_valid_for_sched_fusion_p):
16059 Allow 16-byte modes.
16060 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
16061 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
16063 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
16065 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
16066 * config/aarch64/iterators.md (VQ2): New mode iterator.
16068 2018-06-20 Martin Liska <mliska@suse.cz>
16070 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
16071 Change default ratio from 10 to 8.
16073 2018-06-20 Martin Liska <mliska@suse.cz>
16075 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
16077 (bit_test_cluster::find_bit_tests): Likewise.
16078 (switch_decision_tree::analyze_switch_statement): Find clusters.
16079 * tree-switch-conversion.h (struct jump_table_cluster): Document
16082 2018-06-20 Martin Liska <mliska@suse.cz>
16084 * tree-switch-conversion.c (switch_conversion::collect):
16085 Record m_uniq property.
16086 (switch_conversion::expand): Bail out for special conditions.
16087 (group_cluster::~group_cluster): New.
16088 (group_cluster::group_cluster): Likewise.
16089 (group_cluster::dump): Likewise.
16090 (jump_table_cluster::emit): New.
16091 (switch_decision_tree::fix_phi_operands_for_edges): New.
16092 (struct case_node): Remove struct.
16093 (jump_table_cluster::can_be_handled): New.
16094 (case_values_threshold): Moved to header.
16095 (reset_out_edges_aux): Likewise.
16096 (jump_table_cluster::is_beneficial): New.
16097 (bit_test_cluster::can_be_handled): Likewise.
16098 (add_case_node): Remove.
16099 (bit_test_cluster::is_beneficial): New.
16100 (case_bit_test::cmp): New.
16101 (bit_test_cluster::emit): New.
16102 (expand_switch_as_decision_tree_p): Remove.
16103 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
16104 (fix_phi_operands_for_edge): Likewise.
16105 (switch_decision_tree::analyze_switch_statement): New.
16106 (compute_cases_per_edge): Move ...
16107 (switch_decision_tree::compute_cases_per_edge): ... here.
16108 (try_switch_expansion): Likewise.
16109 (switch_decision_tree::try_switch_expansion): Likewise.
16110 (record_phi_operand_mapping): Likewise.
16111 (switch_decision_tree::record_phi_operand_mapping): Likewise.
16112 (emit_case_decision_tree): Likewise.
16113 (switch_decision_tree::emit): Likewise.
16114 (balance_case_nodes): Likewise.
16115 (switch_decision_tree::balance_case_nodes): Likewise.
16116 (dump_case_nodes): Likewise.
16117 (switch_decision_tree::dump_case_nodes): Likewise.
16118 (emit_jump): Likewise.
16119 (switch_decision_tree::emit_jump): Likewise.
16120 (emit_cmp_and_jump_insns): Likewise.
16121 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
16122 (emit_case_nodes): Likewise.
16123 (switch_decision_tree::emit_case_nodes): Likewise.
16124 (conditional_probability): Remove.
16125 * tree-switch-conversion.h (enum cluster_type): New.
16127 (struct cluster): Likewise.
16128 (cluster::cluster): Likewise.
16129 (struct simple_cluster): Likewise.
16130 (simple_cluster::simple_cluster): Likewise.
16131 (struct group_cluster): Likewise.
16132 (struct jump_table_cluster): Likewise.
16133 (struct bit_test_cluster): Likewise.
16134 (struct min_cluster_item): Likewise.
16135 (struct case_tree_node): Likewise.
16136 (case_tree_node::case_tree_node): Likewise.
16137 (jump_table_cluster::case_values_threshold): Likewise.
16138 (struct case_bit_test): Likewise.
16139 (struct switch_decision_tree): Likewise.
16140 (struct switch_conversion): Likewise.
16141 (switch_decision_tree::reset_out_edges_aux): Likewise.
16143 2018-06-20 Martin Liska <mliska@suse.cz>
16145 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
16146 (hoist_edge_and_branch_if_true): Likewise.
16147 (expand_switch_using_bit_tests_p): Likewise.
16148 (struct case_bit_test): Likewise.
16149 (case_bit_test_cmp): Likewise.
16150 (emit_case_bit_tests): Likewise.
16151 (switch_conversion::switch_conversion): New class.
16152 (struct switch_conv_info): Remove old struct.
16153 (collect_switch_conv_info): More to ...
16154 (switch_conversion::collect): ... this.
16155 (check_range): Likewise.
16156 (switch_conversion::check_range): Likewise.
16157 (check_all_empty_except_final): Likewise.
16158 (switch_conversion::check_all_empty_except_final): Likewise.
16159 (check_final_bb): Likewise.
16160 (switch_conversion::check_final_bb): Likewise.
16161 (create_temp_arrays): Likewise.
16162 (switch_conversion::create_temp_arrays): Likewise.
16163 (free_temp_arrays): Likewise.
16164 (gather_default_values): Likewise.
16165 (switch_conversion::gather_default_values): Likewise.
16166 (build_constructors): Likewise.
16167 (switch_conversion::build_constructors): Likewise.
16168 (constructor_contains_same_values_p): Likewise.
16169 (switch_conversion::contains_same_values_p): Likewise.
16170 (array_value_type): Likewise.
16171 (switch_conversion::array_value_type): Likewise.
16172 (build_one_array): Likewise.
16173 (switch_conversion::build_one_array): Likewise.
16174 (build_arrays): Likewise.
16175 (switch_conversion::build_arrays): Likewise.
16176 (gen_def_assigns): Likewise.
16177 (switch_conversion::gen_def_assigns): Likewise.
16178 (prune_bbs): Likewise.
16179 (switch_conversion::prune_bbs): Likewise.
16180 (fix_phi_nodes): Likewise.
16181 (switch_conversion::fix_phi_nodes): Likewise.
16182 (gen_inbound_check): Likewise.
16183 (switch_conversion::gen_inbound_check): Likewise.
16184 (process_switch): Use the newly created class.
16185 (switch_conversion::expand): New.
16186 (switch_conversion::~switch_conversion): New.
16187 * tree-switch-conversion.h: New file.
16189 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16191 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
16192 tree-vect-patterns.c.
16193 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
16194 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
16195 (vect_recog_sad_pattern): Likewise.
16196 (vect_recog_widen_sum_pattern): Likewise.
16197 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
16198 (vect_recog_widen_shift_pattern): Remove the type_in argument.
16199 (vect_recog_rotate_pattern): Likewise.
16200 (vect_recog_mult_pattern): Likewise.
16201 (vect_recog_vector_vector_shift_pattern): Likewise.
16202 (vect_recog_divmod_pattern): Likewise.
16203 (vect_recog_mixed_size_cond_pattern): Likewise.
16204 (vect_recog_bool_pattern): Likewise.
16205 (vect_recog_mask_conversion_pattern): Likewise.
16206 (vect_try_gather_scatter_pattern): Likewise.
16207 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
16208 (vect_recog_over_widening_pattern): Likewise.
16209 (vect_recog_gather_scatter_pattern): Likewise.
16210 (vect_recog_func_ptr): Move from tree-vectorizer.h
16211 (vect_vect_recog_func_ptrs): Move further down the file.
16212 (vect_recog_func): Likewise. Remove the third argument.
16213 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
16214 (vect_pattern_recog_1): Expect the pattern function to do any
16215 necessary target tests. Also expect it to provide a vector type.
16216 Remove the type_in handling.
16218 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16220 * tree-vect-patterns.c (vect_pattern_detected): New function.
16221 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
16222 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
16223 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
16224 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
16225 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
16226 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
16227 (vect_recog_mask_conversion_pattern)
16228 (vect_try_gather_scatter_pattern): Likewise.
16230 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16232 * tree-vect-patterns.c (vect_get_internal_def): New function.
16233 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
16234 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
16235 (search_type_for_mask_1): Use it.
16237 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16239 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
16240 redundant WIDEN_SUM_EXPR handling.
16241 (vect_recog_sad_pattern): Likewise.
16243 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16245 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
16246 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
16247 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
16248 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
16249 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
16251 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16253 * tree-vect-stmts.c (vectorizable_call): Make sure that we
16254 use the stmt_vec_info of the original bb statement for the
16255 new zero assignment, even if the call is part of a pattern.
16257 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16259 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
16260 that the sequence is attached to the original statement rather
16261 than the pattern statement.
16262 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
16263 PATTERN_DEF_SEQ from the original statement rather than
16264 the main pattern statement.
16265 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
16266 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
16267 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
16269 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16271 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
16272 definition statements before the early exit for statements that aren't
16274 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
16276 (vect_transform_loop): ...here. Process pattern definition
16277 statements without first checking whether the main pattern
16278 statement is live or relevant.
16280 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
16282 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
16283 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
16285 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
16287 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
16288 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
16289 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
16290 (expand_block_compare): Change select_block_compare_mode call.
16291 (expand_strncmp_align_check): Use new functions, fix comment.
16292 (emit_final_str_compare_gpr): New function.
16293 (expand_strn_compare): Refactor and clean up code.
16294 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
16296 2018-06-19 Tony Reix <tony.reix@atos.com>
16297 Damien Bergamini <damien.bergamini@atos.com>
16298 David Edelsohn <dje.gcc@gmail.com>
16300 * collect2.c (static_obj): New variable.
16301 (static_libs): New variable.
16302 (is_in_list): Uncomment declaration.
16303 (main): Track AIX libraries linked statically.
16304 (is_in_list): Uncomment definition.
16305 (scan_prog_file): Don't add AIX shared libraries initializer
16306 to constructor list if linking statically.
16308 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
16310 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
16312 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
16314 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
16316 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
16319 2018-06-19 Martin Liska <mliska@suse.cz>
16321 * config/i386/i386.c (ix86_can_inline_p): Do not use
16322 ipa_fn_summaries::get_create.
16323 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
16325 (devirtualization_time_bonus): Likewise.
16326 (ipcp_propagate_stage): Likewise.
16327 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
16328 (edge_set_predicate): Likewise.
16329 (evaluate_conditions_for_known_args): Likewise.
16330 (evaluate_properties_for_edge): Likewise.
16331 (ipa_call_summary::reset): Tranform to ...
16332 (ipa_call_summary::~ipa_call_summary): ... this.
16333 (ipa_fn_summary::reset): Transform to ...
16334 (ipa_fn_summary::~ipa_fn_summary): ... this.
16335 (ipa_fn_summary_t::remove): Rename to ...
16336 (ipa_fn_summary_t::remove_callees): ... this.
16337 (ipa_fn_summary_t::duplicate): Use placement new
16338 instead of memory copy.
16339 (ipa_call_summary_t::duplicate): Likewise.
16340 (ipa_call_summary_t::remove): Remove.
16341 (dump_ipa_call_summary): Change get_create to get.
16342 (ipa_dump_fn_summary): Dump only when summary exists.
16343 (analyze_function_body): Use symbol_summary::get instead
16345 (compute_fn_summary): Likewise.
16346 (estimate_edge_devirt_benefit): Likewise.
16347 (estimate_edge_size_and_time): Likewise.
16348 (inline_update_callee_summaries): Likewise.
16349 (remap_edge_change_prob): Likewise.
16350 (remap_edge_summaries): Likewise.
16351 (ipa_merge_fn_summary_after_inlining): Likewise.
16352 (write_ipa_call_summary): Likewise.
16353 (ipa_fn_summary_write): Likewise.
16354 (ipa_free_fn_summary): Likewise.
16355 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
16356 (struct ipa_call_summary): Likewise.
16357 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
16359 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
16360 (estimate_size_after_inlining): Likewise.
16361 (estimate_growth): Likewise.
16362 (growth_likely_positive): Likewise.
16363 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
16364 (inline_call): Likewise.
16365 * ipa-inline.c (caller_growth_limits): Likewise.
16366 (can_inline_edge_p): Likewise.
16367 (can_inline_edge_by_limits_p): Likewise.
16368 (compute_uninlined_call_time): Likewise.
16369 (compute_inlined_call_time): Likewise.
16370 (want_inline_small_function_p): Likewise.
16371 (edge_badness): Likewise.
16372 (update_caller_keys): Likewise.
16373 (update_callee_keys): Likewise.
16374 (inline_small_functions): Likewise.
16375 (inline_to_all_callers_1): Likewise.
16376 (dump_overall_stats): Likewise.
16377 (early_inline_small_functions): Likewise.
16378 (early_inliner): Likewise.
16379 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
16380 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
16381 * ipa-pure-const.c (malloc_candidate_p): Likewise.
16382 * ipa-split.c (execute_split_functions): Likewise.
16383 * symbol-summary.h: Likewise.
16384 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
16386 2018-06-19 Richard Biener <rguenther@suse.de>
16388 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
16389 (vectorize_loops): ... here. Fix dbgcnt handling.
16390 (try_vectorize_loop): Wrap try_vectorize_loop_1.
16392 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
16395 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
16396 ieee128 argument takes up only one (vector) register, not two (floating
16399 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
16401 * gimplify.c (gimplify_init_constructor): Really never clear for an
16402 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
16404 2018-06-19 Richard Biener <rguenther@suse.de>
16406 PR tree-optimization/86179
16407 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
16408 after failed recognition.
16410 2018-06-18 Martin Sebor <msebor@redhat.com>
16412 PR middle-end/85602
16413 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
16414 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
16415 Handle integer subtraction.
16416 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
16417 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
16419 2018-06-18 David Malcolm <dmalcolm@redhat.com>
16421 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
16422 param from rtx to rtx_insn *.
16423 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
16425 (frv_ifcvt_modify_insn): Likwise.
16426 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
16427 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
16428 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
16429 as_a <rtx_insn *> cast to local "unprotected_region" once
16430 it's been established that it's not NULL or pc_rtx.
16431 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
16432 param "sethi" from rtx to rtx_insn *.
16433 (nds32_group_float_insns): Likewise for param "insn".
16434 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
16436 (vax_output_int_subtract): Likewise.
16437 * config/vax/vax.c (vax_output_int_add): Likewise for param
16439 (vax_output_int_subtract): Likewise.
16440 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
16441 (emit_pattern_after): Likewise for param "after".
16442 (emit_insn_after): Likewise.
16443 (emit_jump_insn_after): Likewise.
16444 (emit_call_insn_after): Likewise.
16445 (emit_debug_insn_after): Likewise.
16446 (emit_pattern_before): Likewise for param "before".
16447 (emit_insn_before): Likewise.
16448 (emit_jump_insn_before): Likewise.
16449 * final.c (get_insn_template): Likewise for param "insn", removing
16451 * output.h (get_insn_template): Likewise for 2nd param.
16452 * rtl.h (emit_insn_before): Likewise.
16453 (emit_jump_insn_before): Likewise.
16454 (emit_debug_insn_before_noloc): Likewise.
16455 (emit_insn_after): Likewise.
16456 (emit_jump_insn_after): Likewise.
16457 (emit_call_insn_after): Likewise.
16458 (emit_debug_insn_after): Likewise.
16459 (set_insn_deleted): Likewise for param.
16461 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
16464 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
16465 floating point modes, so that IFmode is numerically greater than
16466 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
16467 to declare the ordering. This prevents IFmode from being
16468 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
16469 machine. Include rs6000-modes.h to share the fractional values
16470 between genmodes* and the rest of the compiler.
16471 (IFmode): Likewise.
16472 (KFmode): Likewise.
16473 (TFmode): Likewise.
16474 * config/rs6000/rs6000-modes.h: New file.
16475 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
16476 meaning of rs6000_long_double_size so that 126..128 selects an
16477 appropriate 128-bit floating point type.
16478 (rs6000_option_override_internal): Likewise.
16479 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
16480 (TARGET_LONG_DOUBLE_128): Change the meaning of
16481 rs6000_long_double_size so that 126..128 selects an appropriate
16482 128-bit floating point type.
16483 (LONG_DOUBLE_TYPE_SIZE): Update comment.
16484 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
16485 source and destination to match the standard usage.
16486 (truncifkf2): Likewise.
16487 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
16488 ISA 2.07 to use an explicit clobber, instead of passing in a
16490 (copysign<mode>3_soft): Likewise.
16492 2018-06-18 David Malcolm <dmalcolm@redhat.com>
16494 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
16495 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
16496 (vect_slp_analyze_instance_dependence): Likewise.
16497 (vect_enhance_data_refs_alignment): Likewise.
16498 (vect_analyze_data_refs_alignment): Likewise.
16499 (vect_slp_analyze_and_verify_instance_alignment
16500 (vect_analyze_data_ref_accesses): Likewise.
16501 (vect_prune_runtime_alias_test_list): Likewise.
16502 (vect_analyze_data_refs): Likewise.
16503 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
16504 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
16505 (vect_analyze_scalar_cycles_1): Likewise.
16506 (vect_get_loop_niters): Likewise.
16507 (vect_analyze_loop_form_1): Likewise.
16508 (vect_update_vf_for_slp): Likewise.
16509 (vect_analyze_loop_operations): Likewise.
16510 (vect_analyze_loop): Likewise.
16511 (vectorizable_induction): Likewise.
16512 (vect_transform_loop): Likewise.
16513 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
16514 * tree-vect-slp.c (vect_analyze_slp): Likewise.
16515 (vect_make_slp_decision): Likewise.
16516 (vect_detect_hybrid_slp): Likewise.
16517 (vect_slp_analyze_operations): Likewise.
16518 (vect_slp_bb): Likewise.
16519 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
16520 (vectorizable_bswap): Likewise.
16521 (vectorizable_call): Likewise.
16522 (vectorizable_simd_clone_call): Likewise.
16523 (vectorizable_conversion): Likewise.
16524 (vectorizable_assignment): Likewise.
16525 (vectorizable_shift): Likewise.
16526 (vectorizable_operation): Likewise.
16527 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
16529 2018-06-18 Martin Sebor <msebor@redhat.com>
16531 PR tree-optimization/81384
16532 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
16533 * builtins.c (expand_builtin_strnlen): New function.
16534 (expand_builtin): Call it.
16535 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
16536 * builtins.def (BUILT_IN_STRNLEN): New.
16537 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
16538 Warn for bounds in excess of maximum object size.
16539 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
16540 single-value ranges. Handle strnlen.
16541 (handle_builtin_strlen): Handle strnlen.
16542 (strlen_check_and_optimize_stmt): Same.
16543 * doc/extend.texi (Other Builtins): Document strnlen.
16545 2018-06-18 Maya Rashish <coypu@sdf.org>
16547 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
16548 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
16549 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
16551 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
16553 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
16555 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16557 * tree.c (escaped_string::escape): Replace cast to char * by
16558 const_cast<char *> (unescaped).
16560 2018-06-18 Nick Clifton <nickc@redhat.com>
16563 * tree.c (escaped_string): New class. Converts an unescaped
16564 string into its escaped equivalent.
16565 (warn_deprecated_use): Use the new class to convert the
16566 deprecation message, if present.
16567 (test_escaped_strings): New self test.
16568 (test_c_tests): Add test_escaped_strings.
16569 * doc/extend.texi (deprecated): Add a note that the
16570 deprecation message is affected by the -fmessage-length
16571 option, and that control characters will be escaped.
16572 (#pragma GCC error): Document this pragma.
16573 (#pragma GCC warning): Likewise.
16574 * doc/invoke.texi (-fmessage-length): Document this option's
16575 effect on the #warning and #error preprocessor directives and
16576 the deprecated attribute.
16578 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
16580 * tree.c (decl_value_expr_lookup): Revert latest change.
16581 (decl_value_expr_insert): Likewise.
16583 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
16585 * gimplify.c (nonlocal_vlas): Delete.
16586 (nonlocal_vla_vars): Likewise.
16587 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
16589 (gimplify_body): Do not create and destroy nonlocal_vlas.
16590 * tree-nested.c: Include diagnostic.h.
16591 (use_pointer_in_frame): Tweak.
16592 (lookup_field_for_decl): Add assertion and declare the transformation.
16593 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
16594 internal error when the reference is in a wrong context. Do not
16595 create a debug decl by default.
16596 (note_nonlocal_block_vlas): Delete.
16597 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
16598 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
16599 create a debug decl by default.
16600 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
16601 call is in a wrong context.
16602 (fixup_vla_decls): New function.
16603 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
16604 debug variables were created.
16605 * tree.c (decl_value_expr_lookup): Add checking assertion.
16606 (decl_value_expr_insert): Likewise.
16608 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
16610 PR middle-end/82479
16611 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
16612 * tree-scalar-evolution.c (interpret_expr): Likewise.
16613 (expression_expensive_p): Likewise.
16614 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
16615 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
16616 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
16617 (ssa_defined_by_minus_one_stmt_p): New.
16619 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
16621 PR middle-end/64946
16622 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
16623 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
16624 * dojump.c (do_jump): Likewise.
16625 * expr.c (expand_expr_real_2): Check operand type's sign.
16626 * fold-const.c (const_unop): Handle ABSU_EXPR.
16627 (fold_abs_const): Likewise.
16628 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
16629 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
16630 (strip_sign_op_1): Likesise.
16631 * match.pd: Add new pattern to generate ABSU_EXPR.
16632 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
16633 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
16634 * tree-eh.c (operation_could_trap_helper_p): Likewise.
16635 * tree-inline.c (estimate_operator_cost): Likewise.
16636 * tree-pretty-print.c (dump_generic_node): Likewise.
16637 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
16638 * tree.def (ABSU_EXPR): New.
16640 2018-06-16 Jakub Jelinek <jakub@redhat.com>
16642 PR middle-end/86095
16643 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
16644 documented as preserved for backward compatibility only.
16645 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
16647 PR rtl-optimization/86108
16648 * bb-reorder.c (create_forwarder_block): Renamed to ...
16649 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
16650 jump from new landing pad to the second part.
16651 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
16654 2018-06-15 Jakub Jelinek <jakub@redhat.com>
16656 PR middle-end/85878
16657 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
16658 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
16659 Only call store_expr for halves if the mode is the same.
16661 PR middle-end/86123
16662 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
16663 Fix up comment formatting.
16665 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
16667 * typed-splay-tree.h (typed_splay_tree::remove): New function.
16668 (typed_splay_tree::closure,
16669 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
16670 (typed_splay_tree::typed_splay_tree,
16671 typed_splay_tree::operator =): Declared private.
16672 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
16673 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
16674 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
16675 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
16676 typed_splay_tree::splay_tree_splay,
16677 typed_splay_tree::splay_tree_foreach_helper,
16678 typed_splay_tree::splay_tree_insert,
16679 typed_splay_tree::splay_tree_remove,
16680 typed_splay_tree::splay_tree_lookup,
16681 typed_splay_tree::splay_tree_predecessor,
16682 typed_splay_tree::splay_tree_successor,
16683 typed_splay_tree::splay_tree_min,
16684 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
16685 (typed_splay_tree::root, typed_splay_tree::comp,
16686 typed_splay_tree::delete_key,
16687 typed_splay_tree::delete_value): New data members.
16688 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
16689 typed_splay_tree::remove.
16691 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
16693 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
16694 -mginv and -mno-ginv to the assembler.
16695 * config/mips/mips.opt (-mcrc): New option.
16696 (-mginv): Likewise.
16697 * doc/invoke.text (-mcrc): Document.
16698 (-mginv): Likewise.
16700 2018-06-15 Nick Clifton <nickc@redhat.com>
16703 * tree.c (escaped_string): New class. Converts an unescaped
16704 string into its escaped equivalent.
16705 (warn_deprecated_use): Use the new class to convert the
16706 deprecation message, if present.
16707 (test_escaped_strings): New self test.
16708 (test_c_tests): Add test_escaped_strings.
16709 * doc/extend.texi (deprecated): Add a note that the
16710 deprecation message is affected by the -fmessage-length
16711 option, and that control characters will be escaped.
16712 (#pragma GCC error): Document this pragma.
16713 (#pragma GCC warning): Likewise.
16714 * doc/invoke.texi (-fmessage-length): Document this option's
16715 effect on the #warning and #error preprocessor directives and
16716 the deprecated attribute.
16718 2018-06-15 Richard Biener <rguenther@suse.de>
16720 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
16721 here, also noting vector size used.
16722 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
16723 size used in MSG_OPTIMIZED_LOCATIONS dump.
16724 (pass_slp_vectorize::execute): Adjust.
16726 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
16729 * config/arc/arc.c (arc_return_address_register): Fix
16732 2018-06-15 Richard Biener <rguenther@suse.de>
16734 PR middle-end/86159
16735 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
16736 leave useless conversion stripping to force_gimple_operand_gsi.
16737 (gimplify_build2): Likewise.
16738 (gimplify_build1): Likewise.
16740 2018-06-15 Richard Biener <rguenther@suse.de>
16742 PR middle-end/86076
16743 * tree-cfg.c (move_stmt_op): unshare invariant addresses
16744 before adjusting their block.
16746 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
16748 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
16749 multilibs for *-*-rtems*.
16750 * config/riscv/t-rtems: New file.
16752 2018-06-14 Jakub Jelinek <jakub@redhat.com>
16754 PR middle-end/86122
16755 * match.pd ((A +- CST1) +- CST2): Punt if last resort
16756 unsigned_type_for returns NULL.
16759 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
16760 subregs of multi-word pseudos unless the float mode has word size.
16762 2018-06-14 Richard Biener <rguenther@suse.de>
16764 PR middle-end/86139
16765 * tree-vect-generic.c (build_word_mode_vector_type): Remove
16766 duplicate and harmful type_hash_canon.
16767 * tree.c (type_hash_canon): Assert we didn't find ourselves.
16769 2018-06-14 Richard Biener <rguenther@suse.de>
16772 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
16775 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
16777 * config/rtems.h (STDINT_LONG32): Define.
16779 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
16780 Prachi Godbole <prachi.godbole@imgtec.com>
16782 * config/mips/mips-cpus.def: Define P6600.
16783 * config/mips/mips-tables.opt: Regenerate.
16784 * config/mips/mips.c (mips_ucbranch_type): New enum.
16785 (mips_rtx_cost_data): Add support for P6600.
16786 (mips_issue_rate): Likewise.
16787 (mips_multipass_dfa_lookahead): Likewise.
16788 (mips_avoid_hazard): Likewise.
16789 (mips_reorg_process_insns): Likewise.
16790 (mips_classify_branch_p6600): New function.
16791 * config/mips/mips.h (TUNE_P6600): New define.
16792 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
16793 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
16794 * config/mips/mips.md: Include p6600.md.
16795 (processor): Add p6600.
16796 * config/mips/p6600.md: New file.
16797 * doc/invoke.texi: Add p6600 to supported architectures.
16799 2018-06-13 Martin Sebor <msebor@redhat.com>
16801 PR tree-optimization/86114
16802 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
16804 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
16806 2018-06-13 Richard Biener <rguenther@suse.de>
16808 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
16809 Properly set vector type of the intermediate stmt.
16810 * tree-vect-stmts.c (vectorizable_operation): The destination
16811 var always has vectype_out type.
16813 2018-06-13 Jeff Law <law@redhat.com>
16815 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
16816 integer 0 for argument to print_rtl_with_bb.
16817 (rl78_reorg): Likewise.
16819 2018-06-13 David Malcolm <dmalcolm@redhat.com>
16821 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
16822 from rtx to rtx_insn *.
16823 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
16825 (add_sched_insns_for_speculation): Likewise for local "target",
16826 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
16827 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
16828 from rtx_insn ** to rtx_code_label **.
16829 (reorg_emit_nops): Likewise.
16830 (c6x_reorg): Likewise for local "call_labels".
16831 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
16833 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
16834 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
16835 the loops over LABEL_REFS.
16836 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
16838 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
16839 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
16840 (split_branches): Strengthen local "olabel" from rtx to
16841 rtx_insn *, adding a safe_as_a cast.
16842 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
16844 (add_insn_after): Likewise for first two params.
16845 (add_insn_before): Likewise.
16846 (remove_insn): Likewise for param.
16847 (emit_pattern_before_noloc): Likewise for second and third params.
16848 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
16849 (emit_call_insn_before_noloc): Likewise.
16850 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
16852 (emit_barrier_before): Likewise.
16853 (emit_label_before): Strengthen "label" param from "rtx" to
16854 "rtx_code_label *". Strengthen "before" param from "rtx" to
16856 (emit_insn_after_1): Strengthen "after" param from "rtx" to
16858 (emit_pattern_after_noloc): Likewise.
16859 (emit_insn_after_noloc): Likewise.
16860 (emit_jump_insn_after_noloc): Likewise.
16861 (emit_call_insn_after_noloc): Likewise.
16862 (emit_debug_insn_after_noloc): Likewise.
16863 (emit_barrier_after): Likewise.
16864 (emit_label_after): Likewise for both params.
16865 (emit_pattern_after_setloc): Likewise for "after" param. Convert
16866 "loc" param from "int" to "location_t".
16867 (emit_insn_after_setloc): Likewise.
16868 (emit_jump_insn_after_setloc): Likewise.
16869 (emit_call_insn_after_setloc): Likewise.
16870 (emit_debug_insn_after_setloc): Likewise.
16871 (emit_pattern_before_setloc): Likewise for "before" param. Convert
16872 "loc" param from "int" to "location_t".
16873 (emit_pattern_before): Convert NULL_RTX to NULL.
16874 (emit_insn_before_setloc): Convert "loc" param from "int" to
16876 (emit_jump_insn_before_setloc): Likewise.
16877 (emit_call_insn_before_setloc): Likewise.
16878 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
16879 rtx_insn *. Convert "loc" param from "int" to "location_t".
16880 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
16881 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
16882 Convert 3rd param from "int" to "location_t".
16883 (emit_barrier_before, emit_barrier_after, next_real_insn):
16884 Strengthen param from rtx to rtx_insn *.
16885 (emit_label_before): Strengthen 1st param from "rtx" to
16886 "rtx_code_label *". Strengthen 2nd param from "rtx" to
16888 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
16889 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
16890 Strengthen 2nd param from "rtx" to "rtx_insn *".
16891 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
16892 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
16893 Likewise. Convert 3rd param from "int" to "location_t".
16894 (emit_label_after): Strengthen 1st param from "rtx" to
16895 "rtx_code_label *".
16896 (next_real_insn, remove_insn): Strengthen param from "rtx" to
16898 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
16899 from "rtx" to "rtx_insn *".
16901 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
16903 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
16904 bodies streamed in with -Q.
16905 * dumpfile.c (dump_files): Add lto-stream-out dump file.
16906 * dumpfile.h (tree_dump_index): Add lto_stream_out.
16907 * gimple-streamer-out.c: Include gimple-pretty-print.h
16908 (output_bb): Dump stmts streamed.
16909 * lto-section-out.c: Include print-tree.h
16910 (lto_begin_section): Dump sections created.
16911 (lto_output_decl_index): Dump decl encoded.
16912 * lto-streamer-out.c: Include print-tree.h
16913 (create_output_block): Dump output block created.
16914 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
16915 (output_function): Dump function output.
16916 (output_constructor): Dump constructor streamed.
16917 (write_global_stream): Output indexes encoded.
16918 (produce_asm_for_decls): Dump streams encoded.
16919 * lto-streamer.c (streamer_dump_file): New global var.
16920 * lto-streamer.h (streamer_dump_file): Declare.
16921 * passes.c (ipa_write_summaries): Initialize streamer dump.
16922 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
16925 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
16928 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
16929 offsets for register save directives. Emit a second batch of save
16930 directives, if need be, when the function accesses prior frames.
16932 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16934 * config/arc/fpu.md (fmasf4): Force operand to register.
16935 (fnmasf4): Likewise.
16937 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16939 * config/arc/arc-protos.h (arc_pad_return): Remove.
16940 * config/arc/arc.c (machine_function): Remove force_short_suffix
16942 (arc_print_operand): Adjust printing of '&'.
16943 (arc_verify_short): Remove conditional printing of short suffix.
16944 (arc_final_prescan_insn): Remove reference to size_reason.
16945 (pad_return): New function.
16946 (arc_reorg): Call pad_return.
16947 (arc_pad_return): Remove.
16948 (arc_init_machine_status): Remove reference to force_short_suffix.
16949 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
16950 (attr length): When attribute iscompact is true force to 2
16951 regardless; in the case of maybe check if we want to force the
16952 instruction to have 4 bytes length.
16953 (nopv): Change it to generate 4 byte long nop as well.
16954 (blockage): New pattern.
16955 (simple_return): Remove call to arc_pad_return.
16956 (p_return_i): Likewise.
16958 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16960 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
16962 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16964 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
16967 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16969 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
16970 for ARC700 and ARCv2.
16972 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
16975 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
16976 operands[2] instead of operands[1].
16979 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
16981 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
16982 case, check whether the outer register overlaps an unallocatable
16983 register, not just whether it fits the required class.
16985 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
16987 * poly-int.h (can_div_trunc_p): Add new overload in which all values
16989 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
16990 (memrefs_conflict_p): Likewise.
16991 (init_alias_analysis): Likewise.
16992 * cfgexpand.c (expand_debug_expr): Likewise.
16993 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
16994 * cse.c (fold_rtx): Likewise.
16995 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
16996 * expr.c (emit_block_move_hints): Likewise.
16997 (clear_storage_hints, push_block, emit_push_insn): Likewise.
16998 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
16999 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
17000 (emit_group_store): Likewise.
17001 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
17002 to read the PRE/POST_MODIFY increment.
17003 * calls.c (store_one_arg): Use strip_offset.
17004 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
17006 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
17008 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
17009 (simplify_binary_operation_1): Extend CONST_INT handling to
17011 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
17012 than a HOST_WIDE_INT.
17013 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
17015 (adjust_mems, add_stores): Update accodingly.
17016 (vt_canonicalize_addr): Track polynomial offsets.
17017 (emit_note_insn_var_location): Likewise.
17018 (vt_add_function_parameter): Likewise.
17019 (vt_initialize): Likewise.
17021 2018-06-12 Jeff Law <law@redhat.com>
17023 * config.gcc (alpha*-*-freebsd*): Remove.
17024 * config/alpha/freebsd.h: Remove.
17026 2018-06-12 David Malcolm <dmalcolm@redhat.com>
17029 * spellcheck-tree.c (levenshtein_distance): Rename to...
17030 (get_edit_distance): ...this, and update for underlying renaming.
17031 * spellcheck-tree.h (levenshtein_distance): Rename to...
17032 (get_edit_distance): ...this.
17033 * spellcheck.c (levenshtein_distance): Rename to...
17034 (get_edit_distance): ...this. Convert from Levenshtein distance
17035 to Damerau-Levenshtein distance by supporting transpositions of
17036 adjacent characters. Rename "v1" to "v_next" and "v0" to
17038 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
17039 (selftest::test_edit_distance_unit_test_oneway): ...this, and
17040 update for underlying renaming.
17041 (selftest::levenshtein_distance_unit_test): Rename to...
17042 (selftest::test_get_edit_distance_unit): ...this, and update for
17043 underlying renaming.
17044 (selftest::test_find_closest_string): Add example from PR 69968
17045 where transposition helps
17046 (selftest::test_metric_conditions): Update for renaming.
17047 (selftest::test_metric_conditions): Likewise.
17048 (selftest::spellcheck_c_tests): Likewise.
17049 * spellcheck.h (levenshtein_distance): Rename both overloads to...
17050 (get_edit_distance): ...this.
17051 (best_match::consider): Update for renaming.
17053 2018-06-12 Martin Sebor <msebor@redhat.com>
17055 PR tree-optimization/85259
17056 * builtins.c (compute_objsize): Handle constant offsets.
17057 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
17058 true iff a warning has been issued.
17059 * gimple.h (gimple_nonartificial_location): New function.
17060 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
17061 gimple_nonartificial_location and handle -Wno-system-headers.
17062 (handle_builtin_stxncpy): Same.
17064 2018-06-12 Martin Sebor <msebor@redhat.com>
17067 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
17069 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17071 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17072 BUILTIN_VEC_XST entries for pointer to double and long long.
17074 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
17077 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
17079 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
17082 2018-06-12 Martin Liska <mliska@suse.cz>
17084 * doc/options.texi: Document IntegerRange.
17086 2018-06-12 Martin Liska <mliska@suse.cz>
17088 * config/i386/i386.opt: Make MPX-related options as Deprecated.
17089 * opt-functions.awk: Handle Deprecated flag.
17090 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
17092 (read_cmdline_option): Report warning for a deprecated option.
17093 * opts.h (struct cl_option): Add new field cl_deprecated.
17094 (CL_ERR_DEPRECATED): New.
17096 2018-06-12 Martin Liska <mliska@suse.cz>
17098 * doc/options.texi: Document Deprecated option flag.
17100 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
17102 * config/arc/arc-arch.h (arc_extras): New enum.
17103 (arc_cpu_t):Add field extra.
17104 (arc_cpu_types): Consider the extras.
17105 * config/arc/arc-cpus.def: Add extras info.
17106 * config/arc/arc-opts.h (processor_type): Consider extra field.
17107 * config/arc/arc.c (arc_override_options): Handle extra field.
17109 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
17111 * config/arc/arc-arch.h: Update ARC_OPTX macro.
17112 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
17114 * config/arc/arc.c (arc_init): Update pic warning.
17115 (irq_range): Update irq range parsing warnings.
17116 (arc_override_options): Update various warning messages.
17117 (arc_handle_aux_attribute): Likewise.
17119 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
17121 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
17123 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
17125 * doc/sourcebuild.texi: Document usage of line number 0 in verify
17126 compiler messages directives.
17128 2018-06-12 Matthew Fortune <mfortune@gmail.com>
17130 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
17131 * config/mips/mips-tables.opt: Regenerate.
17132 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
17134 * doc/invoke.texi: Document -march=i6500.
17136 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
17138 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
17139 (i6400_gpmul): Add cpu_unit.
17140 (i6400_gpdiv): Likewise.
17141 (i6400_msa_add_d): Update reservations.
17142 (i6400_msa_int_add) Likewise.
17143 (i6400_msa_short_logic3) Likewise.
17144 (i6400_msa_short_logic2) Likewise.
17145 (i6400_msa_short_logic) Likewise.
17146 (i6400_msa_move) Likewise.
17147 (i6400_msa_cmp) Likewise.
17148 (i6400_msa_short_float2) Likewise.
17149 (i6400_msa_div_d) Likewise.
17150 (i6400_msa_long_logic1) Likewise.
17151 (i6400_msa_long_logic2) Likewise.
17152 (i6400_msa_mult) Likewise.
17153 (i6400_msa_long_float2) Likewise.
17154 (i6400_msa_long_float4) Likewise.
17155 (i6400_msa_long_float5) Likewise.
17156 (i6400_msa_long_float8) Likewise.
17157 (i6400_fpu_fadd): Include frint type.
17158 (i6400_fpu_store): New define_insn_reservation.
17159 (i6400_fpu_load): Likewise.
17160 (i6400_fpu_move): Likewise.
17161 (i6400_fpu_fcmp): Likewise.
17162 (i6400_fpu_fmadd): Likewise.
17163 (i6400_int_mult): Include imul3nc type and update reservation.
17164 (i6400_int_div): Include idiv3 type and update reservation.
17165 (i6400_int_load): Update to check type not move_type.
17166 (i6400_int_store): Likewise.
17167 (i6400_int_prefetch): Set zero latency.
17169 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
17171 * gcc.c: Document new %@{...} sequence.
17172 (LINK_COMMAND_SPEC): Use it for the -L switches.
17173 (cpp_unique_options): Use it for the -I switches.
17174 (at_file_argbuf): New global variable.
17175 (in_at_file): Likewise.
17176 (alloc_args): Create at_file_argbuf.
17177 (clear_args): Truncate at_file_argbuf.
17178 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
17179 (open_at_file): New function.
17180 (close_at_file): Likewise.
17181 (create_at_file): Delete.
17182 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
17185 (validate_switches_from_spec): Deal with %@{...} sequence.
17186 (validate_switches): Likewise.
17187 (driver::finalize): Call clear_args.
17189 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
17191 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
17193 2018-06-11 Martin Sebor <msebor@redhat.com>
17195 * doc/invoke.texi (-Wall): List -Wc++17-compat.
17196 (Wno-class-memaccess): Add @opindex.
17197 (Wno-templates, Wno-multiple-inheritance): Same.
17198 (Wno-virtual-inheritance, Wno-namespaces): Same.
17199 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
17200 (Wno-format-overflow, Wno-format-truncation): Same.
17201 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
17202 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
17203 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
17204 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
17205 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
17206 (Wno-misspelled-isr): Same.
17208 2018-06-11 Martin Sebor <msebor@redhat.com>
17210 * PR tree-optimization/86083
17211 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
17213 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
17215 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
17217 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
17220 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
17221 on the correct operand.
17222 (*movdi_internal64): Ditto.
17224 2018-06-11 Martin Liska <mliska@suse.cz>
17226 PR tree-optimization/86089
17227 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
17229 2018-06-11 Julia Koval <julia.koval@intel.com>
17231 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
17232 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
17233 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
17235 2018-06-11 Olivier Hainque <hainque@adacore.com>
17237 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
17238 for Ada with strict dwarf2.
17240 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
17243 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
17246 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
17248 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
17250 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
17252 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
17255 2018-06-08 Martin Liska <mliska@suse.cz>
17257 * tree-cfg.h (debug_function): Fix argument type to match
17260 2018-06-08 Martin Liska <mliska@suse.cz>
17262 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
17263 Remove usage of MPX-related (and removed) fields.
17264 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
17266 2018-06-08 David Malcolm <dmalcolm@redhat.com>
17268 * cfg.c (debug): Use TDF_NONE rather than 0.
17269 * cfghooks.c (debug): Likewise.
17270 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
17271 (struct dump_option_value_info): Convert to...
17272 (struct kv_pair): ...this template type.
17273 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
17275 (optinfo_verbosity_options): Likewise.
17276 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
17278 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
17279 than int for "optgroup_flags" param.
17280 (dump_generic_expr_loc): Use dump_flags_t rather than int for
17282 (dump_dec): Likewise.
17283 (dump_finish): Use TDF_NONE rather than 0.
17284 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
17285 rather than int for "optgroup_flags" param. Use TDF_NONE rather
17286 than 0. Update for change to option_ptr.
17287 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
17288 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
17289 0. Update for changes to optinfo_verbosity_options and
17291 (opt_info_switch_p): Convert optgroup_flags from int to
17293 (dump_basic_block): Use dump_flags_t rather than int
17294 for "dump_kind" param.
17295 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
17296 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
17297 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
17298 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
17299 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
17300 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
17301 TDF_NONE): Convert from macros to...
17302 (enum dump_flag): ...this new enum.
17303 (dump_flags_t): Update to use enum.
17304 (operator|, operator&, operator~, operator|=, operator&=):
17305 Implement for dump_flags_t.
17306 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
17307 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
17308 Convert from macros to...
17309 (enum optgroup_flag): ...this new enum.
17310 (optgroup_flags_t): New typedef.
17311 (operator|, operator|=): Implement for optgroup_flags_t.
17312 (struct dump_file_info): Convert field "alt_flags" to
17313 dump_flags_t. Convert field "optgroup_flags" to
17315 (dump_basic_block): Use dump_flags_t rather than int for param.
17316 (dump_generic_expr_loc): Likewise.
17317 (dump_dec): Likewise.
17318 (dump_register): Convert param "optgroup_flags" to
17320 (opt_info_enable_passes): Likewise.
17321 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
17323 * gimple-pretty-print.c (debug): Likewise.
17324 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
17325 (merged_store_group::apply_stores): Likewise.
17326 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
17327 * gimple.c (verify_gimple_pp): Likewise.
17328 * graphite-poly.c (print_pbb_body): Likewise.
17329 * passes.c (pass_manager::register_one_dump_file): Convert
17330 local "optgroup_flags" to optgroup_flags_t.
17331 * print-tree.c (print_node): Use TDF_NONE rather than 0.
17333 (debug_body): Likewise.
17334 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
17335 to optgroup_flags_t.
17336 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
17338 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
17339 (convert_mult_to_fma): Likewise.
17340 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
17341 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
17342 * tree-vect-data-refs.c (dump_lower_bound): Convert param
17343 "dump_kind" to dump_flags_t.
17345 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
17347 * config/rs6000/rs6000.c (min, max): Delete.
17349 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
17351 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
17352 -mabi=spe and -mabi=no-spe.
17354 2018-06-08 Martin Liska <mliska@suse.cz>
17356 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
17357 where we expect an existing summary.
17359 2018-06-08 Martin Liska <mliska@suse.cz>
17361 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
17362 * ipa-inline.h (estimate_edge_growth): Likewise.
17364 2018-06-08 Martin Liska <mliska@suse.cz>
17366 * cgraph.c (function_version_hasher::hash): Use
17367 cgraph_node::get_uid ().
17368 (function_version_hasher::equal):
17369 * cgraph.h (cgraph_node::get_uid): New method.
17370 * ipa-inline.c (update_caller_keys): Use
17371 cgraph_node::get_uid ().
17372 (update_callee_keys): Likewise.
17373 * ipa-utils.c (searchc): Likewise.
17374 (ipa_reduced_postorder): Likewise.
17375 * lto-cgraph.c (input_node): Likewise.
17376 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
17377 * symbol-summary.h (symtab_insertion): Likewise.
17378 (symtab_removal): Likewise.
17379 (symtab_duplication): Likewise.
17380 * tree-pretty-print.c (dump_function_header): Likewise.
17381 * tree-sra.c (convert_callers_for_node): Likewise.
17383 2018-06-08 Martin Liska <mliska@suse.cz>
17385 * cgraph.c (symbol_table::create_edge): Always assign a new
17387 (symbol_table::free_edge): Do not recycle numbers.
17388 * cgraph.h (cgraph_edge::get): New method.
17389 * symbol-summary.h (symtab_removal): Use it.
17390 (symtab_duplication): Likewise.
17391 (call_summary::hashable_uid): Remove.
17393 2018-06-08 Martin Liska <mliska@suse.cz>
17395 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
17396 (initialize_growth_caches): Remove.
17397 (free_growth_caches): Likewise.
17398 (do_estimate_edge_time): Use edge_growth_cache.
17399 (do_estimate_edge_size): Likewise.
17400 (do_estimate_edge_hints): Likewise.
17401 * ipa-inline.c (reset_edge_caches): Likewise.
17402 (recursive_inlining): Likewise.
17403 (inline_small_functions): Likewise.
17404 * ipa-inline.h (initialize_growth_caches): Remove.
17405 (estimate_edge_size): Likewise.
17406 (estimate_edge_time): Likewise.
17407 (estimate_edge_hints): Likewise.
17408 (reset_edge_growth_cache): Likewise.
17409 * symbol-summary.h (call_summary::remove): New method.
17411 2018-06-08 Martin Liska <mliska@suse.cz>
17413 * ipa-cp.c (class edge_clone_summary): New summary.
17414 (grow_edge_clone_vectors): Remove.
17415 (ipcp_edge_duplication_hook): Remove.
17416 (class edge_clone_summary_t): New call_summary class.
17417 (ipcp_edge_removal_hook): Remove.
17418 (edge_clone_summary_t::duplicate): New function.
17419 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
17420 (create_specialized_node): Likewise.
17421 (ipcp_driver): Initialize edge_clone_summaries and do not
17424 2018-06-08 Martin Liska <mliska@suse.cz>
17426 * symbol-summary.h (get): New function.
17427 (call_summary::m_initialize_when_cloning): New class member.
17429 2018-06-08 Martin Liska <mliska@suse.cz>
17431 * cgraph.c (cgraph_node::remove): Do not recycle uid.
17432 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
17433 (symbol_table::allocate_cgraph_symbol): Do not set uid.
17434 * passes.c (uid_hash_t): Record removed_nodes by their uids.
17435 (remove_cgraph_node_from_order): Use the removed_nodes set.
17436 (do_per_function_toporder): Likwise.
17437 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
17438 instead of summary_uid.
17439 (symtab_removal): Likewise.
17440 (symtab_duplication): Likewise.
17442 2018-06-08 Martin Liska <mliska@suse.cz>
17444 * ipa-cp.c (ipcp_store_bits_results): Use
17445 ipcp_transformation_sum.
17446 (ipcp_store_vr_results): Likewise.
17447 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
17449 (ipcp_transformation_initialize): ... this.
17450 (ipa_set_node_agg_value_chain):
17451 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
17452 (write_ipcp_transformation_info): Likewise.
17453 (read_ipcp_transformation_info): Likewise.
17454 (ipcp_update_bits): Likewise.
17455 (ipcp_update_vr): Likewise.
17456 (ipcp_transform_function): Likewise.
17457 * ipa-prop.h: Rename ipcp_transformation_summary to
17458 ipcp_transformation.
17459 (class ipcp_transformation_t): New function summary.
17460 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
17461 (ipa_get_agg_replacements_for_node): Likewise.
17463 2018-06-08 Martin Liska <mliska@suse.cz>
17465 * ipa-pure-const.c (struct funct_state_d): Do it class instead
17467 (class funct_state_summary_t): New function_summary class.
17468 (has_function_state): Remove.
17469 (get_function_state): Likewise.
17470 (set_function_state): Likewise.
17471 (add_new_function): Likewise.
17472 (funct_state_summary_t::insert): New function.
17473 (duplicate_node_data): Remove.
17474 (remove_node_data): Remove.
17475 (funct_state_summary_t::duplicate): New function.
17476 (register_hooks): Create new funct_state_summaries.
17477 (pure_const_generate_summary): Use it.
17478 (pure_const_write_summary): Likewise.
17479 (pure_const_read_summary): Likewise.
17480 (propagate_pure_const): Likewise.
17481 (propagate_nothrow): Likewise.
17482 (dump_malloc_lattice): Likewise.
17483 (propagate_malloc): Likewise.
17484 (execute): Do not register hooks, just remove summary
17486 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
17489 2018-06-08 Martin Liska <mliska@suse.cz>
17491 * ipa-reference.c (remove_node_data): Remove.
17492 (duplicate_node_data): Likewise.
17493 (class ipa_ref_var_info_summary_t): New class.
17494 (class ipa_ref_opt_summary_t): Likewise.
17495 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
17496 (get_reference_optimization_summary): Use
17497 ipa_ref_opt_sum_summaries.
17498 (set_reference_vars_info): Remove.
17499 (set_reference_optimization_summary): Likewise.
17500 (ipa_init): Create summaries.
17501 (init_function_info): Use function summary.
17502 (ipa_ref_opt_summary_t::duplicate): New function.
17503 (ipa_ref_opt_summary_t::remove): New function.
17504 (get_read_write_all_from_node): Fix GNU coding style.
17505 (propagate): Use function summary.
17506 (write_node_summary_p): Fix GNU coding style.
17507 (stream_out_bitmap): Likewise.
17508 (ipa_reference_read_optimization_summary): Use function summary.
17509 (ipa_reference_c_finalize): Do not release hooks.
17511 2018-06-08 Martin Liska <mliska@suse.cz>
17513 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
17514 (analyze_function_body): Extract multiple calls of get_create.
17515 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
17516 * ipa-inline.c (recursive_inlining): Use ::get method.
17517 * ipa-inline.h (estimate_edge_growth): Likewise.
17519 2018-06-08 Martin Liska <mliska@suse.cz>
17521 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
17523 (hsa_function_summary::hsa_function_summary): Use the new enum
17525 (hsa_gpu_implementation_p): Use hsa_summaries::get.
17526 * hsa-gen.c (hsa_get_host_function): Likewise.
17527 (get_brig_function_name): Likewise.
17528 * ipa-hsa.c (process_hsa_functions): Likewise.
17529 (ipa_hsa_write_summary): Likewise.
17530 * symbol-summary.h (symtab_duplication): Use ::get function/
17531 (get): New function.
17533 2018-06-08 Martin Liska <mliska@suse.cz>
17535 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
17537 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
17538 (hsa_register_kernel): Likewise.
17539 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
17540 * hsa-gen.c (hsa_get_host_function): Likewise.
17541 (get_brig_function_name): Likewise.
17542 (generate_hsa): Likewise.
17543 (pass_gen_hsail::execute): Likewise.
17544 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
17545 (devirtualization_time_bonus): Likewise.
17546 (ipcp_propagate_stage): Likewise.
17547 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
17548 (edge_set_predicate): Likewise.
17549 (evaluate_conditions_for_known_args): Likewise.
17550 (evaluate_properties_for_edge): Likewise.
17551 (ipa_fn_summary::reset): Likewise.
17552 (ipa_fn_summary_t::duplicate): Likewise.
17553 (dump_ipa_call_summary): Likewise.
17554 (ipa_dump_fn_summary): Likewise.
17555 (analyze_function_body): Likewise.
17556 (compute_fn_summary): Likewise.
17557 (estimate_edge_devirt_benefit): Likewise.
17558 (estimate_edge_size_and_time): Likewise.
17559 (estimate_calls_size_and_time): Likewise.
17560 (estimate_node_size_and_time): Likewise.
17561 (inline_update_callee_summaries): Likewise.
17562 (remap_edge_change_prob): Likewise.
17563 (remap_edge_summaries): Likewise.
17564 (ipa_merge_fn_summary_after_inlining): Likewise.
17565 (ipa_update_overall_fn_summary): Likewise.
17566 (read_ipa_call_summary): Likewise.
17567 (inline_read_section): Likewise.
17568 (write_ipa_call_summary): Likewise.
17569 (ipa_fn_summary_write): Likewise.
17570 (ipa_free_fn_summary): Likewise.
17571 * ipa-hsa.c (process_hsa_functions): Likewise.
17572 (ipa_hsa_write_summary): Likewise.
17573 (ipa_hsa_read_section): Likewise.
17574 * ipa-icf.c (sem_function::merge): Likewise.
17575 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
17576 (do_estimate_edge_time): Likewise.
17577 (estimate_size_after_inlining): Likewise.
17578 (estimate_growth): Likewise.
17579 (growth_likely_positive): Likewise.
17580 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
17581 (inline_call): Likewise.
17582 * ipa-inline.c (caller_growth_limits): Likewise.
17583 (can_inline_edge_p): Likewise.
17584 (can_inline_edge_by_limits_p): Likewise.
17585 (compute_uninlined_call_time): Likewise.
17586 (compute_inlined_call_time): Likewise.
17587 (want_inline_small_function_p): Likewise.
17588 (edge_badness): Likewise.
17589 (update_caller_keys): Likewise.
17590 (update_callee_keys): Likewise.
17591 (recursive_inlining): Likewise.
17592 (inline_small_functions): Likewise.
17593 (inline_to_all_callers_1): Likewise.
17594 (dump_overall_stats): Likewise.
17595 (early_inline_small_functions): Likewise.
17596 (early_inliner): Likewise.
17597 * ipa-inline.h (estimate_edge_growth): Likewise.
17598 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
17599 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
17600 * ipa-prop.h (IPA_NODE_REF): Likewise.
17601 (IPA_EDGE_REF): Likewise.
17602 * ipa-pure-const.c (malloc_candidate_p): Likewise.
17603 (propagate_malloc): Likewise.
17604 * ipa-split.c (execute_split_functions): Likewise.
17605 * symbol-summary.h: Rename get to get_create.
17607 (get_create): Likewise.
17608 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
17610 2018-06-08 Martin Liska <mliska@suse.cz>
17612 * symbol-summary.h (release): Move definition out of class
17614 (symtab_removal): Likewise.
17615 (symtab_duplication): Likewise.
17617 2018-06-08 Martin Liska <mliska@suse.cz>
17619 * symbol-summary.h (function_summary): Move constructor
17620 implementation out of class declaration.
17621 (release): Likewise.
17622 (symtab_insertion): Likewise.
17623 (symtab_removal): Likewise.
17624 (symtab_duplication): Likewise.
17627 2018-06-08 Martin Liska <mliska@suse.cz>
17629 * Makefile.in: Remove support for MPX (macros, related functions,
17630 fields in cgraph_node, ...).
17631 * builtin-types.def (BT_BND): Likewise.
17632 (BT_FN_BND_CONST_PTR): Likewise.
17633 (BT_FN_CONST_PTR_BND): Likewise.
17634 (BT_FN_VOID_PTR_BND): Likewise.
17635 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
17636 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
17637 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
17638 (expand_builtin_mempcpy_with_bounds): Likewise.
17639 (expand_builtin_memset_with_bounds): Likewise.
17640 (expand_builtin_memset_args): Likewise.
17641 (std_expand_builtin_va_start): Likewise.
17642 (expand_builtin): Likewise.
17643 (expand_builtin_with_bounds): Likewise.
17644 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
17645 (DEF_LIB_BUILTIN_CHKP): Likewise.
17646 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
17647 (DEF_CHKP_BUILTIN): Likewise.
17648 (BUILT_IN_MEMCPY): Likewise.
17649 (BUILT_IN_MEMMOVE): Likewise.
17650 (BUILT_IN_MEMPCPY): Likewise.
17651 (BUILT_IN_MEMSET): Likewise.
17652 (BUILT_IN_STPCPY): Likewise.
17653 (BUILT_IN_STRCAT): Likewise.
17654 (BUILT_IN_STRCHR): Likewise.
17655 (BUILT_IN_STRCPY): Likewise.
17656 (BUILT_IN_STRLEN): Likewise.
17657 (BUILT_IN_MEMCPY_CHK): Likewise.
17658 (BUILT_IN_MEMMOVE_CHK): Likewise.
17659 (BUILT_IN_MEMPCPY_CHK): Likewise.
17660 (BUILT_IN_MEMSET_CHK): Likewise.
17661 (BUILT_IN_STPCPY_CHK): Likewise.
17662 (BUILT_IN_STRCAT_CHK): Likewise.
17663 (BUILT_IN_STRCPY_CHK): Likewise.
17664 * calls.c (store_bounds): Likewise.
17665 (emit_call_1): Likewise.
17666 (special_function_p): Likewise.
17667 (maybe_warn_nonstring_arg): Likewise.
17668 (initialize_argument_information): Likewise.
17669 (finalize_must_preallocate): Likewise.
17670 (compute_argument_addresses): Likewise.
17671 (expand_call): Likewise.
17672 * cfgexpand.c (expand_call_stmt): Likewise.
17673 (expand_return): Likewise.
17674 (expand_gimple_stmt_1): Likewise.
17675 (pass_expand::execute): Likewise.
17676 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
17677 (cgraph_node::remove): Likewise.
17678 (cgraph_node::dump): Likewise.
17679 (cgraph_node::verify_node): Likewise.
17680 * cgraph.h (chkp_function_instrumented_p): Likewise.
17681 (symtab_node::get_alias_target): Likewise.
17682 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
17683 (cgraph_local_p): Likewise.
17684 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
17685 (cgraph_edge::rebuild_references): Likewise.
17686 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
17687 (walk_polymorphic_call_targets): Likewise.
17688 (cgraph_node::expand_thunk): Likewise.
17689 (symbol_table::output_weakrefs): Likewise.
17690 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
17691 (ix86_handle_option): Likewise.
17692 * config/i386/constraints.md: Likewise.
17693 * config/i386/i386-builtin-types.def (BND): Likewise.
17697 * config/i386/i386-builtin.def (BDESC_END): Likewise.
17698 (BDESC_FIRST): Likewise.
17700 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
17701 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
17702 * config/i386/i386.c (enum reg_class): Likewise.
17703 (ix86_target_string): Likewise.
17704 (ix86_option_override_internal): Likewise.
17705 (ix86_conditional_register_usage): Likewise.
17706 (ix86_valid_target_attribute_inner_p): Likewise.
17707 (ix86_set_indirect_branch_type): Likewise.
17708 (ix86_set_current_function): Likewise.
17709 (ix86_function_arg_regno_p): Likewise.
17710 (init_cumulative_args): Likewise.
17711 (ix86_function_arg_advance): Likewise.
17712 (ix86_function_arg): Likewise.
17713 (ix86_pass_by_reference): Likewise.
17714 (ix86_function_value_regno_p): Likewise.
17715 (ix86_function_value_1): Likewise.
17716 (ix86_function_value_bounds): Likewise.
17717 (ix86_return_in_memory): Likewise.
17718 (ix86_setup_incoming_vararg_bounds): Likewise.
17719 (ix86_va_start): Likewise.
17720 (indirect_thunk_need_prefix): Likewise.
17721 (print_reg): Likewise.
17722 (ix86_print_operand): Likewise.
17723 (ix86_expand_call): Likewise.
17724 (ix86_output_function_return): Likewise.
17725 (reg_encoded_number): Likewise.
17726 (BDESC_VERIFYS): Likewise.
17727 (ix86_init_mpx_builtins): Likewise.
17728 (ix86_init_builtins): Likewise.
17729 (ix86_emit_cmove): Likewise.
17730 (ix86_emit_move_max): Likewise.
17731 (ix86_expand_builtin): Likewise.
17732 (ix86_builtin_mpx_function): Likewise.
17733 (ix86_get_arg_address_for_bt): Likewise.
17734 (ix86_load_bounds): Likewise.
17735 (ix86_store_bounds): Likewise.
17736 (ix86_load_returned_bounds): Likewise.
17737 (ix86_store_returned_bounds): Likewise.
17738 (ix86_class_likely_spilled_p): Likewise.
17739 (ix86_hard_regno_mode_ok): Likewise.
17740 (x86_order_regs_for_local_alloc): Likewise.
17741 (ix86_mitigate_rop): Likewise.
17742 (ix86_bnd_prefixed_insn_p): Likewise.
17743 (ix86_mpx_bound_mode): Likewise.
17744 (ix86_make_bounds_constant): Likewise.
17745 (ix86_initialize_bounds): Likewise.
17746 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
17747 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
17748 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
17749 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
17750 (TARGET_CHKP_BOUND_MODE): Likewise.
17751 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
17752 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
17753 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
17754 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
17755 * config/i386/i386.h (TARGET_MPX): Likewise.
17756 (TARGET_MPX_P): Likewise.
17757 (VALID_BND_REG_MODE): Likewise.
17758 (FIRST_BND_REG): Likewise.
17759 (LAST_BND_REG): Likewise.
17760 (enum reg_class): Likewise.
17761 (BND_REG_P): Likewise.
17762 (BND_REGNO_P): Likewise.
17763 (BNDmode): Likewise.
17764 (ADJUST_INSN_LENGTH): Likewise.
17765 * config/i386/i386.md: Likewise.
17766 * config/i386/i386.opt: Likewise.
17767 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
17768 (defined): Likewise.
17769 (LINK_MPX): Likewise.
17770 (MPX_SPEC): Likewise.
17771 (LIBMPX_SPEC): Likewise.
17772 (LIBMPXWRAPPERS_SPEC): Likewise.
17773 (CHKP_SPEC): Likewise.
17774 * config/i386/predicates.md: Likewise.
17775 * dbxout.c (dbxout_type): Likewise.
17776 * doc/extend.texi: Likewise.
17777 * doc/invoke.texi: Likewise.
17778 * doc/md.texi: Likewise.
17779 * doc/tm.texi: Likewise.
17780 * doc/tm.texi.in: Likewise.
17781 * dwarf2out.c (is_base_type): Likewise.
17782 (gen_formal_types_die): Likewise.
17783 (gen_subprogram_die): Likewise.
17784 (gen_type_die_with_usage): Likewise.
17785 (gen_decl_die): Likewise.
17786 (dwarf2out_late_global_decl): Likewise.
17787 * expr.c (expand_assignment): Likewise.
17788 (emit_storent_insn): Likewise.
17789 (store_expr_with_bounds): Likewise.
17790 (store_expr): Likewise.
17791 (expand_expr_real_1): Likewise.
17792 * expr.h (store_expr_with_bounds): Likewise.
17793 * function.c (use_register_for_decl): Likewise.
17794 (struct bounds_parm_data): Likewise.
17795 (assign_parms_augmented_arg_list): Likewise.
17796 (assign_parm_find_entry_rtl): Likewise.
17797 (assign_parm_is_stack_parm): Likewise.
17798 (assign_parm_load_bounds): Likewise.
17799 (assign_bounds): Likewise.
17800 (assign_parms): Likewise.
17801 (expand_function_start): Likewise.
17802 * gcc.c (CHKP_SPEC): Likewise.
17803 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
17804 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
17805 (wrestrict_dom_walker::check_call): Likewise.
17806 * gimple.c (gimple_build_call_from_tree): Likewise.
17807 * gimple.h (enum gf_mask): Likewise.
17808 (gimple_call_with_bounds_p): Likewise.
17809 (gimple_call_set_with_bounds): Likewise.
17810 * gimplify.c (gimplify_init_constructor): Likewise.
17811 * ipa-cp.c (initialize_node_lattices): Likewise.
17812 (propagate_constants_across_call): Likewise.
17813 (find_more_scalar_values_for_callers_subset): Likewise.
17814 * ipa-hsa.c (process_hsa_functions): Likewise.
17815 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
17816 * ipa-icf.c (sem_function::merge): Likewise.
17817 * ipa-inline.c (early_inliner): Likewise.
17818 * ipa-pure-const.c (warn_function_noreturn): Likewise.
17819 (warn_function_cold): Likewise.
17820 (propagate_pure_const): Likewise.
17821 * ipa-ref.h (enum GTY): Likewise.
17822 * ipa-split.c (find_retbnd): Likewise.
17823 (consider_split): Likewise.
17824 (split_function): Likewise.
17825 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
17826 * ipa.c (walk_polymorphic_call_targets): Likewise.
17827 (symbol_table::remove_unreachable_nodes): Likewise.
17828 (process_references): Likewise.
17829 (cgraph_build_static_cdtor_1): Likewise.
17830 * lto-cgraph.c (lto_output_node): Likewise.
17831 (output_refs): Likewise.
17832 (compute_ltrans_boundary): Likewise.
17833 (input_overwrite_node): Likewise.
17834 (input_node): Likewise.
17835 (input_cgraph_1): Likewise.
17836 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
17837 * passes.c (pass_manager::execute_early_local_passes): Likewise.
17838 (class pass_chkp_instrumentation_passes): Likewise.
17839 (make_pass_chkp_instrumentation_passes): Likewise.
17840 * passes.def: Likewise.
17841 * rtl.h (struct GTY): Likewise.
17842 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
17843 * stor-layout.c (layout_type): Likewise.
17844 * symtab.c: Likewise.
17845 * target.def: Likewise.
17846 * targhooks.c (default_chkp_bound_type): Likewise.
17847 (default_chkp_bound_mode): Likewise.
17848 (default_builtin_chkp_function): Likewise.
17849 (default_chkp_function_value_bounds): Likewise.
17850 (default_chkp_make_bounds_constant): Likewise.
17851 (default_chkp_initialize_bounds): Likewise.
17852 * targhooks.h (default_chkp_bound_type): Likewise.
17853 (default_chkp_bound_mode): Likewise.
17854 (default_builtin_chkp_function): Likewise.
17855 (default_chkp_function_value_bounds): Likewise.
17856 (default_chkp_make_bounds_constant): Likewise.
17857 (default_chkp_initialize_bounds): Likewise.
17858 * toplev.c (compile_file): Likewise.
17859 (process_options): Likewise.
17860 * tree-core.h (DEF_BUILTIN): Likewise.
17861 (DEF_BUILTIN_CHKP): Likewise.
17862 * tree-inline.c (declare_return_variable): Likewise.
17863 (remap_gimple_stmt): Likewise.
17864 (copy_bb): Likewise.
17865 (initialize_inlined_parameters): Likewise.
17866 (expand_call_inline): Likewise.
17867 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
17868 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
17869 (make_pass_ipa_chkp_produce_thunks): Likewise.
17870 (make_pass_chkp): Likewise.
17871 (make_pass_chkp_opt): Likewise.
17872 (make_pass_chkp_instrumentation_passes): Likewise.
17873 * tree-pretty-print.c (dump_generic_node): Likewise.
17874 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
17875 * tree-ssa-dce.c (propagate_necessity): Likewise.
17876 (eliminate_unnecessary_stmts): Likewise.
17877 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
17878 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
17879 * tree-ssa-sccvn.h: Likewise.
17880 * tree-ssa-strlen.c (get_string_length): Likewise.
17881 (valid_builtin_call): Likewise.
17882 (adjust_last_stmt): Likewise.
17883 (handle_builtin_strchr): Likewise.
17884 (handle_builtin_strcpy): Likewise.
17885 (handle_builtin_stxncpy): Likewise.
17886 (handle_builtin_memcpy): Likewise.
17887 (handle_builtin_strcat): Likewise.
17888 (strlen_check_and_optimize_stmt): Likewise.
17889 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
17890 * tree-streamer-in.c: Likewise.
17891 * tree-streamer.c (record_common_node): Likewise.
17892 * tree.c (tree_code_size): Likewise.
17893 (wide_int_to_tree_1): Likewise.
17894 (type_contains_placeholder_1): Likewise.
17895 (build_common_tree_nodes): Likewise.
17896 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
17897 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
17898 (POINTER_BOUNDS_P): Likewise.
17899 (BOUNDED_TYPE_P): Likewise.
17900 (BOUNDED_P): Likewise.
17901 (CALL_WITH_BOUNDS_P): Likewise.
17902 (pointer_bounds_type_node): Likewise.
17903 * value-prof.c (gimple_ic): Likewise.
17904 * var-tracking.c (vt_add_function_parameters): Likewise.
17905 * varasm.c (make_decl_rtl): Likewise.
17906 (assemble_start_function): Likewise.
17907 (output_constant): Likewise.
17908 (maybe_assemble_visibility): Likewise.
17909 * varpool.c (ctor_for_folding): Likewise.
17910 * chkp-builtins.def: Remove.
17911 * ipa-chkp.c: Remove.
17912 * ipa-chkp.h: Remove.
17913 * rtl-chkp.c: Remove.
17914 * rtl-chkp.h: Remove.
17915 * tree-chkp-opt.c: Remove.
17916 * tree-chkp.c: Remove.
17917 * tree-chkp.h: Remove.
17919 2018-06-07 Carl Love <cel@us.ibm.com>
17921 * config/rs6000/vsx.md (vextract_fp_from_shorth,
17922 vextract_fp_from_shortl): Add BE support.
17924 2018-06-07 Paul Koning <ni1d@arrl.net>
17926 * compare-elim.c (try_merge_compare): Don't merge compare if
17927 address contains a side effect.
17928 (try_eliminate_compare): Likewise.
17930 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
17932 * config.gcc: Support "tremont".
17933 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
17934 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17936 * config/i386/i386.c (m_TREMONT): Define.
17937 (processor_target_table): Add "tremont".
17938 (PTA_TREMONT): Define.
17939 (ix86_lea_outperforms): Add TARGET_TREMONT.
17940 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
17941 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
17942 and M_INTEL_GOLDMONT_PLUS.
17943 (fold_builtin_cpu): Add "tremont".
17944 (ix86_add_stmt_cost): Add TARGET_TREMONT.
17945 (ix86_option_override_internal): Add "tremont".
17946 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
17947 (processor_type): Add PROCESSOR_TREMONT.
17948 * config/i386/x86-tune.def: Add m_TREMONT.
17949 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
17951 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
17953 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
17954 symbol defined for msp430i* devices to be lower case.
17956 2018-06-07 Richard Biener <rguenther@suse.de>
17958 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
17959 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
17960 Properly wrap signed arithmetic if overflow wraps.
17962 2018-06-07 Jakub Jelinek <jakub@redhat.com>
17964 PR tree-optimization/69615
17965 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
17966 of a cast from a same precision integral SSA_NAME in a bb dominated
17967 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
17968 cast to utype if rhs2 has already a compatible type.
17970 2018-06-07 Richard Biener <rguenther@suse.de>
17972 PR tree-optimization/85935
17973 * graphite-scop-detection.c (find_params_in_bb): Analyze
17974 condition operands with respect to the correct loop. Assert
17975 the analysis doesn't fail.
17977 2018-06-04 Carl Love <cel@us.ibm.com>
17979 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
17980 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
17981 as it is slightly cheaper.
17982 (first_match_or_eos_index_<mode>):
17983 Calculate index using natural element order.
17984 (first_match_index_<mode>):
17985 Calculate index using natural element order.
17986 (first_match_or_eos_index_<mode>):
17987 Calculate index using natural order.
17988 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
17989 for BE and LE modes.
17990 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
17991 P9V_BUILTIN_VCLZLSBB_V16QI.
17992 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
17995 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
17997 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
17998 indentation and line wrap for many prototypes. Add missing
17999 @smallexample directives around block of prototypes for vec_xl and
18002 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
18004 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
18005 track if we pass or return IEEE 128-bit floating point.
18006 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
18007 C++ mangling that is compatible with GCC 8.1.
18008 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
18009 (init_cumulative_args): Note if we pass or return IEEE 128-bit
18010 floating point types.
18011 (rs6000_function_arg_advance_1): Likewise.
18012 (rs6000_mangle_type): Optionally generate mangled names that match
18013 what GCC 8.1 generated for IEEE 128-bit floating point types.
18014 (rs6000_globalize_decl_name): If we have an external function that
18015 passes or returns IEEE 128-bit types, generate a weak reference
18016 from the mangled name used in GCC 8.1 to the current mangled
18018 (rs6000_init_builtins): Make __ibm128 use the long double type if
18019 long double is IBM extended double. Make __float128 use the long
18020 double type if long double is IEEE 128-bit.
18023 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
18024 macro for __ibm128 built-in functions.
18025 (PACK_IF): Add __ibm128 pack/unpack functions.
18026 (UNPACK_IF): Likewise.
18027 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
18028 enable long double built-in functions if long double is IEEE
18029 128-bit floating point.
18030 (rs6000_invalid_builtin): Update long double built-in function
18032 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
18033 functions, adjust the built-in function to use the long double
18034 built-in function if __ibm128 and long double are the same type.
18035 * doc/extend.texi (PowerPC builtins): Update documention for
18036 __builtin_{,un}pack_longdouble. Add documentation for
18037 __builtin_{,un}pack_ibm128.
18039 2018-06-06 Jim Wilson <jimw@sifive.com>
18041 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
18042 (struct machine_function): New field interrupt_mode.
18043 (riscv_handle_type_attribute): New function. Add forward declaration.
18044 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
18045 (riscv_expand_epilogue): Check interrupt_mode field.
18046 (riscv_set_current_function): Check interrupt attribute args and
18047 set interrupt_mode field.
18048 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
18049 (riscv_sret, riscv_uret): New.
18050 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
18051 new arguments to interrupt attribute.
18053 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
18056 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
18057 Don't handle -mcpu=power8 if -mpower9-vector is also used.
18059 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
18061 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
18062 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
18063 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
18064 several redundant entries.
18066 2018-06-06 David Malcolm <dmalcolm@redhat.com>
18068 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
18069 type from "rtx" to "rtx_insn *".
18070 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
18071 for local "call_insn", removing cast.
18072 (ix86_expand_call): Likewise, introducing a "call_insn" local.
18074 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
18076 PR tree-optimization/86066
18077 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
18078 for BIT_INSERT_EXPR stores.
18080 2018-06-06 Richard Biener <rguenther@suse.de>
18082 PR tree-optimization/86062
18083 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
18084 component refs ontop
18085 of to be offsetted base.
18087 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
18089 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
18090 to be static and remove check on interrupt attribute name.
18092 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
18094 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
18095 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
18097 2018-06-05 Steve Ellcey <sellcey@cavium.com>
18100 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
18102 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
18103 Remove second argument, change how error is called.
18104 (aarch64_layout_arg): Remove second argument from
18105 aarch64_err_no_fpadvsimd call.
18106 (aarch64_init_cumulative_args): Ditto.
18107 (aarch64_gimplify_va_arg_expr): Ditto.
18108 * config/aarch64/aarch64.md (mov<mode>): Ditto.
18110 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
18112 * config/i386/i386.md (simple_return_indirect_internal): New expander.
18113 (*simple_return_indirect_internal<mode>): Rename from
18114 simple_return_indirect_internal. Use W mode iterator.
18115 (rstorssp): New expander.
18116 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
18117 (clrssbsy): New expander.
18118 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
18120 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
18122 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
18124 (cmse_check_pointed_object): Likewise.
18126 2018-06-05 Martin Liska <mliska@suse.cz>
18128 PR gcov-profile/47618
18129 * doc/invoke.texi: Document how -fprofile-dir format
18132 2018-06-05 Richard Biener <rguenther@suse.de>
18134 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
18135 removal pretend DOM info isn't available so we do not update
18136 it and only remove edges, not dominated blocks. Actually free
18137 DOM info in case we removed something. Remove unreachable blocks.
18138 (mfb_keep_latches): Work with either DOM info or marked backedges.
18139 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
18140 first. Mark backedges if DOM info isn't available.
18141 (Re-)compute DOM info after cleanup_control_flow_pre.
18143 2018-06-05 Richard Biener <rguenther@suse.de>
18145 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
18146 (locus_discrim_hasher::hash): Adjust.
18147 (locus_discrim_hasher::equal): Likewise.
18148 (next_discriminator_for_locus): Work on line directly.
18149 (same_line_p): Pass in expanded locus1 as well.
18150 (assign_discriminators): Avoid redundant location expansions.
18152 2018-06-05 Richard Biener <rguenther@suse.de>
18154 PR tree-optimization/86046
18155 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
18156 if required after clearing TREE_ADDRESSABLE.
18158 2018-06-05 Richard Biener <rguenther@suse.de>
18160 PR tree-optimization/86047
18161 * tree-ssa-loop.c (for_each_index): Glob handling of all
18162 decls and constants and really handle all of them.
18164 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18167 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
18168 qualifier_void_pointer and qualifier_const_void_pointer.
18169 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
18170 (arm_init_builtins): Handle the above.
18171 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
18172 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
18173 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
18176 2018-06-05 Martin Liska <mliska@suse.cz>
18178 * auto-profile.c (read_autofdo_file): Do not use
18179 gcov_ctr_summary struct.
18180 (afdo_callsite_hot_enough_for_early_inline): Likewise.
18181 * coverage.c (struct counts_entry): Likewise.
18182 (read_counts_file): Read just single summary entry.
18183 (get_coverage_counts): Use gcov_summary struct.
18184 * coverage.h (get_coverage_counts): Likewise.
18185 * gcov-dump.c (dump_working_sets): Likewise.
18186 (tag_summary): Dump just single summary.
18187 * gcov-io.c (gcov_write_summary): Write just histogram
18189 (gcov_read_summary): Read just single summary.
18190 (compute_working_sets): Use gcov_summary struct.
18191 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
18192 of GCOV_COUNTERS_SUMMABLE.
18193 (GCOV_COUNTERS_SUMMABLE): Remove.
18194 (GCOV_FIRST_VALUE_COUNTER): Replace with
18195 GCOV_COUNTER_V_INTERVAL.
18196 (struct gcov_ctr_summary): Remove.
18197 (struct gcov_summary): Directly use fields of former
18199 (compute_working_sets): Use gcov_summary struct.
18200 * gcov.c (read_count_file): Do not use ctrs fields.
18201 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
18203 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
18205 * profile.c: Likewise.
18206 * profile.h: Likewise.
18208 2018-06-05 Martin Liska <mliska@suse.cz>
18210 PR gcov-profile/84846
18211 * gcov.c (output_lines): Print working directory only
18212 in intermediate format.
18214 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
18216 * config/s390/s390-builtin-types.def: Add void function type.
18217 * config/s390/s390-builtins.def: Use the function type for the
18220 2018-06-04 Jim Wilson <jimw@sifive.com>
18222 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
18224 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
18225 and maybe_eh_return. Change regno to unsigned int. Use new args to
18226 handle EH_RETURN_DATA_REGNO registers properly.
18227 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
18228 (riscv_expand_epilogue): Update comment. Change argument name and
18229 type. Update code to use new name and type. Pass new args to
18230 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
18232 * config/riscv/riscv.md (NORMAL_RETURN): New.
18233 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
18234 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
18235 (eh_return): Call gen_eh_return_internal and emit barrier.
18236 (eh_return_internal): Call riscv_expand_epilogue.
18238 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
18240 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
18241 bit_insertion field and declare can_be_merged_into method.
18242 (merged_store_group::can_be_merged_into): New method.
18243 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
18244 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
18245 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
18247 2018-06-04 Richard Biener <rguenther@suse.de>
18249 PR tree-optimization/85955
18250 * builtins.c (fold_builtin_sincos): Convert pointers to
18251 destination to appropriate type before dereferencing.
18253 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
18255 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
18257 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
18259 * expr.c (expand_expr_real_1): Force the operand into memory if
18260 its TYPE_MODE is BLKmode and if there is no integer mode for
18261 the number of bits being extracted.
18263 2018-06-04 Jakub Jelinek <jakub@redhat.com>
18267 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
18268 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
18270 2018-06-04 Richard Biener <rguenther@suse.de>
18272 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
18273 (cleanup_tree_cfg_noloop): ... single caller. Do
18274 start_recording_case_labels later.
18276 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
18278 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
18279 to _IMMINTRIN_H_INCLUDED.
18280 * config/i386/pconfigintrin.h: Ditto.
18281 * config/i386/waitpkgintrin.h: Ditto.
18282 * config/i386/immintrin.h: Add includes for sgxintrin.h,
18283 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
18284 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
18285 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
18286 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
18287 waitpkgintrin.h and cldemoteintrin.h.
18289 2018-06-04 Richard Biener <rguenther@suse.de>
18291 PR tree-optimization/86038
18292 * tracer.c (find_best_successor): Check probability for
18293 being initialized, bail out if not.
18295 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
18298 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
18299 of bits to ignore when comparing architectures.
18301 2018-06-04 Jakub Jelinek <jakub@redhat.com>
18303 PR tree-optimization/69615
18304 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
18305 maximum or minimum of the type, try to merge it also as if
18306 range1 is + [-, x - 1] or + [x + 1, -].
18309 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
18311 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
18313 PR tree-optimization/86034
18314 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
18315 the unsigned bitfield type in a bit insertion sequence if it does not
18316 have a larger precision than the bitfield size.
18317 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
18319 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
18321 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
18323 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
18325 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
18326 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
18327 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
18328 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
18330 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
18332 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
18333 Disable -fdelete-null-pointer-checks for ELF toolchain.
18335 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
18336 Kito Cheng <kito.cheng@gmail.com>
18338 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
18339 (nds32le-*-*, nds32be-*-*): Integrate checking process.
18340 (nds32*-*-*): Add glibc and uclibc conditions.
18341 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
18342 (TARGET_EXCEPT_UNWIND_INFO): Define.
18343 * config/nds32/elf.h: New file.
18344 * config/nds32/linux.h: New file.
18345 * config/nds32/nds32-elf.opt: New file.
18346 * config/nds32/nds32-linux.opt: New file.
18347 * config/nds32/nds32-fp-as-gp.c
18348 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
18349 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
18351 (nds32_asm_file_end): Ditto.
18352 (nds32_print_operand): Ditto.
18353 (nds32_insert_attributes): Ditto.
18354 (nds32_init_libfuncs): New function.
18355 (TARGET_HAVE_TLS): Define.
18356 (TARGET_INIT_LIBFUNCS): Define.
18357 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
18359 (TARGET_ELF): Apply different mcmodel setting.
18360 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
18361 been migrated into elf.h and linux.h files.
18362 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
18363 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
18364 (mcmodel): The content has been migrated into nds32-elf.opt and
18365 nds32-linux.opt files.
18366 * config/nds32/t-elf: New file.
18367 * config/nds32/t-linux: New file.
18369 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
18370 Shiva Chen <shiva0217@gmail.com>
18372 * config/nds32/constants.md (unspec_volatile_element): Add
18373 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
18374 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
18376 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
18377 (make_pass_nds32_fp_as_gp): Declare.
18378 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
18380 (nds32_asm_function_end_prologue): Remove unused asm output.
18381 (nds32_asm_function_begin_epilogue): Remove unused asm output.
18382 (nds32_asm_file_start): Output necessary fp_as_gp information.
18383 (nds32_option_override): Adjust register usage.
18384 (nds32_expand_prologue): Consider fp_as_gp situation.
18385 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
18386 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
18389 (simple_return): Ditto.
18390 (omit_fp_begin): Output special directive for fp_as_gp.
18391 (omit_fp_end): Output special directive for fp_as_gp.
18392 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
18393 mforbid-fp-as-gp): New options.
18395 2018-06-01 Mark Wielaard <mark@klomp.org>
18397 * dwarf2out.c (dwarf2out_finish): Remove generation of
18398 DW_AT_loclists_base.
18400 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
18402 * gimple-ssa-store-merging.c: Include gimple-fold.h.
18403 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
18404 (struct merged_store_group): Add bit_insertion field.
18405 (dump_char_array): Use standard hexadecimal format.
18406 (merged_store_group::merged_store_group): Set bit_insertion to false.
18407 (merged_store_group::apply_stores): Use optimal buffer size. Deal
18408 with BIT_INSERT_EXPR stores. Move up code updating the mask and
18409 also print the mask in the dump file.
18410 (pass_store_merging::gate): Minor tweak.
18411 (imm_store_chain_info::coalesce_immediate): Fix wrong association
18412 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
18413 stores with INTEGER_CST stores.
18414 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
18415 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
18416 and use it throughout. Generate bit insertion sequences if need be.
18417 (pass_store_merging::process_store): Remove redundant condition.
18418 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
18420 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
18422 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
18423 the 128-bit floating point types. Fix function comment.
18425 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18427 * config/aarch64/aarch64-simd.md
18428 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
18430 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
18433 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
18435 PR tree-optimization/85989
18436 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
18438 (backprop::intersect_uses): Check it when deciding whether this
18439 is a backedge reference.
18440 (backprop::process_block): Add each phi to m_visited_phis
18441 after visiting it, then clear it at the end.
18443 2018-06-01 Richard Biener <rguenther@suse.de>
18445 * tree-vectorizer.h (vect_dr_stmt): New function.
18446 (vect_get_load_cost): Adjust.
18447 (vect_get_store_cost): Likewise.
18448 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
18449 Use vect_dr_stmt instead of DR_SMTT.
18450 (vect_record_base_alignments): Likewise.
18451 (vect_calculate_target_alignment): Likewise.
18452 (vect_compute_data_ref_alignment): Likewise and make static.
18453 (vect_update_misalignment_for_peel): Likewise.
18454 (vect_verify_datarefs_alignment): Likewise.
18455 (vector_alignment_reachable_p): Likewise.
18456 (vect_get_data_access_cost): Likewise. Pass down
18457 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
18458 (vect_get_peeling_costs_all_drs): Likewise.
18459 (vect_peeling_hash_get_lowest_cost): Likewise.
18460 (vect_enhance_data_refs_alignment): Likewise.
18461 (vect_find_same_alignment_drs): Likewise.
18462 (vect_analyze_data_refs_alignment): Likewise.
18463 (vect_analyze_group_access_1): Likewise.
18464 (vect_analyze_group_access): Likewise.
18465 (vect_analyze_data_ref_access): Likewise.
18466 (vect_analyze_data_ref_accesses): Likewise.
18467 (vect_vfa_segment_size): Likewise.
18468 (vect_small_gap_p): Likewise.
18469 (vectorizable_with_step_bound_p): Likewise.
18470 (vect_prune_runtime_alias_test_list): Likewise.
18471 (vect_analyze_data_refs): Likewise.
18472 (vect_supportable_dr_alignment): Likewise.
18473 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
18474 (vect_gen_prolog_loop_niters): Likewise.
18475 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
18476 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
18478 (vect_recog_mask_conversion_pattern): Likewise.
18479 (vect_try_gather_scatter_pattern): Likewise.
18480 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
18481 to vect_get_store_cost.
18482 (vect_get_store_cost): Get stmt_info instead of DR.
18483 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
18484 (vect_get_load_cost): Get stmt_info instead of DR.
18486 2018-06-01 Richard Biener <rguenther@suse.de>
18488 PR middle-end/86017
18489 * gimple-fold.c (var_decl_component_p): Also allow offsetted
18490 vars wrapped in MEM_REFs.
18492 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
18494 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18495 Fix subreg tests so that we only return a choice between
18496 GENERAL_REGS and FP_REGS if the original classes included both.
18498 2018-06-01 Richard Biener <rguenther@suse.de>
18501 * tree-ssa-structalias.c (get_function_part_constraint):
18502 Handle NULL fi->decl.
18503 (find_func_aliases_for_call): Properly handle indirect
18504 fi from direct call.
18505 (find_func_clobbers): Likewise.
18506 (ipa_pta_execute): Likewise.
18507 (create_variable_info_for): For functions that are ifunc_resolver
18508 resolve to a varinfo that contains the result of the resolver call.
18509 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
18512 2018-05-31 Michael Collison <michael.collison@arm.com>
18514 * config/aarch64/aarch64.md:
18515 (*fix_to_zero_extenddfdi2): New pattern.
18516 * gcc.target/aarch64/fix_extend1.c: New testcase.
18518 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
18520 PR middle-end/78809
18521 PR middle-end/83026
18522 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
18523 and BUILT_IN_STRNCMP_EQ.
18524 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
18525 BUILT_IN_STRNCMP_EQ.
18526 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
18527 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
18528 (gimple_fold_builtin): Likewise.
18529 * tree-ssa-strlen.c (compute_string_length): New function.
18530 (determine_min_obsize): New function.
18531 (handle_builtin_string_cmp): New function to handle calls to
18532 string compare functions.
18533 (strlen_optimize_stmt): Add handling to builtin string compare
18535 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
18536 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
18537 * tree.c (build_common_builtin_nodes): Add new defines of
18538 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
18540 2018-05-31 Jakub Jelinek <jakub@redhat.com>
18543 * bb-reorder.c (pass_partition_blocks::gate): Return false for
18544 functions with naked attribute.
18546 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
18548 * config/i386/sse.md (avx_vec_concat<mode>):
18549 Substitute concat_tg_mode mode attribute with xtg_mode.
18550 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
18551 (concat_tg_mode): Remove mode attribute.
18553 2018-05-31 Martin Sebor <msebor@redhat.com>
18556 * calls.c (alloc_max_size): Correct a logic error/typo.
18557 Treat excessive arguments as infinite. Warn for invalid arguments.
18558 * doc/invoke.texi (-Walloc-size-larger-than): Update.
18560 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
18563 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
18564 and movx for Haswell.
18566 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
18567 Cesar Philippidis <cesar@codesourcery.com>
18569 PR middle-end/85879
18570 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
18571 when emitting error on private/firstprivate reductions.
18572 * omp-low.c (lower_omp_target): Avoid reference-type processing
18573 on pointers for firstprivate clause.
18575 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
18577 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
18580 * config/aarch64/aarch64-simd.md
18581 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
18582 (aarch64_ld1_x3_<mode>): Likewise
18583 (aarch64_st1x2<VALLDIF:mode>): Likewise
18584 (aarch64_st1_x2_<mode>): Likewise
18585 (aarch64_st1x3<VALLDIF:mode>): Likewise
18586 (aarch64_st1_x3_<mode>): Likewise
18587 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
18588 (vld1_s8_x3): Likewise.
18589 (vld1_u16_x3): Likewise.
18590 (vld1_s16_x3): Likewise.
18591 (vld1_u32_x3): Likewise.
18592 (vld1_s32_x3): Likewise.
18593 (vld1_u64_x3): Likewise.
18594 (vld1_s64_x3): Likewise.
18595 (vld1_f16_x3): Likewise.
18596 (vld1_f32_x3): Likewise.
18597 (vld1_f64_x3): Likewise.
18598 (vld1_p8_x3): Likewise.
18599 (vld1_p16_x3): Likewise.
18600 (vld1_p64_x3): Likewise.
18601 (vld1q_u8_x3): Likewise.
18602 (vld1q_s8_x3): Likewise.
18603 (vld1q_u16_x3): Likewise.
18604 (vld1q_s16_x3): Likewise.
18605 (vld1q_u32_x3): Likewise.
18606 (vld1q_s32_x3): Likewise.
18607 (vld1q_u64_x3): Likewise.
18608 (vld1q_s64_x3): Likewise.
18609 (vld1q_f16_x3): Likewise.
18610 (vld1q_f32_x3): Likewise.
18611 (vld1q_f64_x3): Likewise.
18612 (vld1q_p8_x3): Likewise.
18613 (vld1q_p16_x3): Likewise.
18614 (vld1q_p64_x3): Likewise.
18615 (vst1_s64_x2): Likewise.
18616 (vst1_u64_x2): Likewise.
18617 (vst1_f64_x2): Likewise.
18618 (vst1_s8_x2): Likewise.
18619 (vst1_p8_x2): Likewise.
18620 (vst1_s16_x2): Likewise.
18621 (vst1_p16_x2): Likewise.
18622 (vst1_s32_x2): Likewise.
18623 (vst1_u8_x2): Likewise.
18624 (vst1_u16_x2): Likewise.
18625 (vst1_u32_x2): Likewise.
18626 (vst1_f16_x2): Likewise.
18627 (vst1_f32_x2): Likewise.
18628 (vst1_p64_x2): Likewise.
18629 (vst1q_s8_x2): Likewise.
18630 (vst1q_p8_x2): Likewise.
18631 (vst1q_s16_x2): Likewise.
18632 (vst1q_p16_x2): Likewise.
18633 (vst1q_s32_x2): Likewise.
18634 (vst1q_s64_x2): Likewise.
18635 (vst1q_u8_x2): Likewise.
18636 (vst1q_u16_x2): Likewise.
18637 (vst1q_u32_x2): Likewise.
18638 (vst1q_u64_x2): Likewise.
18639 (vst1q_f16_x2): Likewise.
18640 (vst1q_f32_x2): Likewise.
18641 (vst1q_f64_x2): Likewise.
18642 (vst1q_p64_x2): Likewise.
18643 (vst1_s64_x3): Likewise.
18644 (vst1_u64_x3): Likewise.
18645 (vst1_f64_x3): Likewise.
18646 (vst1_s8_x3): Likewise.
18647 (vst1_p8_x3): Likewise.
18648 (vst1_s16_x3): Likewise.
18649 (vst1_p16_x3): Likewise.
18650 (vst1_s32_x3): Likewise.
18651 (vst1_u8_x3): Likewise.
18652 (vst1_u16_x3): Likewise.
18653 (vst1_u32_x3): Likewise.
18654 (vst1_f16_x3): Likewise.
18655 (vst1_f32_x3): Likewise.
18656 (vst1_p64_x3): Likewise.
18657 (vst1q_s8_x3): Likewise.
18658 (vst1q_p8_x3): Likewise.
18659 (vst1q_s16_x3): Likewise.
18660 (vst1q_p16_x3): Likewise.
18661 (vst1q_s32_x3): Likewise.
18662 (vst1q_s64_x3): Likewise.
18663 (vst1q_u8_x3): Likewise.
18664 (vst1q_u16_x3): Likewise.
18665 (vst1q_u32_x3): Likewise.
18666 (vst1q_u64_x3): Likewise.
18667 (vst1q_f16_x3): Likewise.
18668 (vst1q_f32_x3): Likewise.
18669 (vst1q_f64_x3): Likewise.
18670 (vst1q_p64_x3): Likewise.
18672 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
18674 * config/msp430/msp430.c (msp430_output_labelref): Prepend
18675 user_label_prefix to name.
18677 * tree-core.h: Update comment about the format of NAME string
18678 passed to handler in attribute_spec.
18680 * config/msp430/msp430.md: Remove erroneous subreg expression from
18681 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
18682 zero_extend{q,h}isi2.
18684 2018-05-30 Borislav Petkov <bp@suse.de>
18686 * doc/extend.texi: Document some architecture specific
18687 constraints and sort entries.
18689 2018-05-30 Martin Sebor <msebor@redhat.com>
18691 PR middle-end/85369
18692 * builtins.c (expand_builtin_stpcpy_1): New function.
18693 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
18694 only if the former succeeds.
18696 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
18698 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
18701 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
18703 * doc/invoke.texi (-flinker-output): Document
18705 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
18707 * passes.c (ipa_write_summaries): Only modify statements if body
18709 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
18710 incrementally linking.
18711 (ipa_passes): Likewise.
18712 * lto-cgraph.c (lto_output_node): When incrementally linking do not
18713 pass down resolution info.
18714 * common.opt (flag_incremental_link): Update info.
18715 * gcc.c (plugin specs): Turn flinker-output=* to
18716 -plugin-opt=-linker-output-known
18717 * toplev.c (compile_file): Also cut compilation when doing incremental
18719 * flag-types. (enum lto_partition_model): Add
18720 LTO_LINKER_OUTPUT_NOLTOREL.
18721 (invoke.texi): Add -flinker-output docs.
18722 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
18723 link same way as WPA; do not stream in dead initializers.
18725 * dwarf2out.c (dwarf2out_die_ref_for_decl,
18726 darf2out_register_external_decl): Support incremental link.
18728 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
18730 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
18732 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
18734 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
18735 it down to simple_object_copy_lto_debug_sections.
18736 (run_gcc): Determine incremental LTO link time and configure
18737 lto1 into non-wpa mode, disable renaming of debug sections.
18739 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
18741 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
18742 descriptions of various incorrectly documented functions.
18744 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
18747 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
18748 address check not strict.
18750 2018-05-30 Richard Biener <rguenther@suse.de>
18752 PR tree-optimization/85964
18753 * tracer.c (better_p): Drop initialized count check, we only
18754 call the function with initialized counts now.
18755 (find_best_successor): Do find a best edge if one
18756 has uninitialized count.
18757 (find_best_predecessor): Likewise. Do BB frequency check only
18758 if count is initialized.
18760 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
18762 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
18763 (aarch64_ldrstr_offset_compare): New.
18764 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
18765 load/store orderings.
18766 (aarch64_gen_adjusted_ldpstp): Likewise.
18768 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
18770 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18771 Check for subset of GENERAL_REGS and FP_REGS.
18772 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
18775 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
18777 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
18778 and wi::to_poly_offset. Add the current offset and then check
18779 whether the sum fits, rather than using an unchecked addition of
18780 a checked term. Check for a shwi rather than a uhwi.
18781 * expr.c (get_bit_range): Use tree_to_poly_uint64.
18782 (store_constructor): Use poly_int_tree_p.
18783 (expand_expr_real_1): Likewise.
18784 * function.c (assign_temp): Likewise.
18785 * fold-const.c (const_binop): Use poly_int_tree_p and
18786 wi::to_poly_offset.
18787 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
18789 * ipa-icf-gimple.c (func_checker::compare_operand): Use
18790 to_poly_offset for MEM offsets.
18791 * ipa-icf.c (sem_variable::equals): Likewise.
18792 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
18793 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
18794 wi::to_poly_offset for BIT_FIELD_REF offsets.
18795 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
18796 wi::to_poly_offset.
18797 * var-tracking.c (emit_note_insn_var_location): Use
18798 tree_to_poly_uint64.
18800 2018-05-29 Jim Wilson <jimw@sifive.com>
18802 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
18804 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
18807 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
18808 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
18810 (sse4_1_round<mode>2): Use nonimmediate_operand
18811 for operand 1 predicate.
18813 2018-05-29 Martin Sebor <msebor@redhat.com>
18814 Richard Biener <rguenther@suse.de>
18817 * calls.c (get_size_range): Call determine_value_range instead
18818 of get_value_range..
18819 * tree-vrp.h (determine_value_range): Declared new function.
18820 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
18822 2018-05-29 Richard Biener <rguenther@suse.de>
18824 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
18825 sure to use non-pattern stmts for get_earlier_stmt arguments.
18826 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
18827 called on pattern stmts.
18828 (get_later_stmt): Likewise.
18830 2018-05-29 Martin Liska <mliska@suse.cz>
18832 PR gcov-profile/85759
18833 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
18836 2018-05-29 Jakub Jelinek <jakub@redhat.com>
18838 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
18840 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
18844 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
18845 VEC_PACK_FLOAT_EXPR): New tree codes.
18846 * tree-pretty-print.c (op_code_prio): Handle
18847 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
18848 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
18849 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
18850 * tree-inline.c (estimate_operator_cost): Likewise.
18851 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
18852 * fold-const.c (const_binop): Likewise.
18853 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
18854 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
18855 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
18856 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
18857 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
18858 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
18859 * expr.c (expand_expr_real_2): Likewise.
18860 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
18861 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
18862 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
18864 * optabs.c (expand_widen_pattern_expr): For
18865 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
18866 sign from result type rather than operand's type.
18867 (expand_binop_directly): For vec_packu_float_optab and
18868 vec_packs_float_optab allow result type to be different from operand's
18870 * optabs-tree.c (optab_for_tree_code): Handle
18871 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
18872 VEC_PACK_FLOAT_EXPR. Formatting fixes.
18873 * tree-vect-generic.c (expand_vector_operations_1): Handle
18874 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
18875 VEC_PACK_FLOAT_EXPR.
18876 * tree-vect-stmts.c (supportable_widening_operation): Handle
18878 (supportable_narrowing_operation): Handle FLOAT_EXPR.
18879 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
18880 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
18881 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
18882 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
18884 (vec_pack<floatprefix>_float_<mode>): New expander.
18885 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
18887 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
18888 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
18889 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
18890 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
18891 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
18893 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
18894 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
18895 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
18896 VEC_PACK_FLOAT_EXPR): Document.
18898 2018-05-29 Richard Biener <rguenther@suse.de>
18900 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
18902 (stmt_vec_info_vec): Make pointer.
18903 (init_stmt_vec_info_vec): Remove.
18904 (free_stmt_vec_info_vec): Likewise.
18905 (set_stmt_vec_info_vec): New function.
18906 (free_stmt_vec_infos): Likewise.
18907 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
18908 (set_vinfo_for_stmt): Likewise.
18909 (get_earlier_stmt): Likewise.
18910 (get_later_stmt): Likewise.
18911 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
18912 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
18913 (vec_info::~vec_info): Free stmt_vec_infos.
18914 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
18915 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
18916 (pass_slp_vectorize::execute): Likewise.
18917 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
18918 (free_stmt_vec_info_vec): Likewise.
18919 (set_stmt_vec_info_vec): New function.
18920 (free_stmt_vec_infos): Likewise.
18921 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
18922 the global stmt_vec_info_vec.
18923 * tree-parloops.c (gather_scalar_reductions): Use
18924 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
18927 2018-05-29 Richard Biener <rguenther@suse.de>
18929 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
18931 2018-05-29 Martin Liska <mliska@suse.cz>
18932 David Malcolm <dmalcolm@redhat.com>
18934 * vec.c (test_reverse): New.
18935 (vec_c_tests): Add new test.
18936 * vec.h (vl_ptr>::reverse): New function.
18938 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
18940 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
18942 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
18945 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
18947 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
18949 2018-05-28 Richard Biener <rguenther@suse.de>
18951 PR tree-optimization/85933
18952 * tree-vect-data-refs.c (vect_record_base_alignments): Only
18953 look at stmts marked as vectorizable.
18955 2018-05-28 Richard Biener <rguenther@suse.de>
18957 PR tree-optimization/85934
18958 * tree-vect-generic.c (expand_vector_operations_1): Hoist
18959 vector boolean check before scalar optimization.
18961 2018-05-28 Jakub Jelinek <jakub@redhat.com>
18963 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
18966 2018-05-28 Mark Wielaard <mark@klomp.org>
18968 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
18969 if it is an expression containing a minus sign.
18971 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
18973 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
18975 2018-05-27 Paul Koning <ni1d@arrl.net>
18977 * config/pdp11/pdp11.md (truncsihi2): Remove.
18979 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
18980 Chung-Ju Wu <jasonwucj@gmail.com>
18982 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
18984 (unaligned_store_dw): Ditto.
18985 * config/nds32/nds32-memory-manipulation.c
18986 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
18987 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
18988 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
18989 (emit_setmem_word_loop): Rename to ...
18990 (emit_setmem_doubleword_loop): ... this.
18991 (nds32_gen_dup_4_byte_to_word_value): New function.
18992 (nds32_gen_dup_8_byte_to_double_word_value): New function.
18993 (nds32_expand_setmem_loop): Refine implementation.
18994 (nds32_expand_setmem_loop_v3m): Ditto.
18995 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
18998 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
19000 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
19002 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
19004 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
19005 (nds32_init_machine_status): Initialize machine->attr_naked_p and
19006 machine->attr_no_prologue_p.
19007 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
19008 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
19009 (nds32_expand_epilogue): Consider attr_naked_p.
19010 (nds32_expand_epilogue_v3pop): Likewise.
19011 (nds32_can_use_return_insn): Likewise.
19012 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
19013 attr_no_prologue_p fields.
19014 * config/nds32/nds32.opt (mret-in-naked-func): New option.
19016 2018-05-27 Jakub Jelinek <jakub@redhat.com>
19019 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
19021 * config/i386/sse.md
19022 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
19024 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
19026 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
19028 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
19030 (*<floatsuffix>floatv2div2sf2): Rename to ...
19031 (*float<floatunssuffix>v2div2sf2): ... this.
19032 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
19033 (float<floatunssuffix>v2div2sf2_mask): ... this.
19034 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
19035 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
19036 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
19038 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
19040 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
19042 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
19044 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
19046 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
19048 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
19049 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
19050 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
19051 gen_ufix_truncv8dfv8si2.
19052 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
19053 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
19054 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
19055 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
19056 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
19057 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
19058 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
19059 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
19061 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
19065 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
19067 2018-05-25 Jim Wilson <jimw@sifive.com>
19069 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
19070 * config/riscv/riscv.c (struct machine_function): Add
19071 interrupt_handler_p and attribute_checked_p fields.
19072 (riscv_attribute_table): Add interrupt.
19073 (riscv_interrupt_type_p): New.
19074 (riscv_save_reg_p): Save extra regs for interrupt handler.
19075 (riscv_use_save_libcall): Return false for interrupt handler.
19076 (riscv_first_stack_step): Add forward declaration.
19077 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
19078 for interrupt handler with large frame. Use it for saved reg list.
19079 (riscv_expand_prologue): Move flag_stack_usage_info support to
19080 eliminate duplication.
19081 (riscv_expand_epilogue): Generate mret for interrupt handler.
19082 (riscv_epilogue_uses): New.
19083 (riscv_can_use_return_insn): Return false for interrupt handler.
19084 (riscv_function_ok_for_sibcall): Likewise.
19085 (riscv_set_current_function): Add interrupt handler support.
19086 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
19087 * config/riscv/riscv.md (UNSPECV_MRET): New.
19089 (riscv_frflags, riscv_fsflags): Use tab after opcode.
19091 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
19093 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
19095 PR tree-optimization/85712
19096 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
19097 this candidate has already been replaced in-situ by a copy.
19099 2018-05-25 Jason Merrill <jason@redhat.com>
19101 PR c++/80485 - inline function non-zero address.
19102 * symtab.c (nonzero_address): Check DECL_COMDAT.
19104 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
19107 * config/alpha/alpha.md (ashlsi3): New insn pattern.
19108 (*ashlsi_se): Rename from *ashldi_se. Define as sign
19109 extension of SImode operation. Use const123_operand predicate.
19110 (*saddsi_1): Remove.
19111 (*saddl_se_1): Ditto.
19112 (*ssubsi_1): Ditto.
19113 (*ssubl_se_1): Ditto.
19114 * config/alpha/predicates.md (const123_operand): New predicate.
19115 * config/alpha/constraints.md (P): Use IN_RANGE.
19117 2018-05-25 Richard Biener <rguenther@suse.de>
19119 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
19121 (ref_maybe_used_by_stmt_p): Likewise.
19122 (stmt_may_clobber_ref_p): Likewise.
19123 (stmt_may_clobber_ref_p_1): Likewise.
19124 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
19126 (ref_maybe_used_by_stmt_p): Likewise.
19127 (stmt_may_clobber_ref_p): Likewise.
19128 (stmt_may_clobber_ref_p_1): Likewise.
19129 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
19130 the alias oracle to disambiguate DRs with stmts DR analysis
19132 (vect_analyze_data_refs): Do not give up on not analyzable
19133 DRs for BB vectorization. Remove code truncating the dataref
19136 2018-05-25 Jakub Jelinek <jakub@redhat.com>
19139 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
19140 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
19141 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
19143 2018-05-25 Richard Biener <rguenther@suse.de>
19145 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
19146 function, combining stmt data ref gathering and fatal analysis
19148 (vect_analyze_data_refs): Remove now redudnant code and simplify.
19149 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
19150 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
19151 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
19152 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
19154 2018-05-25 Bin Cheng <bin.cheng@arm.com>
19156 PR tree-optimization/85720
19157 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
19158 SCC if all partitions are builtins.
19159 (version_loop_by_alias_check): New parameter. Generate cancelable
19160 runtime alias check if all partitions are builtins.
19161 (distribute_loop): Update call to above function.
19163 2018-05-25 Bin Cheng <bin.cheng@arm.com>
19165 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
19166 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
19167 (parm_default_def_partition_arg): Ditto.
19168 (set_parm_default_def_partition): Ditto.
19169 (get_parm_default_def_partitions): Ditto and make it static.
19170 (get_undefined_value_partitions): Ditto and make it static.
19171 (remove_ssa_form): Refactor call to init_var_map here.
19172 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
19173 computation for loop region.
19174 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
19175 (register_default_def): Delete.
19176 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
19177 (parm_default_def_partition_arg): Ditto.
19178 (set_parm_default_def_partition): Ditto.
19179 (get_parm_default_def_partitions): Ditto and make it static.
19180 (get_undefined_value_partitions): Ditto and make it static.
19181 (coalesce_with_default, coalesce_with_default): Update comment.
19182 (create_coalesce_list_for_region): New func factored out from
19183 create_outofssa_var_map.
19184 (populate_coalesce_list_for_outofssa): New func factored out from
19185 create_outofssa_var_map and coalesce_ssa_name.
19186 (create_outofssa_var_map): Delete.
19187 (coalesce_ssa_name): Refactor to support live range computation.
19188 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
19189 (get_parm_default_def_partitions): Delete.
19190 (get_undefined_value_partitions): Ditto.
19191 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
19192 computation for loop region.
19193 (new_tree_live_info, loe_visit_block): Ditto.
19194 (live_worklist, set_var_live_on_entry): Ditto.
19195 (calculate_live_on_exit, verify_live_on_entry): Ditto.
19196 * tree-ssa-live.h (struct _var_map): New fields.
19197 (init_var_map): Change decl.
19198 (region_contains_p): New.
19200 2018-05-25 Bin Cheng <bin.cheng@arm.com>
19202 * tree-ssa-live.h (live_merge_and_clear): Delete.
19204 2018-05-25 Richard Biener <rguenther@suse.de>
19207 * tree-dump.c (dequeue_and_dump): Remove access to removed
19208 operand 2 of a SWITCH_EXPR.
19210 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
19212 * doc/sourcebuild.texi (vect_double_cond_arith): Include
19213 multiplication and division.
19214 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
19215 (cond_udiv@var{m}, cond_umod@var{m}): Document.
19216 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
19217 (cond_udiv_optab, cond_umod_optab): New optabs.
19218 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
19219 (IFN_COND_RDIV): New internal functions.
19220 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
19221 TRUNC_MOD_EXPR and RDIV_EXPR.
19222 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
19223 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
19225 (SVE_INT_BINARY): Include mult.
19226 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
19227 (optab, sve_int_op): Handle mult.
19228 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
19230 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
19231 for SVE_INT_BINARY_SD.
19233 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
19235 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
19236 (optab, sve_int_op): Handle div and udiv.
19237 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
19238 for SVE_INT_BINARY_SD.
19239 (*<optab><mode>3): New insn for the same.
19241 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
19243 * tree-vect-patterns.c: Include predict.h.
19244 (vect_recog_divmod_pattern): Restrict check for division support
19245 to when optimizing for size.
19247 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
19249 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
19250 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
19251 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
19252 (gimple_match_op::set_op): Likewise.
19253 (gimple_resimplify4): Declare.
19254 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
19255 (expr::gen_transform): Likewise.
19256 (decision_tree::gen): Generate a simplification routine for 4 operands.
19257 * gimple-match-head.c (gimple_simplify): Add an overload for
19258 4 operands. In the top-level function, handle up to 4 call
19259 arguments and call gimple_resimplify4.
19260 (gimple_resimplify4): New function.
19261 (build_call_internal): Pass a fourth operand.
19262 (maybe_push_to_seq): Likewise.
19263 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
19264 Fold VEC_COND_EXPRs of an operation and a default value into
19265 an IFN_COND_* function if possible.
19266 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
19268 (SVE_COND_FP_BINARY): Include them.
19269 (optab, sve_fp_op): Handle them.
19270 (SVE_INT_BINARY_REV): New code iterator.
19271 (SVE_COND_FP_BINARY_REV): New int iterator.
19272 (commutative): New int attribute.
19273 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
19275 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
19277 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
19278 (*cond_<optab><mode>): New patterns for reversed operands.
19280 2018-05-25 Richard Biener <rguenther@suse.de>
19282 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
19283 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
19284 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
19285 (STMT_VINFO_GROUPED_ACCESS): Adjust.
19286 * tree-vect-data-refs.c (everywhere): Adjust users.
19287 * tree-vect-loop.c (everywhere): Likewise.
19288 * tree-vect-slp.c (everywhere): Likewise.
19289 * tree-vect-stmts.c (everywhere): Likewise.
19290 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
19292 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19294 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
19296 (gcc_cv_as_section_exclude): ... this.
19297 Try Solaris as #exclude syntax.
19298 * configure: Regenerate.
19299 * config.in: Regenerate.
19300 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
19302 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
19303 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
19305 * varasm.c (default_elf_asm_named_section): Don't check if
19306 HAVE_GAS_SECTION_EXCLUDE is defined.
19308 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
19310 * doc/md.texi: Update the documentation of the cond_* optabs
19311 to mention the new final operand. Fix GET_MODE_NUNITS call.
19312 Describe the scalar case too.
19313 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
19314 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
19316 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
19317 (get_conditional_internal_fn): Update comment.
19318 * tree-vect-loop.c (vectorizable_reduction): Pass the original
19319 accumulator value as a final argument to conditional functions.
19320 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
19321 a define_expand and add an "else" operand. Assert for now that
19322 the else operand is equal to operand 2. Use SVE_INT_BINARY and
19323 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
19324 (*cond_<optab><mode>): New patterns.
19325 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
19326 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
19327 (UNSPEC_COND_EOR): Delete.
19328 (optab): Remove associated mappings.
19329 (SVE_INT_BINARY): New code iterator.
19330 (sve_int_op): Remove int attribute and add "minus" to the code
19332 (SVE_COND_INT_OP): Delete.
19333 (SVE_COND_FP_OP): Rename to...
19334 (SVE_COND_FP_BINARY): ...this.
19336 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
19338 * optabs.c (can_reuse_operands_p): New function.
19339 (maybe_legitimize_operands): Try to reuse the results for
19342 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
19344 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
19345 Add {q} suffix to insn mnemonic.
19347 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
19349 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
19350 (msp430_warn_func_return): New.
19352 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
19354 * fold-const.c (tree_nonzero_bits): New function.
19355 * fold-const.h (tree_nonzero_bits): Likewise.
19356 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
19357 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
19359 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
19363 * varasm.c (assemble_alias): Check ifunc_resolver only on
19366 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
19369 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
19370 when memory input operand is handled.
19372 2018-05-24 Luis Machado <luis.machado@linaro.org>
19374 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
19376 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
19378 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
19380 * match.pd: Delay FMA folds until after vectorization.
19382 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
19385 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
19386 address check not strict.
19388 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
19390 * gimple-match.h (gimple_match_op): New class.
19391 (mprts_hook): Replace parameters with a gimple_match_op *.
19392 (maybe_build_generic_op): Likewise.
19393 (gimple_simplified_result_is_gimple_val): Replace parameters with
19394 a const gimple_match_op *.
19395 (gimple_simplify): Replace code_helper * and tree * parameters with
19396 a gimple_match_op * parameter.
19397 (gimple_resimplify1): Replace code_helper *, tree and tree *
19398 parameters with a gimple_match_op * parameter.
19399 (gimple_resimplify2): Likewise.
19400 (gimple_resimplify3): Likewise.
19401 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
19402 parameters with a gimple_match_op * parameter.
19403 * gimple-match-head.c (gimple_simplify): Change prototypes of
19404 auto-generated functions to take a gimple_match_op * instead of
19405 separate code_helper * and tree * parameters. Make the same
19406 change in the top-level overload and update calls to the
19407 gimple_resimplify routines. Update calls to the auto-generated
19408 functions and to maybe_push_res_to_seq in the publicly-facing
19409 operation-specific gimple_simplify overloads.
19410 (gimple_match_op::MAX_NUM_OPS): Define.
19411 (gimple_resimplify1): Replace rcode and ops with a single res_op
19412 parameter. Update call to gimple_simplify.
19413 (gimple_resimplify2): Likewise.
19414 (gimple_resimplify3): Likewise.
19415 (mprts_hook): Replace parameters with a gimple_match_op *.
19416 (maybe_build_generic_op): Likewise.
19417 (build_call_internal): Replace type, nargs and ops with
19418 a gimple_match_op *.
19419 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
19420 with a single gimple_match_op *. Update calls to mprts_hook,
19421 build_call_internal and gimple_simplified_result_is_gimple_val.
19422 Factor out code that is common to the tree_code and combined_fn cases.
19423 * genmatch.c (expr::gen_transform): Replace tem_code and
19424 tem_ops with a gimple_match_op called tem_op. Update calls
19425 to the gimple_resimplify functions and maybe_push_res_to_seq.
19426 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
19427 res_ops. Update call to the gimple_resimplify functions.
19428 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
19429 (decision_tree::gen): Make the functions take a gimple_match_op *
19430 called res_op instead of separate res_code and res_ops parameters.
19431 Update call accordingly.
19432 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
19433 and ops with a single res_op parameter. Update calls to
19434 maybe_build_generic_op and maybe_push_res_to_seq.
19435 (fold_stmt_1): Update calls to gimple_simplify and
19436 replace_stmt_with_simplification.
19437 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
19438 and gimple_simplified_result_is_gimple_val.
19439 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
19441 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
19442 with a gimple_match_op *.
19443 (vn_nary_build_or_lookup): Likewise. Update call to
19444 vn_nary_build_or_lookup_1.
19445 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
19446 gimple_match_op *. Update calls to the gimple_resimplify routines
19447 and to gimple_simplified_result_is_gimple_val.
19448 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
19449 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
19450 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
19451 (visit_nary_op): Likewise.
19452 (visit_reference_op_load): Likewise.
19454 2018-05-23 Luis Machado <luis.machado@linaro.org>
19456 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
19457 modifier for printing the step amount.
19459 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19462 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
19465 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
19467 * doc/sourcebuild.texi (Endianness): New subsubsection.
19469 2018-05-23 Luis Machado <luis.machado@linaro.org>
19471 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
19472 <prefetch_dynamic_strides>: New const bool field.
19473 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
19474 prefetch_dynamic_strides.
19475 (exynosm1_prefetch_tune): Likewise.
19476 (thunderxt88_prefetch_tune): Likewise.
19477 (thunderx_prefetch_tune): Likewise.
19478 (thunderx2t99_prefetch_tune): Likewise.
19479 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
19481 (aarch64_override_options_internal): Update to set
19482 PARAM_PREFETCH_DYNAMIC_STRIDES.
19483 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
19484 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
19485 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
19486 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
19487 prefetch-dynamic-strides setting.
19489 2018-05-23 Luis Machado <luis.machado@linaro.org>
19491 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
19492 <minimum_stride>: New const int field.
19493 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
19494 minimum_stride field defaulting to -1.
19495 (exynosm1_prefetch_tune): Likewise.
19496 (thunderxt88_prefetch_tune): Likewise.
19497 (thunderx_prefetch_tune): Likewise.
19498 (thunderx2t99_prefetch_tune): Likewise.
19499 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
19500 <default_opt_level>: Set to 3.
19501 (aarch64_override_options_internal): Update to set
19502 PARAM_PREFETCH_MINIMUM_STRIDE.
19503 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
19504 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
19505 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
19506 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
19507 stride is constant and is below the minimum stride threshold.
19509 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19511 * config/arm/arm-cpus.in (mode26): Delete.
19512 (armv4): Delete mode26 reference.
19513 * config/arm/arm.c (arm_configure_build_target): Delete use of
19516 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
19518 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
19520 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
19521 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
19523 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
19524 (floatunsdidf2): Ditto.
19526 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
19528 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
19529 (fixuns_trunc<mode>si2_avx512f): Ditto.
19530 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
19531 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
19532 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
19534 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
19536 PR rtl-optimization/79985
19537 * df-scan.c (df_insn_refs_collect): Remove special case for
19538 global registers and asm statements.
19540 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
19542 * extend.texi (Global Register Variables): Rewrite the bullet list.
19543 Note that the register is available for allocation. Note that access
19544 via inline asm must use constraints. Add note about async-signal
19545 handlers. Remove paragraph about automagic register selection.
19547 2018-05-23 Richard Biener <rguenther@suse.de>
19549 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
19550 of fixed offset from memset VN.
19552 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
19554 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
19555 first_interp field.
19556 (alloc_cand_and_find_basis): Initialize first_interp field.
19557 (slsr_process_mul): Modify first_interp field.
19558 (slsr_process_add): Likewise.
19559 (slsr_process_cast): Modify first_interp field for each new
19561 (slsr_process_copy): Likewise.
19562 (dump_candidate): Dump first_interp field.
19563 (replace_mult_candidate): Process all interpretations, not just
19565 (replace_rhs_if_not_dup): Likewise.
19566 (replace_one_candidate): Likewise.
19568 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
19570 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
19572 (aarch64_needs_frame_chain): New function.
19573 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
19575 2018-05-23 Sudakshina Das <sudi.das@arm.com>
19578 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
19579 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
19580 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
19581 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
19582 as true for strict-align.
19583 (aarch64_can_inline_p): Perform checks even when callee has no
19584 attributes to check for strict alignment.
19585 * doc/extend.texi (AArch64 Function Attributes): Document
19587 * doc/invoke.texi: (AArch64 Options): Likewise.
19589 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
19591 PR tree-optimization/85853
19592 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
19593 the handling of the root of the node to...
19594 (vect_slp_analyze_node_operations_1): ...this new function,
19595 and run the whole thing with the child nodes' def types
19596 set according to their SLP node's def type.
19598 2018-05-23 Richard Biener <rguenther@suse.de>
19600 PR middle-end/85874
19601 * tree-data-ref.c (create_runtime_alias_checks): Defer
19602 and ignore overflow warnings.
19604 2018-05-23 Yury Gribov <tetra2005@gmail.com>
19606 PR tree-optimization/85822
19607 * tree-vrp.c (is_masked_range_test): Fix handling of negative
19610 2018-05-23 Richard Biener <rguenther@suse.de>
19612 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
19613 memset constants via native_interpret_expr.
19615 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
19618 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
19620 (cgraph_node::create_alias): Likewise.
19621 (cgraph_node::get_availability): Check ifunc_resolver instead
19622 of looking up ifunc attribute.
19623 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
19624 * varasm.c (do_assemble_alias): Likewise.
19625 (assemble_alias): Likewise.
19626 (default_binds_local_p_3): Likewise.
19627 * cgraph.h (cgraph_node): Add ifunc_resolver.
19628 (cgraph_node::only_called_directly_or_aliased_p): Return false
19629 for IFUNC resolver.
19630 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
19632 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
19633 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
19634 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
19635 instead of looking up ifunc attribute.
19637 2018-05-22 Luis Machado <luis.machado@linaro.org>
19639 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
19641 2018-05-22 Martin Sebor <msebor@redhat.com>
19643 PR middle-end/85359
19644 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
19645 only when expasion succeeds.
19646 (expand_builtin_strcmp): Same.
19647 (expand_builtin_strncmp): Same.
19649 2018-05-22 Martin Sebor <msebor@redhat.com>
19651 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
19653 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
19654 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19656 * config/aarch64/aarch64-ldpstp.md: Replace uses of
19657 aarch64_mem_pair_operand with memory_operand and delete operand swapping
19659 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
19660 Add check for legitimate_address.
19661 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
19662 (aarch64_swap_ldrstr_operands): New.
19663 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
19666 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
19667 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19669 * config/aarch64/aarch64.md: New patterns to generate stp
19671 (store_pair_sw, store_pair_dw): New patterns to generate stp for
19672 single words and double words.
19673 (load_pair_sw, load_pair_dw): Likewise.
19674 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
19676 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
19678 * config/aarch64/aarch64-ldpstp.md: Modify peephole
19679 for different mode ldpstp and add peephole for merged zero stores.
19680 Likewise for loads.
19681 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
19683 (aarch64_gen_store_pair): Rename calls to match new patterns.
19684 (aarch64_gen_load_pair): Rename calls to match new patterns.
19685 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
19686 (load_pair<DREG:mode><DREG2:mode>): ... This.
19687 (store_pair<mode>): Rename to...
19688 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
19689 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
19690 New mode iterators.
19691 (V_INT_EQUIV): Handle SImode.
19692 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
19695 2018-05-22 Martin Sebor <msebor@redhat.com>
19698 * calls.c (maybe_warn_nonstring_arg): Use string length to set
19699 or ajust the presumed bound on an operation to avoid unnecessary
19702 2018-05-22 Martin Sebor <msebor@redhat.com>
19704 PR tree-optimization/85826
19705 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
19706 assuming that a DECL necesarily has a constant size.
19708 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
19710 PR middle-end/85862
19711 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
19713 2018-05-22 Richard Biener <rguenther@suse.de>
19715 PR tree-optimization/85834
19716 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
19717 non-constant and non-zero memset arguments.
19719 2018-05-22 Martin Liska <mliska@suse.cz>
19722 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
19724 2018-05-22 Richard Biener <rguenther@suse.de>
19726 PR tree-optimization/85863
19727 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
19728 comparisons when vectype is specified.
19729 (vectorizable_condition): Do not specify vectype for
19730 vect_is_simple_cond when SLP vectorizing.
19732 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
19735 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
19736 define __ibm128 as long double.
19737 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
19738 as a distinct type when IEEE 128-bit support is enabled.
19739 (init_float128_ieee): Fix up conversions between IFmode and IEEE
19740 128-bit types to use the correct functions.
19741 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
19742 convert between 128-bit floating point types that have different
19743 modes but the same representation, instead of using gen_lowpart to
19745 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
19747 (IFKF_reg): New attributes to give the register constraints for
19749 (extend<mode>tf2_internal): New insns to mark an explicit
19750 conversion between 128-bit floating point types that have a
19751 different mode but share the same representation.
19753 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
19755 PR tree-optimization/85814
19756 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
19757 a null return from get_strinfo when unsharing the next
19758 strinfo in the chain.
19760 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
19763 * varasm.c (weak_finish): Clean up weak_decls.
19765 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19767 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
19768 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
19769 UNSPEC_UADALP values.
19770 * config/aarch64/iterators.md (ABAL): New int iterator.
19773 (sur): Add mappings for the above.
19774 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
19776 (aarch64_<sur>abal<mode>_4): Likewise.
19777 (aarch64_<sur>adalp<mode>_3): Likewise.
19778 (<sur>sadv16qi): New define_expand.
19780 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
19782 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
19783 (*movdf_internal): Ditto.
19784 (*rcpsf2_sse): Ditto.
19785 (*rsqrtsf2_sse): Ditto.
19786 (*sqrt<mode>2_sse): Ditto.
19788 2018-05-21 Tamar Christina <tamar.christina@arm.com>
19790 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
19792 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
19793 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
19794 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
19795 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
19797 * config/aarch64/arm_neon.h: Likewise.
19798 * config/aarch64/iterators.md (VQ_I): New.
19800 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
19802 * config.gcc: Add arc/t-multilib-linux to tmake_file for
19804 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
19807 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
19809 * config/nds32/constraints.md (S): New constraint.
19810 * config/nds32/nds32.md (call_internal): Use constraint S.
19811 (call_value_internal): Likewise.
19812 (sibcall_internal): Likewise.
19813 (sibcall_value_internal): Likewise.
19815 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
19816 Chung-Ju Wu <jasonwucj@gmail.com>
19818 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
19819 into consideration.
19821 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
19822 Chung-Ju Wu <jasonwucj@gmail.com>
19824 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
19825 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
19826 (nds32_rtx_costs_impl): Simplify.
19827 (nds32_address_cost_impl): Simplify.
19828 (nds32_init_rtx_costs): New function.
19829 (nds32_rtx_costs_speed_prefer): Likewise.
19830 (nds32_rtx_costs_size_prefer): Likewise.
19831 (nds32_address_cost_speed_prefer): Likewise.
19832 (nds32_address_cost_speed_fwprop): Likewise.
19833 (nds32_address_cost_size_prefer): Likewise.
19834 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
19835 * config/nds32/nds32.c (nds32_option_override): Use
19836 nds32_init_rtx_costs function.
19838 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
19840 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
19841 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
19842 (TARGET_PIPELINE_N8): Likewise.
19843 (TARGET_PIPELINE_N10): Likewise.
19844 (TARGET_PIPELINE_N13): Likewise.
19845 (TARGET_PIPELINE_GRAYWOLF): Likewise.
19847 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
19849 * config/nds32/nds32-fpu.md: Update copyright year.
19851 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
19853 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
19855 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
19857 * config/nds32/nds32.c
19858 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
19859 * config/nds32/nds32.opt (minline-asm-r15): New option.
19861 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
19863 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
19865 * config/nds32/nds32.md (abssi2): New pattern.
19867 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
19869 * config/i386/i386.md (rex64namesuffix): New mode attribute.
19870 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
19871 Merge insn pattern from sse_cvtsi2ss<round_name> and
19872 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
19873 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
19874 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
19875 using SWI48 mode iterator.
19876 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
19877 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
19878 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
19879 pattern from sse_cvttss2si<round_saeonly_name>
19880 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
19881 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
19882 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
19883 using SWI48 mode iterator.
19884 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
19885 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
19886 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
19887 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
19888 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
19889 using SWI48 mode iterator.
19890 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
19891 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
19892 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
19893 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
19894 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
19895 SWI48 mode iterator.
19896 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
19897 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
19898 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
19899 pattern from sse_cvttsd2si<round_saeonly_name>
19900 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
19902 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
19904 * config/nds32/nds32-md-auxiliary.c
19905 (nds32_valid_smw_lwm_base_p): Refine.
19906 (nds32_output_smw_single_word): Refine.
19907 (nds32_output_smw_double_word): New.
19908 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
19910 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
19912 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
19913 (nds32_output_stack_pop): Refine.
19914 (nds32_expand_unaligned_load): Refine.
19915 (nds32_expand_unaligned_store): Refine.
19917 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
19918 Chung-Ju Wu <jasonwucj@gmail.com>
19920 * config/nds32/constants.md: Add TP_REGNUM constant.
19921 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
19922 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
19924 * config/nds32/nds32-doubleword.md: Consider flag_pic.
19925 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
19926 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
19927 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
19928 and PIC code generation.
19929 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
19931 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
19933 * config/nds32/nds32.md: Support TLS and PIC.
19934 * config/nds32/nds32.c: Support TLS and PIC.
19935 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
19936 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
19939 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
19941 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
19942 mode with E_ prefix.
19944 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
19945 Chung-Ju Wu <jasonwucj@gmail.com>
19947 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
19948 * config/nds32/nds32-md-auxiliary.c
19949 (symbolic_reference_mentioned_p): New.
19950 (nds32_legitimize_ict_address): New.
19951 (nds32_expand_ict_move): New.
19952 (nds32_indirect_call_referenced_p): New.
19953 (nds32_symbol_binds_local_p): Delete.
19954 (nds32_long_call_p): Modify.
19955 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
19956 * config/nds32/nds32-protos.h
19957 (symbolic_reference_mentioned_p): Declare.
19958 (nds32_legitimize_ict_address): Declare.
19959 (nds32_expand_ict_move): Declare.
19960 (nds32_indirect_call_referenced_p): Declare.
19961 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
19962 (nds32_relax_group): Use nds32_ict_const_p as condition.
19963 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
19964 (nds32_asm_file_start): Output ict_model directive in asm code.
19965 (nds32_legitimate_address_p): Consider indirect call.
19966 (nds32_print_operand): Consider indirect call.
19967 (nds32_print_operand_address): Consider indirect call.
19968 (nds32_insert_attributes): Handle "indirect_call" attribute.
19969 (TARGET_LEGITIMATE_ADDRESS_P): Define.
19970 (TARGET_LEGITIMATE_CONSTANT_P): Define.
19971 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
19972 (TARGET_DELEGITIMIZE_ADDRESS): Define.
19973 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
19974 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
19975 (TARGET_ICT_MODEL_SMALL): Define.
19976 (TARGET_ICT_MODEL_LARGE): Define.
19977 * config/nds32/nds32.md (movsi): Consider ict model.
19978 (call, call_value): Consider ict model.
19979 (sibcall, sibcall_value): Consider ict model.
19980 * config/nds32/nds32.opt (mict-model): New option.
19981 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
19984 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
19985 Monk Chiang <sh.chiang04@gmail.com>
19986 Jim Wilson <jimw@sifive.com>
19988 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
19989 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
19990 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
19991 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
19992 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
19993 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
19994 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
19995 compute save_libcall_adjustment properly.
19996 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
19997 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
19998 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
19999 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
20000 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
20001 (ABI_SPEC): Handle mabi=ilp32e.
20002 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
20003 (RVE): Add RVE mask.
20004 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
20005 <-march>: Add rv32e as an example.
20007 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
20010 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
20011 (intra_create_variable_infos): Handle C++ constructors.
20013 2018-05-18 Martin Liska <mliska@suse.cz>
20015 * passes.def: Remove a redundant pass.
20017 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
20020 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
20022 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20024 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
20026 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
20028 (armv2, armv2a, armv3, armv3m): Delete architectures.
20029 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
20030 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
20031 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
20033 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
20034 (*mulsidi3adddi): Likewise.
20035 (mulsidi3): Likewise.
20036 (*mulsidi3_nov6): Likewise.
20037 (umulsidi3): Likewise.
20038 (umulsidi3_nov6): Likewise.
20039 (umaddsidi4): Likewise.
20040 (*umulsidi3adddi): Likewise.
20041 (smulsi3_highpart): Likewise.
20042 (*smulsi3_highpart_nov6): Likewise.
20043 (umulsi3_highpart): Likewise.
20044 (*umulsi3_highpart_nov6): Likewise.
20045 * config/arm/arm.h (arm_arch3m): Delete.
20046 * config/arm/arm.c (arm_arch3m): Delete.
20047 (arm_option_override_internal): Update armv3-related comment.
20048 (arm_configure_build_target): Delete use of isa_bit_mode32.
20049 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
20050 (arm_rtx_costs_internal): Delete check of arm_arch3m.
20051 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
20052 (mulsa3): Likewise.
20053 (mulusa3): Likewise.
20054 * config/arm/arm-protos.h (arm_arch3m): Delete.
20055 * config/arm/arm-tables.opt: Regenerate.
20056 * config/arm/arm-tune.md: Likewise.
20057 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
20058 deleted architectures.
20060 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20062 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
20063 (armv5t, armv5te): New features.
20064 (ARMv5, ARMv5e): Delete fgroups.
20065 (ARMv5t, ARMv5te): Adjust for above changes.
20066 (ARMv6m): Likewise.
20067 (armv5, armv5e): Delete arches.
20068 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
20070 (*call_reg_arm): Likewise.
20071 (*call_value_reg_armv5): Likewise.
20072 (*call_value_reg_arm): Likewise.
20073 (*call_symbol): Likewise.
20074 (*call_value_symbol): Likewise.
20075 (*sibcall_insn): Likewise.
20076 (*sibcall_value_insn): Likewise.
20077 (clzsi2): Likewise.
20078 (prefetch): Likewise.
20079 (define_split and define_peephole2 dependent on arm_arch5):
20081 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
20083 (TARGET_ARM_QBIT): Likewise.
20084 (TARGET_DSP_MULTIPLY): Likewise.
20085 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
20086 (arm_arch5, arm_arch5e): Delete.
20087 (arm_arch5t, arm_arch5te): Declare.
20088 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
20089 (arm_arch5t): Declare.
20090 (arm_option_reconfigure_globals): Update for the above.
20091 (arm_options_perform_arch_sanity_checks): Update comment, replace
20092 use of arm_arch5 with arm_arch5t.
20093 (use_return_insn): Likewise.
20094 (arm_emit_call_insn): Likewise.
20095 (output_return_instruction): Likewise.
20096 (arm_final_prescan_insn): Likewise.
20097 (arm_coproc_builtin_available): Likewise.
20098 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
20099 arm_arch5e with arm_arch5t and arm_arch5te.
20100 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
20101 (arm_arch5t, arm_arch5te): Declare.
20102 * config/arm/arm-tables.opt: Regenerate.
20103 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
20104 * config/arm/t-multilib: Likewise.
20105 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
20106 instead of arm_arch5.
20107 (*call_reg_thumb1): Likewise.
20108 (*call_value_reg_thumb1_v5): Likewise.
20109 (*call_value_reg_thumb1): Likewise.
20110 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
20112 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
20114 2018-05-18 Martin Liska <mliska@suse.cz>
20116 PR gcov-profile/84846
20117 * doc/gcov.texi: Document -t option of gcov tool.
20119 2018-05-18 Martin Liska <mliska@suse.cz>
20121 PR gcov-profile/84846
20122 * gcov.c (print_usage): Add new -t option.
20123 (process_args): Handle the option.
20124 (generate_results): Use stdout as output when requested by
20127 2018-05-18 Martin Liska <mliska@suse.cz>
20129 PR gcov-profile/84846
20130 * coverage.c (coverage_init): Write PWD to .gcno file.
20131 * doc/gcov.texi: Document how working directory is printed.
20132 * gcov-dump.c (dump_gcov_file): Print PWD.
20133 * gcov.c (output_intermediate_file): Likewise.
20134 (read_graph_file): Read PWD string.
20135 (output_lines): Print PWD.
20137 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20139 PR middle-end/85817
20140 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
20141 for retval and return false if all args to phi are zero.
20143 2018-05-18 Richard Biener <rguenther@suse.de>
20145 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
20147 (evrp_dom_walker::before_dom_children): Call it.
20149 2018-05-18 Richard Biener <rguenther@suse.de>
20151 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
20152 results when processing array refs with variable index.
20154 2018-05-18 Toon Moene <toon@moene.org>
20156 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
20157 directly after that of -floop-interchange. Indicate that both
20158 options are enabled by default when specifying -O3.
20160 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20162 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
20163 iterator. Delete separate integer-mode vec_set<mode> expander.
20164 (aarch64_simd_vec_setv2di): Delete.
20165 (vec_setv2di): Delete.
20166 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
20167 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
20168 the "w, r" alternative.
20170 2018-05-18 Martin Liska <mliska@suse.cz>
20172 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
20173 * tree-pass.h (make_pass_lower_switch_O0): New function.
20174 * tree-switch-conversion.c (node_has_low_bound): Remove.
20175 (node_has_high_bound): Likewise.
20176 (node_is_bounded): Likewise.
20177 (class pass_lower_switch): Make it a template type and create
20179 (pass_lower_switch::execute): Add template argument.
20180 (make_pass_lower_switch): New function.
20181 (make_pass_lower_switch_O0): New function.
20182 (do_jump_if_equal): Remove.
20183 (emit_case_nodes): Simplify to just handle all 3 cases and leave
20184 all the hard work to tree optimization passes.
20186 2018-05-18 Martin Liska <mliska@suse.cz>
20188 * dbgcnt.c (limit_low): Renamed from limit.
20189 (limit_high): New variable.
20190 (dbg_cnt_is_enabled): Check for upper limit.
20191 (dbg_cnt): Adjust dumping.
20192 (dbg_cnt_set_limit_by_index): Add new argument for high
20194 (dbg_cnt_set_limit_by_name): Likewise.
20195 (dbg_cnt_process_single_pair): Parse new format.
20196 (dbg_cnt_process_opt): Use strtok.
20197 (dbg_cnt_list_all_counters): Remove 'value' and add
20199 * doc/invoke.texi: Document changes.
20201 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
20203 * doc/sourcebuild.texi (scalar_all_fma): Document.
20204 * tree.def (FMA_EXPR): Delete.
20205 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
20206 * internal-fn.c (ternary_direct): New macro.
20207 (expand_ternary_optab_fn): Likewise.
20208 (direct_ternary_optab_supported_p): Likewise.
20209 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
20210 * builtins.c (fold_builtin_fma): Delete.
20211 (fold_builtin_3): Don't call it.
20212 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
20213 * expr.c (expand_expr_real_2): Likewise.
20214 * fold-const.c (operand_equal_p): Likewise.
20215 (fold_ternary_loc): Likewise.
20216 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
20217 * gimple.c (DEFTREECODE): Likewise.
20218 * gimplify.c (gimplify_expr): Likewise.
20219 * optabs-tree.c (optab_for_tree_code): Likewise.
20220 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
20221 * tree-eh.c (operation_could_trap_p): Likewise.
20222 (stmt_could_throw_1_p): Likewise.
20223 * tree-inline.c (estimate_operator_cost): Likewise.
20224 * tree-pretty-print.c (dump_generic_node): Likewise.
20225 (op_code_prio): Likewise.
20226 * tree-ssa-loop-im.c (stmt_cost): Likewise.
20227 * tree-ssa-operands.c (get_expr_operands): Likewise.
20228 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
20229 * fold-const-call.h (fold_fma): Delete.
20230 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
20231 CFN_FNMA and CFN_FNMS.
20232 (fold_fma): Delete.
20233 * genmatch.c (combined_fn): New enum.
20234 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
20235 (commutative_op): New function.
20236 (commutate): Use it. Handle more than 2 operands.
20237 (dt_operand::gen_gimple_expr): Use commutative_op.
20238 (parser::parse_expr): Allow :c to be used with non-binary
20239 operators if the commutative operand is known.
20240 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
20241 CFN_FMS, CFN_FNMA and CFN_FNMS.
20242 (backprop::process_assign_use): Remove FMA_EXPR handling.
20243 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
20244 (gen_hsa_fma): New function.
20245 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
20246 IFN_FNMA and IFN_FNMS.
20247 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
20248 * gimple-fold.h (follow_all_ssa_edges): Declare.
20249 * gimple-fold.c (follow_all_ssa_edges): New function.
20250 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
20251 gimple_build interface and use follow_all_ssa_edges to fold the result.
20252 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
20253 instead of checking for optabs directly.
20254 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
20255 rather than FMA_EXPRs.
20256 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
20257 call to IFN_FMA instead of an FMA_EXPR.
20259 2018-05-17 Jim Wilson <jimw@sifive.com>
20261 * expr.c (do_tablejump): When converting index to Pmode, if we have a
20262 sign extended promoted subreg, and the range does not have the sign bit
20263 set, then do a sign extend.
20265 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
20266 test, check for sign extended subreg and/or constant operands, and
20267 do a sign extend in that case.
20269 2018-05-17 Steve Ellcey <sellcey@cavium.com>
20271 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
20272 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
20274 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
20275 Change logics_shift_reg to logics_shift_imm.
20276 (thunderx2t99_fp_loadpair_basic): Delete.
20277 (thunderx2t99_fp_storepair_basic): Delete.
20278 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
20279 (thunderx2t99_asimd_polynomial): Delete.
20280 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
20281 and neon_fp_mul_d_scalar_q.
20282 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
20283 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
20284 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
20285 (thunderx2t99_asimd_lut): Add missing tbl types.
20286 (thunderx2t99_asimd_ext): Delete.
20287 (thunderx2t99_asimd_load1_1_mult): Delete.
20288 (thunderx2t99_asimd_load1_2_mult): Delete.
20289 (thunderx2t99_asimd_load1_ldp): New.
20290 (thunderx2t99_asimd_load1): New.
20291 (thunderx2t99_asimd_load2): Add missing *load2* types.
20292 (thunderx2t99_asimd_load3): New.
20293 (thunderx2t99_asimd_load4): New.
20294 (thunderx2t99_asimd_store1_1_mult): Delete.
20295 (thunderx2t99_asimd_store1_2_mult): Delete.
20296 (thunderx2t99_asimd_store2_mult): Delete.
20297 (thunderx2t99_asimd_store2_onelane): Delete.
20298 (thunderx2t99_asimd_store_stp): New.
20299 (thunderx2t99_asimd_store1): New.
20300 (thunderx2t99_asimd_store2): New.
20301 (thunderx2t99_asimd_store3): New.
20302 (thunderx2t99_asimd_store4): New.
20304 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
20306 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
20307 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
20309 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
20310 Segher Boessenkool <segher@kernel.crashing.org>
20313 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
20316 2018-05-17 Richard Biener <rguenther@suse.de>
20318 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
20319 for pruning loop and prune defs feeding only already visited PHIs.
20321 2018-05-17 Richard Biener <rguenther@suse.de>
20323 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
20325 2018-05-17 Bin Cheng <bin.cheng@arm.com>
20326 Richard Biener <rguenther@suse.de>
20328 PR tree-optimization/85793
20329 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
20330 for VMAT_ELEMENTWISE.
20332 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
20334 * internal-fn.h (lookup_internal_fn): Declare
20335 * internal-fn.c (lookup_internal_fn): New function.
20336 * gimple.c (gimple_build_call_from_tree): Handle calls to
20337 internal functions.
20338 * gimple-pretty-print.c (dump_gimple_call): Print "." before
20339 internal function names.
20340 * tree-pretty-print.c (dump_generic_node): Likewise.
20341 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
20343 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
20345 * gimple-fold.h (gimple_build): Make the function forms take
20346 combined_fn rather than built_in_function.
20347 (gimple_simplify): Likewise.
20348 * gimple-match-head.c (gimple_simplify): Likewise.
20349 * gimple-fold.c (gimple_build): Likewise.
20350 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
20351 rather than gimple_build_call_internal.
20352 (get_initial_defs_for_reduction): Likewise.
20353 (vect_create_epilog_for_reduction): Likewise.
20354 (vectorizable_live_operation): Likewise.
20356 2018-05-17 Martin Liska <mliska@suse.cz>
20358 * gimple-ssa-sprintf.c (format_directive): Do not use
20359 space in between 'G_' and '('.
20361 2018-05-17 Jakub Jelinek <jakub@redhat.com>
20364 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
20365 even if the mask is not all ones.
20368 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
20370 (ix86_gimple_fold_builtin): Likewise.
20373 * config/i386/i386.c: Include tree-vector-builder.h.
20374 (ix86_vector_shift_count): New function.
20375 (ix86_fold_builtin): Fold shift builtins by scalar count.
20376 (ix86_gimple_fold_builtin): Likewise.
20378 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
20379 _mm512_setzero): New intrinsics.
20381 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
20382 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20384 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
20385 code generation for cases where splatting a value is not useful.
20386 * simplify-rtx.c (simplify_ternary_operation): Simplify
20387 vec_merge across a vec_duplicate and a paradoxical subreg forming
20388 a vector mode to a vec_concat.
20390 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
20392 * config.gcc: Support "goldmont-plus".
20393 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
20395 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
20396 PROCESSOR_GOLDMONT_PLUS.
20397 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
20398 (processor_target_table): Add "goldmont-plus".
20399 (PTA_GOLDMONT_PLUS): Define.
20400 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
20401 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
20402 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
20403 (fold_builtin_cpu): Add "goldmont-plus".
20404 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
20405 (ix86_option_override_internal): Add "goldmont-plus".
20406 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
20407 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
20408 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
20409 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
20411 2018-05-17 Richard Biener <rguenther@suse.de>
20413 PR tree-optimization/85757
20414 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
20415 remove defs that only feed that PHI from further processing.
20417 2018-05-16 Jim Wilson <jimw@sifive.com>
20419 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
20421 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
20423 2018-05-16 Mark Wielaard <mark@klomp.org>
20425 * dwarf2out.c (count_index_strings): New function.
20426 (output_indirect_strings): Call count_index_strings and generate
20427 header for dwarf_version >= 5.
20429 2018-05-16 Mark Wielaard <mark@klomp.org>
20431 * dwarf2out.c (dwarf_FORM): New function.
20432 (set_indirect_string): Use dwarf_FORM.
20433 (reset_indirect_string): Likewise.
20434 (size_of_die): Likewise.
20435 (value_format): Likewise.
20436 (output_die): Likewise.
20437 (add_skeleton_AT_string): Likewise.
20438 (output_macinfo_op): Likewise.
20439 (index_string): Likewise.
20440 (output_index_string_offset): Likewise.
20441 (output_index_string): Likewise.
20442 (count_index_strings): Likewise.
20444 2018-05-16 Carl Love <cel@us.ibm.com>
20446 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
20447 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
20449 2018-05-16 Martin Jambor <mjambor@suse.cz>
20451 * ipa-prop.c (ipa_free_all_edge_args): Remove.
20452 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
20454 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
20456 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
20457 (fnma<mode>4): Likewise.
20458 (fms<mode>4): Likewise.
20459 (fnms<mode>4): Likewise.
20460 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
20461 (aarch64_fnma<mode>4): Likewise.
20462 (aarch64_fms<mode>4): Likewise.
20463 (aarch64_fnms<mode>4): Likewise.
20464 (aarch64_fnmadd<mode>4): Likewise.
20466 2018-05-16 Jason Merrill <jason@redhat.com>
20468 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
20470 2018-05-16 Richard Biener <rguenther@suse.de>
20472 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
20473 (dump_stmt_cost): Declare.
20474 (add_stmt_cost): Dump cost we add.
20475 (add_stmt_costs): New function.
20476 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
20477 No longer exported.
20478 (vect_analyze_stmt): Adjust prototype.
20479 (vectorizable_condition): Likewise.
20480 (vectorizable_live_operation): Likewise.
20481 (vectorizable_reduction): Likewise.
20482 (vectorizable_induction): Likewise.
20483 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
20484 cost vector to pass to vectorizable_ and record afterwards.
20485 (vect_model_reduction_cost): Take cost vector argument and adjust.
20486 (vect_model_induction_cost): Likewise.
20487 (vectorizable_reduction): Likewise.
20488 (vectorizable_induction): Likewise.
20489 (vectorizable_live_operation): Likewise.
20490 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
20491 SLP_TREE_NUMBER_OF_VEC_STMTS.
20492 (vect_analyze_slp_cost_1): Remove.
20493 (vect_analyze_slp_cost): Likewise.
20494 (vect_slp_analyze_node_operations): Take visited args and
20495 a target cost vector. Avoid processing already visited stmt sets.
20496 (vect_slp_analyze_operations): Use a local cost vector to gather
20497 costs and register those of non-discarded instances.
20498 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
20499 (vect_schedule_slp_instance): Remove copying of
20500 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
20502 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
20503 adding cost. Record cost entry location.
20504 (vect_prologue_cost_for_slp_op): Function to compute cost of
20505 a constant or invariant generated for SLP vect in the prologue,
20506 split out from vect_analyze_slp_cost_1.
20507 (vect_model_simple_cost): Make static. Adjust for SLP costing.
20508 (vect_model_promotion_demotion_cost): Likewise.
20509 (vect_model_store_cost): Likewise, make static.
20510 (vect_model_load_cost): Likewise.
20511 (vectorizable_bswap): Add cost vector arg and adjust.
20512 (vectorizable_call): Likewise.
20513 (vectorizable_simd_clone_call): Likewise.
20514 (vectorizable_conversion): Likewise.
20515 (vectorizable_assignment): Likewise.
20516 (vectorizable_shift): Likewise.
20517 (vectorizable_operation): Likewise.
20518 (vectorizable_store): Likewise.
20519 (vectorizable_load): Likewise.
20520 (vectorizable_condition): Likewise.
20521 (vectorizable_comparison): Likewise.
20522 (can_vectorize_live_stmts): Likewise.
20523 (vect_analyze_stmt): Likewise.
20524 (vect_transform_stmt): Adjust calls to vectorizable_*.
20525 * tree-vectorizer.c: Include gimple-pretty-print.h.
20526 (dump_stmt_cost): New function.
20528 2018-05-16 Richard Biener <rguenther@suse.de>
20530 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
20531 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
20532 * tree-ssa-dse.c: Include tree-ssa-loop.h.
20533 (check_name): New callback.
20534 (dse_classify_store): Track cycles via a visited bitmap of PHI
20535 defs and simplify handling of in-loop and across loop dead stores
20536 and properly fail for loop-variant refs. Handle byte-tracking with
20537 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
20540 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
20542 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
20543 (vect_get_mask_type_for_stmt): Likewise.
20544 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
20546 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
20547 to determine the statement's vector type and the vector type that
20548 should be used for calculating nunits. Deal with cases in which
20549 the type has to be deferred.
20550 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
20551 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
20552 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
20553 (vect_determine_vf_for_stmt): New functions, split out from...
20554 (vect_determine_vectorization_factor): ...here.
20555 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
20556 (vect_get_mask_type_for_stmt): New functions, split out from
20557 vect_determine_vectorization_factor.
20559 2018-05-16 Richard Biener <rguenther@suse.de>
20561 * tree-cfg.c (verify_gimple_assign_ternary): Properly
20562 verify the [VEC_]COND_EXPR embedded comparison.
20564 2018-05-15 Martin Sebor <msebor@redhat.com>
20566 PR tree-optimization/85753
20567 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
20568 RECORD_TYPE in addition to ARRAY_TYPE.
20570 2018-05-15 Martin Sebor <msebor@redhat.com>
20572 PR middle-end/85643
20573 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
20575 2018-05-15 Richard Biener <rguenther@suse.de>
20577 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
20578 add by_clobber_p one. Change algorithm to collect all defs
20579 representing uses we need to walk and try reducing them to
20580 a single one before failing.
20581 (dse_dom_walker::dse_optimize_stmt): Adjust.
20583 2018-05-13 Mark Wielaard <mark@klomp.org>
20585 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
20586 (size_of_loc_descr): Likewise.
20587 (output_loc_operands): Likewise.
20588 (output_loc_operands_raw): Likewise.
20589 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
20590 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
20591 (hash_loc_operands): Likewise.
20592 (compare_loc_operands): Likewise.
20594 2018-05-14 Mark Wielaard <mark@klomp.org>
20596 * dwarf2out.c (count_index_addrs): New function.
20597 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
20599 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20601 PR tree-optimization/83648
20602 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
20603 return value as malloc candidate.
20605 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20608 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
20609 param as true in call to suggest_attribute.
20611 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
20613 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
20614 -mreadonly-in-sdata.
20616 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20618 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
20620 (aarch64_crypto_aesd_fused): Likewise.
20622 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
20624 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
20625 (movsi_aarch64): Likewise.
20626 (load_pairsi): Likewise.
20627 (load_pairdi): Likewise.
20628 (store_pairsi): Likewise.
20629 (store_pairdi): Likewise.
20630 (load_pairsf): Likewise.
20631 (load_pairdf): Likewise.
20632 (store_pairsf): Likewise.
20633 (store_pairdf): Likewise.
20634 (zero_extend): Likewise.
20635 (trunc): Swap alternatives.
20636 (fcvt_target): Add '?' to prefer w over r.
20638 2018-05-14 Jakub Jelinek <jakub@redhat.com>
20641 * config/i386/i386.md: Disallow non-commutative arithmetics in
20642 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
20643 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
20644 in the peephole2 before it.
20646 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
20648 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
20649 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
20650 (ix86_handle_option): Handle -mcldemote.
20651 * config.gcc: New header.
20652 * config/i386/cldemoteintrin.h: New file.
20653 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
20654 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
20656 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
20657 OPTION_MASK_ISA_CLDEMOTE.
20658 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
20659 (ix86_valid_target_attribute_inner_p): Ditto.
20660 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
20661 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
20662 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
20663 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
20664 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
20666 * config/i386/i386.opt: Add -mcldemote.
20667 * config/i386/x86intrin.h: New header.
20668 * doc/invoke.texi: Add -mcldemote.
20670 2018-05-14 Richard Biener <rguenther@suse.de>
20672 * doc/match-and-simplify.texi: Adjust :s documentation.
20674 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
20676 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
20677 intended memcpy size.
20678 (REORDER_45): Likewise.
20680 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
20682 * sort.cc: New file.
20683 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
20684 * vec.c (qsort_chk): Use gcc_qsort.
20685 * Makefile.in (OBJS-libcommon): Add sort.o.
20686 (build/sort.o): New target. Use it...
20687 (BUILD_RTL): ... here, and...
20688 (build/gencfn-macros): ... here, and...
20689 (build/genmatch): ... here.
20691 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
20692 Chung-Ju Wu <jasonwucj@gmail.com>
20694 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
20695 * config/nds32/nds32-graywolf.md: New file.
20696 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
20697 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
20699 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
20700 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
20701 * config/nds32/nds32.md (pipeline_model): Add graywolf.
20702 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
20703 * config/nds32/pipelines.md: Include n15 settings.
20705 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
20706 Chung-Ju Wu <jasonwucj@gmail.com>
20708 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
20709 * config/nds32/nds32-n13.md: New file.
20710 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
20711 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
20713 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
20714 * config/nds32/nds32.md (pipeline_model): Add n13.
20715 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
20716 * config/nds32/pipelines.md: Include n13 settings.
20718 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
20719 Chung-Ju Wu <jasonwucj@gmail.com>
20721 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
20722 * config/nds32/nds32-n10.md: New file.
20723 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
20724 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
20726 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
20727 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
20728 * config/nds32/nds32.md (pipeline_model): Add n10.
20729 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
20730 * config/nds32/pipelines.md: Include n10 settings.
20732 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
20733 Kito Cheng <kito.cheng@gmail.com>
20734 Chung-Ju Wu <jasonwucj@gmail.com>
20736 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
20737 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
20738 Add enum values for DSP extension instructions.
20739 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
20741 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
20742 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
20743 New code iterators.
20744 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
20745 * config/nds32/nds32-dspext.md: New file for DSP implementation.
20746 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
20747 * config/nds32/nds32-intrinsic.md: Likewise.
20748 * config/nds32/nds32_intrinsic.h: Likewise.
20749 * config/nds32/nds32-md-auxiliary.c: Likewise.
20750 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
20751 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
20752 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
20753 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
20754 * config/nds32/nds32-protos.h: New declarations for DSP extension.
20755 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
20756 TYPE_DMAC in switch statement.
20757 * config/nds32/nds32.c: New checking and implementation for DSP
20758 extension instructions.
20759 * config/nds32/nds32.h: Likewise.
20760 * config/nds32/nds32.md: Likewise.
20761 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
20762 * config/nds32/predicates.md: Implement new predicates for DSP
20765 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
20767 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
20768 Reformat alternatives and attributes so it is easier to identify
20769 which constraints/attributes go with which instruction.
20770 (mov<mode>_hardfloat32, FMOVE64): Likewise.
20771 (mov<mode>_softfloat32, FMOVE64): Likewise.
20772 (mov<mode>_hardfloat64, FMOVE64): Likewise.
20773 (mov<mode>_softfloat64, FMOVE64): Likewise.
20775 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
20777 * doc/extend.texi (PowerPC Built-in Functions): Rename this
20779 (Basic PowerPC Built-in Functions): The new name of the
20780 subsection previously known as "PowerPC Built-in Functions".
20781 (Basic PowerPC Built-in Functions Available on all Configurations):
20783 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
20784 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
20785 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
20786 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
20788 2018-05-11 Martin Jambor <mjambor@suse.cz>
20791 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
20794 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
20797 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
20799 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
20801 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
20802 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
20803 (ix86_handle_option): Handle -mwaitpkg.
20804 * config.gcc: New header.
20805 * config/i386/cpuid.h (bit_WAITPKG): New bit.
20806 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
20807 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
20809 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
20810 OPTION_MASK_ISA_WAITPKG.
20811 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
20812 (ix86_option_override_internal): Add PTA_WAITPKG.
20813 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
20814 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
20815 IX86_BUILTIN_TPAUSE.
20816 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
20817 __builtin_ia32_umwait and __builtin_ia32_tpause.
20818 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
20819 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
20820 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
20821 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
20822 UNSPECV_TPAUSE): New.
20823 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
20824 * config/i386/i386.opt: Add -mwaitpkg.
20825 * config/i386/waitpkgintrin.h: New file.
20826 * config/i386/x86intrin.h: New header.
20827 * doc/invoke.texi: Add -mwaitpkg.
20829 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
20832 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
20834 (cortex-m0): Use armv6s-m isa.
20835 (cortex-m0plus): Likewise.
20836 (cortex-m1): Likewise.
20837 (cortex-m0.small-multiply): Likewise.
20838 (cortex-m0plus.small-multiply): Likewise.
20839 (cortex-m1.small-multiply): Likewise.
20841 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
20842 Jakub Jelinek <jakub@redhat.com>
20844 PR tree-optimization/85692
20845 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
20846 source permute as well.
20848 2018-05-11 Martin Liska <mliska@suse.cz>
20851 * doc/extend.texi: Document LLVM style format for no_sanitize
20854 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
20856 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
20857 mode_supports_vsx_dform_quad to mode_supports_dq_form.
20858 (mode_supports_vsx_dform_quad): Likewise.
20859 (mode_supports_vmx_dform): Move these functions to be next to the
20860 other mode_supports functions.
20861 (mode_supports_dq_form): Likewise.
20862 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
20863 mode_supports_dq_form.
20864 (reg_offset_addressing_ok_p): Likewise.
20865 (offsettable_ok_by_alignment): Likewise.
20866 (rs6000_legitimate_offset_address_p): Likewise.
20867 (legitimate_lo_sum_address_p): Likewise.
20868 (rs6000_legitimize_address): Likewise.
20869 (rs6000_legitimize_reload_address): Likewise.
20870 (rs6000_secondary_reload_inner): Likewise.
20871 (rs6000_preferred_reload_class): Likewise.
20872 (rs6000_output_move_128bit): Likewise.
20874 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
20876 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
20877 Generate SImode target register for null target.
20878 <case IX86_BUILTIN_XGETBV>: Ditto.
20879 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
20880 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
20882 2018-05-10 Carl Love <cel@us.ibm.com>
20884 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
20885 dcbtt and dcbtstt if operands[2] is 0.
20887 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
20890 * config/i386/sse.md (usadv64qi): New expander.
20892 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
20894 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
20895 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
20896 -maltivec=be support.
20897 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
20898 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
20899 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20900 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
20901 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
20902 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
20903 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
20904 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
20905 altivec_vsumsws): Adjust.
20906 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
20907 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
20908 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
20910 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
20911 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
20912 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
20913 (altivec_lve<VI_char>x): Delete expand.
20914 (*altivec_lve<VI_char>x_internal): Rename to...
20915 (altivec_lve<VI_char>x): ... this.
20916 (altivec_lvxl_<mode>): Delete expand.
20917 (*altivec_lvxl_<mode>_internal): Rename to ...
20918 (altivec_lvxl_<mode>): ... this.
20919 (altivec_stvxl_<mode>): Delete expand.
20920 (*altivec_stvxl_<mode>_internal): Rename to ...
20921 (altivec_stvxl_<mode>): ... this.
20922 (altivec_stve<VI_char>x): Delete expand.
20923 (*altivec_stve<VI_char>x_internal): Rename to ...
20924 (altivec_stve<VI_char>x): ... this.
20925 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
20926 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
20927 reduc_plus_scal_<mode>): Adjust.
20928 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
20930 (rs6000_cpu_cpp_builtins): Adjust.
20931 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
20932 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
20933 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
20934 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
20935 -maltivec=be support.
20936 (rs6000_split_vec_extract_var): Adjust.
20937 (rs6000_split_v4si_init): Adjust.
20938 (swap_selector_for_mode): Delete.
20939 (altivec_expand_lvx_be, altivec_expand_stvx_be,
20940 altivec_expand_stvex_be): Delete.
20941 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
20942 -maltivec=be support.
20943 (rs6000_gimple_fold_builtin): Ditto.
20944 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
20946 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
20947 (TARGET_DIRECT_MOVE_64BIT): Adjust.
20948 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
20949 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
20950 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
20951 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
20952 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
20953 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
20954 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
20955 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
20956 anonymous split): Adjust.
20957 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
20958 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
20960 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
20962 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
20963 when --with-gxx-include-dir is also specified.
20964 * configure: Regenerate.
20966 2018-05-09 Jim Wilson <jimw@sifive.com>
20969 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
20970 * config/riscv/t-withmultilib: New.
20971 * config/riscv/withmultilib.h: New.
20972 * doc/install.texi: Document RISC-V --with-multilib-list support.
20974 2018-05-09 Richard Biener <rguenther@suse.de>
20976 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
20978 (vect_bb_vectorization_profitable_p): Adjust. Compute
20979 actual scalar cost using the cost vector and the add_stmt_cost
20982 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
20984 PR rtl-optimization/85645
20985 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
20986 in the REG_CFA_REGISTER note for LR, don't leave it empty.
20988 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
20990 PR rtl-optimization/85645
20991 * shrink-wrap.c (spread_components): Return a boolean saying if
20992 anything was changed.
20993 (try_shrink_wrapping_separate): Iterate spread_components until
20994 nothing changes anymore.
20996 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
20998 PR rtl-optimization/85645
20999 * regrename.c (build_def_use): Also kill the chains that include the
21000 destination of a REG_CFA_REGISTER note.
21002 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
21004 PR rtl-optimization/85645
21005 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
21006 insn that has a REG_CFA_REGISTER note.
21008 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
21010 * cfgexpand.c (expand_clobber): New function.
21011 (expand_gimple_stmt_1): Use it.
21012 * tree-vect-stmts.c (vect_clobber_variable): New function,
21014 (vectorizable_simd_clone_call): ...here.
21015 (vectorizable_store): Emit a clobber either side of an
21016 IFN_STORE_LANES sequence.
21017 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
21019 2018-05-09 Tom de Vries <tom@codesourcery.com>
21022 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
21023 (define_insn "trap_if_false"): Add exit after trap.
21025 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
21027 PR rtl-optimization/85638
21028 * bb-reorder.c: Include common/common-target.h.
21029 (create_forwarder_block): New function extracted from...
21030 (fix_up_crossing_landing_pad): ...here. Rename into...
21031 (dw2_fix_up_crossing_landing_pad): ...this.
21032 (sjlj_fix_up_crossing_landing_pad): New function.
21033 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
21034 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
21035 from both partitions and exit the loop after one iteration.
21037 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
21040 * doc/extend.texi (PowerPC Built-in Functions): Rename this
21042 (Basic PowerPC Built-in Functions): The new name of the
21043 subsection previously known as "PowerPC Built-in Functions".
21044 (Basic PowerPC Built-in Functions Available on all Configurations):
21046 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
21048 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
21050 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
21052 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
21055 2018-05-08 Jim Wilson <jimw@sifive.com>
21057 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
21058 (LD_EMUL_SUFFIX): New.
21059 (LINK_SPEC): Use it.
21061 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
21063 * doc/extend.texi (PowerPC Built-in Functions): Rename this
21065 (Basic PowerPC Built-in Functions): The new name of the
21066 subsection previously known as "PowerPC Built-in Functions".
21067 (Basic PowerPC Built-in Functions Available on all Configurations):
21069 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
21071 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
21073 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
21075 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
21078 2018-05-08 Jakub Jelinek <jakub@redhat.com>
21081 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
21082 after cmpelim optimization.
21084 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
21086 * config.gcc: Support "goldmont".
21087 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
21088 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
21089 PROCESSOR_GOLDMONT.
21090 * config/i386/i386.c (m_GOLDMONT): Define.
21091 (processor_target_table): Add "goldmont".
21092 (PTA_GOLDMONT): Define.
21093 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
21094 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
21095 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
21096 (fold_builtin_cpu): Add "goldmont".
21097 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
21098 (ix86_option_override_internal): Add "goldmont".
21099 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
21100 (processor_type): Add PROCESSOR_GOLDMONT.
21101 * config/i386/i386.md: Add CPU "glm".
21102 * config/i386/glm.md: New file.
21103 * config/i386/x86-tune.def: Add m_GOLDMONT.
21104 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
21106 2018-05-08 Jakub Jelinek <jakub@redhat.com>
21109 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
21111 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
21112 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
21113 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
21116 * config/i386/i386.c (ix86_fold_builtin): Handle
21117 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
21120 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
21121 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
21123 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
21126 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
21127 (check_arch): Likewise.
21128 (check_fpu): Return the result rather than printing it.
21129 (end arch): Fix operator precedence.
21130 (end cpu): Likewise.
21131 (END): Print the result from check_fpu.
21133 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
21134 Alan Hayward <alan.hayward@arm.com>
21135 David Sherwood <david.sherwood@arm.com>
21137 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
21138 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
21139 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
21140 (*fcmuo<mode>_and): New patterns.
21142 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
21144 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
21145 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
21146 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
21147 (cmp_op, sve_imm_con): New code attributes.
21148 (SVE_COND_INT_CMP, imm_con): Delete.
21149 (cmp_op): Remove above unspecs from int attribute.
21150 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
21152 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
21153 comparison-specific unspecs.
21154 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
21155 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
21156 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
21157 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
21158 (*vec_fcm<cmp_op><mode>): Rename to...
21159 (*fcm<cmp_op><mode>): ...this and adjust likewise.
21160 (*vec_fcmuo<mode>): Rename to...
21161 (*fcmuo<mode>): ...this and adjust likewise.
21162 (*pred_fcm<cmp_op><mode>): New pattern.
21163 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
21164 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
21166 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
21168 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
21169 (aarch64_emit_sve_predicated_cond): New function.
21170 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
21171 (aarch64_emit_unspec_cond_or): Replace with...
21172 (aarch64_emit_sve_or_conds): ...this new function. Use
21173 aarch64_emit_sve_ptrue_op for the individual comparisons and
21174 aarch64_emit_binop to OR them together.
21175 (aarch64_emit_inverted_unspec_cond): Replace with...
21176 (aarch64_emit_sve_inverted_cond): ...this new function. Use
21177 aarch64_emit_sve_ptrue_op for the comparison and
21178 aarch64_emit_unop to invert the result.
21179 (aarch64_expand_sve_vec_cmp_float): Update after the above
21180 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
21182 2018-05-07 Nathan Sidwell <nathan@acm.org>
21184 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
21185 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
21186 (Backwards Compatibility): Likewise.
21188 2018-05-07 Luis Machado <luis.machado@linaro.org>
21192 2018-05-07 Luis Machado <luis.machado@linaro.org>
21194 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
21195 <prefetch_dynamic_strides>: New const bool field.
21196 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
21197 prefetch_dynamic_strides.
21198 (exynosm1_prefetch_tune): Likewise.
21199 (thunderxt88_prefetch_tune): Likewise.
21200 (thunderx_prefetch_tune): Likewise.
21201 (thunderx2t99_prefetch_tune): Likewise.
21202 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
21204 (aarch64_override_options_internal): Update to set
21205 PARAM_PREFETCH_DYNAMIC_STRIDES.
21206 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
21207 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
21208 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
21209 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
21210 prefetch-dynamic-strides setting.
21212 2018-05-07 Luis Machado <luis.machado@linaro.org>
21214 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
21215 <minimum_stride>: New const int field.
21216 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
21217 minimum_stride field.
21218 (exynosm1_prefetch_tune): Likewise.
21219 (thunderxt88_prefetch_tune): Likewise.
21220 (thunderx_prefetch_tune): Likewise.
21221 (thunderx2t99_prefetch_tune): Likewise.
21222 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
21223 (aarch64_override_options_internal): Update to set
21224 PARAM_PREFETCH_MINIMUM_STRIDE.
21225 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
21226 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
21227 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
21228 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
21229 stride is constant and is below the minimum stride threshold.
21231 2018-05-07 Luis Machado <luis.machado@linaro.org>
21233 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
21236 2018-05-07 Luis Machado <luis.machado@linaro.org>
21238 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
21239 <prefetch_dynamic_strides>: New const bool field.
21240 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
21241 prefetch_dynamic_strides.
21242 (exynosm1_prefetch_tune): Likewise.
21243 (thunderxt88_prefetch_tune): Likewise.
21244 (thunderx_prefetch_tune): Likewise.
21245 (thunderx2t99_prefetch_tune): Likewise.
21246 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
21248 (aarch64_override_options_internal): Update to set
21249 PARAM_PREFETCH_DYNAMIC_STRIDES.
21250 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
21251 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
21252 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
21253 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
21254 prefetch-dynamic-strides setting.
21256 2018-05-07 Luis Machado <luis.machado@linaro.org>
21258 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
21259 <minimum_stride>: New const int field.
21260 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
21261 minimum_stride field.
21262 (exynosm1_prefetch_tune): Likewise.
21263 (thunderxt88_prefetch_tune): Likewise.
21264 (thunderx_prefetch_tune): Likewise.
21265 (thunderx2t99_prefetch_tune): Likewise.
21266 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
21267 (aarch64_override_options_internal): Update to set
21268 PARAM_PREFETCH_MINIMUM_STRIDE.
21269 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
21270 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
21271 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
21272 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
21273 stride is constant and is below the minimum stride threshold.
21275 2018-05-06 Jakub Jelinek <jakub@redhat.com>
21278 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
21279 the type is addressable. Don't force op into register if it has
21282 2018-05-05 Roland McGrath <mcgrathr@google.com>
21285 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
21286 any section for which we don't know a specific type it should have,
21287 regardless of name. Previously this was done only for the exact
21288 names ".init_array", ".fini_array", and ".preinit_array".
21289 (default_elf_asm_named_section): Add comment about
21290 relationship with default_section_type_flags and SECTION_NOTYPE.
21291 (get_section): Don't consider it a type conflict if one side has
21292 SECTION_NOTYPE and the other doesn't, as long as neither has the
21293 SECTION_BSS et al used in the default_section_type_flags logic.
21295 2018-05-05 Tom de Vries <tom@codesourcery.com>
21298 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
21299 (workaround_barsyncs): New function.
21300 (nvptx_reorg): Use workaround_barsyncs.
21301 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
21302 (define_expand "nvptx_membar_cta"): New define_expand.
21303 (define_insn "*nvptx_membar_cta"): New insn.
21305 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21307 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
21308 To improve optimization opportunities.
21309 * builtin-types.def: The new needed builtin types for the above.
21311 2018-05-04 Richard Biener <rguenther@suse.de>
21313 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
21314 * gimple-ssa-store-merging.c
21315 (imm_store_chain_info::output_merged_store): Remove redundant create,
21316 release split_store vector contents on failure.
21317 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
21318 scalar stmt vector on cache hit.
21320 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
21322 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
21324 * config.gcc (powerpc-xilinx-eabi*): Remove.
21325 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
21327 (fusion_addis_mem_combo_load): Ditto.
21328 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
21330 (rs6000_cpu_cpp_builtins): Ditto.
21331 * config/rs6000/rs6000-linux.c
21332 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
21333 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
21334 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
21336 (rs6000_setup_reg_addr_masks): Ditto.
21337 (rs6000_init_hard_regno_mode_ok): Ditto.
21338 (rs6000_option_override_internal): Ditto.
21339 (legitimate_lo_sum_address_p): Ditto.
21340 (rs6000_legitimize_address): Ditto.
21341 (rs6000_legitimize_reload_address): Ditto.
21342 (rs6000_legitimate_address_p): Ditto.
21343 (abi_v4_pass_in_fpr): Ditto.
21344 (setup_incoming_varargs): Ditto.
21345 (rs6000_gimplify_va_arg): Ditto.
21346 (rs6000_split_multireg_move): Ditto.
21347 (rs6000_savres_strategy): Ditto.
21348 (rs6000_emit_prologue_components): Ditto.
21349 (rs6000_emit_epilogue_components): Ditto.
21350 (rs6000_emit_prologue): Ditto.
21351 (rs6000_emit_epilogue): Ditto.
21352 (rs6000_elf_file_end): Ditto.
21353 (rs6000_function_value): Ditto.
21354 (rs6000_libcall_value): Ditto.
21355 * config/rs6000/rs6000.h: Ditto.
21356 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
21357 (TARGET_MINMAX): ... this. New.
21358 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
21359 * config/rs6000/rs6000.md: Remove Xilinx FP support.
21360 (*movsi_internal1_single): Delete.
21361 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
21362 mfpu=, mxilinx-fpu): Delete.
21363 * config/rs6000/singlefp.h: Delete.
21364 * config/rs6000/sysv4.h: Remove Xilinx FP support.
21365 * config/rs6000/t-rs6000: Ditto.
21366 * config/rs6000/t-xilinx: Delete.
21367 * config/rs6000/titan.md: Adjust for fp_type removal.
21368 * config/rs6000/vsx.md: Remove Xilinx FP support.
21369 (VStype_simple): Delete.
21370 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
21371 * config/rs6000/xfpu.h: Delete.
21372 * config/rs6000/xfpu.md: Delete.
21373 * config/rs6000/xilinx.h: Delete.
21374 * config/rs6000/xilinx.opt: Delete.
21375 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
21376 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
21378 2018-05-04 Tom de Vries <tom@codesourcery.com>
21381 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
21384 2018-05-04 Richard Biener <rguenther@suse.de>
21386 PR middle-end/85627
21387 * tree-complex.c (update_complex_assignment): We are always in SSA form.
21388 (expand_complex_div_wide): Likewise.
21389 (expand_complex_operations_1): Likewise.
21390 (expand_complex_libcall): Preserve EH info of the original stmt.
21391 (tree_lower_complex): Handle removed blocks.
21392 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
21393 on complex multiplication and division libcall builtins.
21395 2018-05-04 Richard Biener <rguenther@suse.de>
21397 PR middle-end/85574
21398 * fold-const.c (negate_expr_p): Restrict negation of operand
21399 zero of a division to when we know that can happen without
21401 (fold_negate_expr_1): Likewise.
21403 2018-05-04 Jakub Jelinek <jakub@redhat.com>
21406 * real.h (real_nextafter): Declare.
21407 * real.c (real_nextafter): New function.
21408 * fold-const-call.c (fold_const_nextafter): New function.
21409 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
21410 CASE_CFN_NEXTTOWARD.
21411 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
21412 even when arg1_mode is different from arg0_mode.
21414 2018-05-03 Nathan Sidwell <nathan@acm.org>
21416 * doc/extend.texi (Deprecated Features): Remove
21417 -ffriend-injection.
21418 (Backwards Compatibility): Likewise.
21419 * doc/invoke.texi (C++ Language Options): Likewise.
21420 (C++ Dialect Options): Likewise.
21422 2018-05-03 Jakub Jelinek <jakub@redhat.com>
21425 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
21426 _mm512_mask_mullox_epi64): New intrinsics.
21428 2018-05-03 Tom de Vries <tom@codesourcery.com>
21431 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
21432 dump files): Add offload-tree.
21434 2018-05-03 Richard Biener <rguenther@suse.de>
21436 PR tree-optimization/85615
21437 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
21438 to loops not nested in BBs loop father to avoid creating multi-entry
21441 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21443 PR tree-optimization/70291
21444 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
21445 arguments. Change return type to tree. Emit libcall as a new
21446 statement rather than replacing existing one when inplace_p is true.
21447 (expand_complex_multiplication_components): New function.
21448 (expand_complex_multiplication): Expand floating-point complex
21449 multiplication using the above.
21450 (expand_complex_division): Rename inner_type parameter to type.
21451 Update expand_complex_libcall call-site.
21452 (expand_complex_operations_1): Update expand_complex_multiplication
21453 and expand_complex_division call-sites.
21455 2018-05-02 Jakub Jelinek <jakub@redhat.com>
21458 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
21459 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
21460 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
21461 the highest significant bit of the shift count mask is clear. In
21462 check whether and[sq]i3 is needed verify that all significant bits
21463 of the shift count other than the highest are set.
21465 2018-05-02 Tom de Vries <tom@codesourcery.com>
21468 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
21469 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
21470 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
21471 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
21472 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
21473 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
21474 __builtin_goacc_parlevel_size.
21476 2018-05-02 Richard Biener <rguenther@suse.de>
21478 PR tree-optimization/85597
21479 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
21480 do not use split vect_get_vec_defs call but call vect_get_slp_defs
21483 2018-05-02 Tom de Vries <tom@codesourcery.com>
21486 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
21487 dump files): Add ltrans-tree.
21489 2018-05-02 Tom de Vries <tom@codesourcery.com>
21492 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
21493 dump files): Add wpa-ipa.
21495 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
21497 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
21498 powerpc*-*-linux*paired* target.
21499 * config/rs6000/750cl.h: Delete.
21500 * config/rs6000/paired.h: Delete.
21501 * config/rs6000/paired.md: Delete.
21502 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
21504 * config/rs6000/rs6000-builtin.def: Remove paired float support.
21505 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
21506 comment. Remove paired float support.
21507 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
21508 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
21510 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
21511 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
21513 * config/rs6000/rs6000.c: Remove paired float support.
21514 (paired_expand_vector_init, paired_expand_vector_move,
21515 paired_emit_vector_compare, paired_emit_vector_cond_expr,
21516 (paired_expand_lv_builtin, paired_expand_stv_builtin,
21517 paired_expand_builtin, paired_expand_predicate_builtin,
21518 paired_init_builtins): Delete.
21519 * config/rs6000/rs6000.h: Remove paired float support.
21520 * config/rs6000/rs6000.md: Remove paired float support.
21521 (move_from_CR_ov_bit): Delete.
21522 * config/rs6000/rs6000.opt (mpaired): Delete.
21523 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
21524 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
21526 2018-05-02 Richard Biener <rguenther@suse.de>
21528 PR middle-end/85567
21529 * gimplify.c (gimplify_save_expr): When in SSA form allow
21530 SAVE_EXPRs to compute to SSA vars.
21532 2018-05-02 Jakub Jelinek <jakub@redhat.com>
21535 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
21536 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
21537 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
21538 clobber operands[2], instead use a new pseudo. Formatting fixes.
21540 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
21542 PR tree-optimization/85586
21543 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
21544 exit early for statements in the same group if the accesses are
21547 2018-05-02 Tom de Vries <tom@codesourcery.com>
21550 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
21553 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
21555 PR tree-optimization/85143
21556 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
21558 2018-05-01 Tom de Vries <tom@codesourcery.com>
21561 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
21562 not found" error message.
21564 2018-05-01 Tom de Vries <tom@codesourcery.com>
21567 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
21568 * vec.c (test_ordered_remove_if): New function.
21569 (vec_c_tests): Call test_ordered_remove_if.
21570 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
21571 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
21572 * tree-vect-patterns.c (vect_pattern_recog_1): Use
21573 VEC_ORDERED_REMOVE_IF.
21575 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21577 PR tree-optimization/82665
21578 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
21579 pointer subtraction where arguments come from a memchr call.
21581 2018-05-01 Jakub Jelinek <jakub@redhat.com>
21583 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
21584 --push-state --as-needed and --pop-state instead of --as-needed and
21585 --no-as-needed if ld supports it.
21586 * configure: Regenerated.
21589 * doc/install.texi2html: Replace _002d with - and _002a with * in
21590 generated html files using sed.
21592 2018-04-30 David Malcolm <dmalcolm@redhat.com>
21595 * gcc-rich-location.c (blank_line_before_p): New function.
21596 (use_new_line): New function.
21597 (gcc_rich_location::add_fixit_insert_formatted): New function.
21598 * gcc-rich-location.h
21599 (gcc_rich_location::add_fixit_insert_formatted): New function.
21601 2018-04-30 David Malcolm <dmalcolm@redhat.com>
21603 * selftest.c (assert_streq): Rename "expected" and "actual" to
21604 "val1" and "val2". Extend NULL-handling to cover both inputs
21605 symmetrically, while still requiring both to be non-NULL for a pass.
21606 * selftest.h (assert_streq): Rename "expected" and "actual" to
21608 (ASSERT_EQ): Likewise.
21609 (ASSERT_EQ_AT): Likewise.
21610 (ASSERT_KNOWN_EQ): Likewise.
21611 (ASSERT_KNOWN_EQ_AT): Likewise.
21612 (ASSERT_NE): Likewise.
21613 (ASSERT_MAYBE_NE): Likewise.
21614 (ASSERT_MAYBE_NE_AT): Likewise.
21615 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
21616 the assertion to pass.
21617 (ASSERT_STREQ_AT): Likewise.
21619 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
21621 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
21622 interaction with -pie.
21624 2018-04-30 David Malcolm <dmalcolm@redhat.com>
21626 * selftest.h: Fix alphabetization of per-source-file selftest
21629 2018-04-30 Jason Merrill <jason@redhat.com>
21631 PR c++/61982 - dead stores to destroyed objects.
21632 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
21635 2018-04-30 Jason Merrill <jason@redhat.com>
21637 * tree.c (build_clobber): New.
21638 * tree.h: Declare it.
21639 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
21641 2018-04-30 David Malcolm <dmalcolm@redhat.com>
21643 * diagnostic-show-locus.c (layout::layout): Update for
21644 location_get_source_line returning a char_span.
21645 (struct char_span): Move to input.h.
21646 (struct correction): Update for fields in char_span becoming
21648 (struct source_line): Update for location_get_source_line
21649 returning a char_span.
21650 (layout::print_line): Likewise.
21651 * edit-context.c (edited_file::print_content): Likewise.
21652 (edited_file::print_diff_hunk): Likewise.
21653 (edited_file::print_run_of_changed_lines): Likewise.
21654 (edited_file::get_num_lines): Likewise.
21655 (edited_line::edited_line): Likewise.
21656 * final.c (asm_show_source): Likewise.
21657 * input.c (location_get_source_line): Convert return type
21658 from const char * to char_span, losing the final "line_len"
21660 (dump_location_info): Update for the above.
21661 (get_substring_ranges_for_loc): Likewise. Use a char_span
21662 when handling the literal within the line.
21663 (test_reading_source_line): Update for location_get_source_line
21664 returning a char_span.
21665 * input.h (class char_span): Move here from
21666 diagnostic-show-locus.c, converting from a struct to a class.
21667 Make data members private.
21668 (char_span::operator bool): New.
21669 (char_span::length): New.
21670 (char_span::get_buffer): New.
21671 (char_span::operator[]): New.
21672 (char_span::subspan): Make const.
21673 (char_span::xstrdup): New.
21674 (location_get_source_line): Convert return type from const char *
21675 to char_span, losing the final "line_size" param.
21677 2018-04-30 Jan Hubicka <jh@suse.cz>
21679 * lto-wrapper.c (ltrans_priorities): New static var.
21680 (cmp_priority): New.
21681 (run_gcc): Read priorities and if doing parallel build order
21682 the Makefile by them.
21684 2018-04-30 David Malcolm <dmalcolm@redhat.com>
21686 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
21688 2018-04-30 Richard Biener <rguenther@suse.de>
21690 * tree-cfg.c (verify_address): Remove base argument, add
21691 flag whether to check TREE_ADDRESSABLE and do that.
21692 (verify_expr): Remove.
21693 (verify_types_in_gimple_reference): Add pieces from verify_expr.
21694 (verify_gimple_assign_single): Likewise.
21695 (verify_gimple_switch): Likewise.
21696 (verify_expr_location_1): Dereference tp once. Add (disabled)
21697 piece from verify_expr.
21698 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
21700 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
21702 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
21704 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
21706 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
21707 (small_data_pattern): Likewise.
21708 (arc_rewrite_small_data): Likewise.
21709 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
21710 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
21711 (get_symbol_alignment): New function.
21712 (legitimate_small_data_address_p): Likewise.
21713 (legitimate_scaled_address): Update, call
21714 legitimate_small_data_address_p.
21715 (output_sdata): New static variable.
21716 (arc_print_operand): Update how we handle small data operands.
21717 (arc_print_operand_address): Likewise.
21718 (arc_legitimate_address_p): Update, use
21719 legitimate_small_data_address_p.
21720 (arc_rewrite_small_data_p): Remove.
21721 (arc_rewrite_small_data_1): Likewise.
21722 (arc_rewrite_small_data): Likewise.
21723 (small_data_pattern): Likewise.
21724 (compact_sda_memory_operand): Update to use
21725 legitimate_small_data_address_p and get_symbol_alignment.
21726 (prepare_move_operands): Don't rewite sdata pattern.
21727 (prepare_extend_operands): Remove.
21728 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
21730 (zero_extendqisi2): Likewise.
21731 (zero_extendhisi2): Likewise.
21732 (extendqihi2): Likewise.
21733 (extendqisi2): Likewise.
21734 (extendhisi2): Likewise.
21735 (addsi3): Likewise.
21736 (subsi3): Likewise.
21737 (andsi3): Likewise.
21738 * config/arc/constraints.md (Usd): Change it to memory constraint.
21740 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
21742 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
21743 as source of std instructions.
21744 * config/arc/arc.md (movsi_insn): Update pattern predicate to
21745 allow 6-bit constants as source for store instructions.
21746 (movdi_insn): Update instruction pattern to allow 6-bit constants
21747 as source for store instructions.
21749 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
21751 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
21753 2018-04-30 Nathan Sidwell <nathan@acm.org>
21754 Sandra Loosemore <sandra@codesourcery.com>
21756 * dumpfile.c (dump_open): Allow '-' for stdout.
21757 * doc/invoke.texi (Developer Options): Document dump filename
21758 determination early. Document stdin/stdout selection.
21760 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
21762 Microblaze Target: PIC data text relative
21764 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
21765 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
21767 * config/microblaze/microblaze.h (microblaze_constant_address_p):
21768 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
21769 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
21770 New addressing mode for data-text relative position indepenedent code.
21771 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
21772 'ADDRESS_SYMBOLIC_TXT_REL'.
21773 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
21774 (microblaze_legitimate_pic_operand): Exclude function calls from
21775 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
21776 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
21778 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
21779 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
21780 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
21781 for 'address + offset'.
21782 (microblaze_expand_prologue): Add new function prologue call for
21784 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
21785 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
21786 table in case of TARGET_PIC_DATA_TEXT_REL.
21787 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
21788 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
21789 Add new macros 'UNSPEC_TEXT',
21790 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
21791 + exclude function calls from 'UNSPEC_PLT' in case of data text
21793 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
21794 new target hook for generating address diff vector tables in case of
21796 * doc/tm.texi : Regenerate.
21797 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
21798 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
21799 of addr diff vector generation.
21800 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
21801 target hook definition.
21802 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
21803 Add default function for generate_pic_addr_diff_vec -> flag_pic.
21804 * doc/invoke.texi (Add new pic option): Add new microblaze pic
21805 option for data text relative.
21807 2018-04-30 Richard Biener <rguenther@suse.de>
21809 * tree-chrec.h (evolution_function_is_constant_p): Remove
21811 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
21813 2018-04-30 Richard Biener <rguenther@suse.de>
21816 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
21818 2018-04-30 Richard Biener <rguenther@suse.de>
21820 PR tree-optimization/28364
21821 PR tree-optimization/85275
21822 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
21823 copying first exit test.
21825 2018-04-28 Mark Wielaard <mark@klomp.org>
21827 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
21828 dwarf_version >= 5.
21829 (dwarf_AT): Handle DW_AT_addr_base.
21830 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
21832 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
21835 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
21836 (*ashl<dwi>3_doubleword_mask_1): Ditto.
21837 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
21838 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
21840 2018-04-28 Richard Biener <rguenther@suse.de>
21842 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
21843 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
21844 to reflect use. Only add interesting stmts.
21846 2018-04-27 Martin Jambor <mjambor@suse.cz>
21849 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
21850 the jump function allows for passing through aggregate values.
21852 2018-04-27 David Malcolm <dmalcolm@redhat.com>
21854 * input.h (in_system_header_at): Convert from macro to inline
21856 (from_macro_expansion_at): Likewise.
21857 (from_macro_definition_at): Likewise.
21859 2018-04-27 Jeff Law <law@redhat.com>
21861 * config.gcc: Mark tile* targets as deprecated/obsolete.
21863 2018-04-27 Richard Biener <rguenther@suse.de>
21865 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
21868 2018-04-27 Richard Biener <rguenther@suse.de>
21870 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
21872 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
21874 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
21875 with Yd constraint. Set "preferred_for_speed" attribute from
21876 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
21877 with Yd constraint.
21878 (*movdi_internal): Ditto.
21879 (movti_interunit splitters): Remove
21880 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
21881 (movdi_interunit splitters): Ditto.
21882 * config/i386/constraints.md (Ye): Remove.
21883 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
21885 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21888 * config/aarch64/constraints.md (Usg): Limit to 31.
21889 (Usj): Limit to 63.
21891 2018-04-27 Jakub Jelinek <jakub@redhat.com>
21893 PR tree-optimization/85529
21894 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
21895 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
21896 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
21897 zero extension or masking of the MSB bit.
21898 (optimize_range_tests): Add FIRST_BB argument, pass it through
21899 to optimize_range_tests_var_bound.
21900 (maybe_optimize_range_tests, reassociate_bb): Adjust
21901 optimize_range_tests callers.
21903 2018-04-26 Richard Biener <rguenther@suse.de>
21904 Jakub Jelinek <jakub@redhat.com>
21906 * cgraph.h (symbol_table): Just declare debug method here.
21907 * symtab.c (symbol_table::debug): Define.
21909 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
21911 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
21913 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
21915 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
21916 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
21917 (*movdi_internal): Substitute Yi and Yj constraint with x
21918 and Ym and Yn constraint with y constraint. Update "isa"
21919 attribute and set "preferred_for_speed" attribute from
21920 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
21921 (*movsi_internal): Ditto.
21922 (*movdf_internal): Ditto.
21923 (*movsf_internal): Ditto.
21924 (*zero_extendsidi2): Ditto.
21925 * config/i386/sse.md (vec_set<mode>_0): Ditto.
21926 (sse2_loadld): Ditto.
21927 (*vec_extract<ssevecmodelower>_0): Ditto.
21928 (*vec_extractv4si_0_zext_sse4): Ditto.
21929 (vec_concatv2di): Ditto.
21930 (*vec_dup<mode>): Ditto.
21931 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
21932 * config/i386/constraints.md (Yi): Remove.
21937 2018-04-26 Nathan Sidwell <nathan@acm.org>
21939 * dumpfile.c (dump_open): New.
21940 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
21941 (dump_finish): Detect stdio/stderr by value not name.
21943 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
21945 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
21947 2018-04-26 Tom de Vries <tom@codesourcery.com>
21950 * config/nvptx/nvptx.c (verify_neutering_jumps)
21951 (verify_neutering_labels): New function
21952 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
21954 2018-04-26 Tom de Vries <tom@codesourcery.com>
21957 * config/nvptx/nvptx.c (needs_neutering_p): New function.
21958 (nvptx_single): Use needs_neutering_p to skip over insns that do not
21961 2018-04-26 Richard Biener <rguenther@suse.de>
21962 Tom de Vries <tom@codesourcery.com>
21965 * lto-streamer-out.c (output_function): Fixup loops if required to match
21966 discovery done in the reader.
21968 2018-04-26 Richard Biener <rguenther@suse.de>
21970 PR tree-optimization/85116
21971 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
21972 have a loop exit from the single latch predecessor. Remove
21973 case of header with just condition.
21974 (ch_base::copy_headers): Exclude infinite loops from any
21976 (pass_ch::execute): Record exits.
21978 2018-04-26 Richard Biener <rguenther@suse.de>
21980 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
21981 prologue cost vector and pass it to vect_get_load_cost.
21982 (vect_get_peeling_costs_all_drs): Likewise.
21983 (vect_peeling_hash_get_lowest_cost): Likewise.
21984 (vect_enhance_data_refs_alignment): Likewise.
21986 2018-04-26 Richard Biener <rguenther@suse.de>
21988 PR middle-end/85450
21989 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
21990 checking of integer<->pointer conversions.
21991 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
21992 sign-/zero-extending pointer types.
21993 (expand_omp_for_static_chunk): Likewise.
21995 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
21996 Jean Lee <xiaoyur347@gmail.com>
21998 * config/mips/mips.c (mips_asan_shadow_offset): New function.
21999 (TARGET_ASAN_SHADOW_OFFSET): Define.
22000 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
22001 true for -fsanitize=address.
22003 2018-04-25 Mark Wielaard <mark@klomp.org>
22005 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
22008 2018-04-25 Jakub Jelinek <jakub@redhat.com>
22010 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
22011 than "alu", remove explicit "memory" and "imm_disp" attributes.
22012 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
22014 PR middle-end/85414
22015 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
22016 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
22017 gen_lowpart_no_emit.
22019 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
22022 * config/i386/i386.c (ix86_expand_builtin): Change memory
22023 operand to XI, extend p0 to Pmode.
22024 * config/i386/i386.md: Change unspec volatile and operand
22025 1 mode to XI, change operand 0 mode to P.
22027 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
22029 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
22030 GET_MODE_MASK before any checking.
22031 (nds32_can_use_bset_p): Likewise.
22032 (nds32_can_use_btgl_p): Likewise.
22034 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
22036 * config/nds32/nds32-doubleword.md: New define_split pattern for
22037 illegal register number.
22039 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
22041 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
22043 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
22045 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
22047 2018-04-25 Richard Biener <rguenther@suse.de>
22049 * lto-streamer.h (LTO_major_version): Bump to 8.
22051 2018-04-25 Jakub Jelinek <jakub@redhat.com>
22053 * BASE-VER: Set to 9.0.0.
22055 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
22057 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
22058 in __abskf2 and __powikf2.
22060 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22063 * config/aarch64/constraints.md (Usg, Usj): New constraints.
22064 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
22065 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
22066 Use the above on operand 2. Reindent.
22067 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
22069 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
22072 * common/config/i386/i386-common.c (ix86_handle_option): Don't
22074 * config/i386/i386.opt (mcet): Removed.
22075 * doc/install.texi: Remove -mcet documentation.
22076 * doc/invoke.texi: Likewise.
22078 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
22081 * doc/install.texi: Remove -mcet from bootstrap-cet.
22083 2018-04-24 Jakub Jelinek <jakub@redhat.com>
22086 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
22087 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
22091 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
22092 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
22093 containing a CONST_VECTOR.
22095 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
22097 * doc/install.texi: Update newlib dependency for nvptx.
22099 2018-04-24 Jakub Jelinek <jakub@redhat.com>
22102 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
22103 instead of INTVAL when shifting x left.
22105 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
22107 PR tree-optimization/85478
22108 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
22109 vect_grouped_store_supported for single element vectors.
22111 2018-04-24 Richard Biener <rguenther@suse.de>
22114 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
22115 load cost increase to the case of non-constant step.
22117 2018-04-24 Jakub Jelinek <jakub@redhat.com>
22120 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
22121 destination if any_malformed_asm.
22123 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
22125 PR middle-end/85496
22126 * expr.c (store_field): In the bitfield case, if the value comes from
22127 a function call and is returned in registers by means of a PARALLEL,
22128 do not change the mode of the temporary unless BLKmode and VOIDmode.
22130 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
22132 PR rtl-optimization/85423
22133 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
22134 dependencies to debug insns when the previous insn is non-debug.
22136 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
22138 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
22139 enums into a single definition.
22140 (fls): Fix predicates and printing.
22143 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
22145 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
22146 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
22147 and short u6 immediate.
22148 (check_if_valid_sleep_operand): Remove.
22149 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
22151 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
22153 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
22154 flag_always_save_lp condition.
22155 * config/nds32/nds32.opt (malways-save-lp): New option.
22157 2018-04-22 Shiva Chen <shiva0217@gmail.com>
22159 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
22160 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
22161 * config/nds32/nds32.h
22162 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
22163 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
22165 2018-04-22 Shiva Chen <shiva0217@gmail.com>
22167 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
22168 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
22170 2018-04-22 Shiva Chen <shiva0217@gmail.com>
22171 Chung-Ju Wu <jasonwucj@gmail.com>
22173 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
22175 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
22176 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
22178 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
22180 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
22182 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
22184 * config/nds32/nds32-protos.h (nds32_data_alignment,
22185 nds32_local_alignment): Declare.
22186 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
22187 nds32_local_alignment): New functions.
22188 (TARGET_CONSTANT_ALIGNMENT): Define.
22189 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
22191 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
22193 * config/nds32/nds32.c
22194 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
22195 (TARGET_MODES_TIEABLE_P): Likewise.
22197 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
22199 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
22200 level Ofast and Og.
22202 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
22203 Chung-Ju Wu <jasonwucj@gmail.com>
22205 * config/nds32/constants.md (unspec_volatile_element): Add enum values
22206 for unaligned access.
22207 * config/nds32/nds32-intrinsic.c: Implementation of expanding
22209 * config/nds32/nds32-intrinsic.md: Likewise.
22210 * config/nds32/nds32_intrinsic.h: Likewise.
22211 * config/nds32/nds32.h (nds32_builtins): Likewise.
22212 * config/nds32/nds32.opt (munaligned-access): New option.
22213 * config/nds32/nds32.c (nds32_asm_file_start): Display
22214 flag_unaligned_access status.
22216 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
22218 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
22219 -mno-relax is present.
22220 * config/riscv/linux.h (LINK_SPEC): Ditto.
22222 2018-04-20 Martin Sebor <msebor@redhat.com>
22225 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
22227 (gimple_fold_builtin_stxcpy_chk): Same.
22228 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
22230 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
22233 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
22234 __powikf2 when long double is IEEE 128-bit.
22236 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
22238 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
22239 step to make sure stack always aligned.
22241 2018-04-20 Carl Love <cel@us.ibm.com>
22244 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
22245 size check for arg0.
22247 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
22248 Tom de Vries <tom@codesourcery.com>
22251 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
22252 Emit insns for calls too.
22253 (nvptx_find_par): Always look for worker-level predecessor insn.
22254 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
22256 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
22257 (nvptx_process_pars): Propagate frames for calls.
22259 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
22262 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
22264 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
22265 (ix86_handle_option): Don't handle OPT_mibt.
22266 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
22268 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
22270 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
22271 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
22272 (ix86_target_macros): Define __CET__ with flag_cf_protection
22273 for -fcf-protection.
22274 * config/i386/i386.c (isa2_opts): Remove -mibt.
22275 * config/i386/i386.h (TARGET_IBT): Removed.
22276 (TARGET_IBT_P): Likewise.
22277 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
22278 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
22279 * config/i386/i386.opt (mcet): Update help message.
22280 (mshstk): Likewise.
22282 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
22283 -mcet as an alias for -mshstk.
22285 2018-04-20 Richard Biener <rguenther@suse.de>
22287 PR middle-end/85475
22288 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
22289 complexity by forcing a single use of the multiply operand.
22291 2018-04-20 Martin Jambor <mjambor@suse.cz>
22294 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
22295 recursion dependency to only apply to non-clones.
22297 2018-04-20 Martin Jambor <mjambor@suse.cz>
22300 * ipa-cp.c (create_specialized_node): Check that clones of
22301 self-recursive edges exist during IPA-CP.
22303 2018-04-19 Toon Moene <toon@moene.org>
22305 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
22308 2018-04-19 Jakub Jelinek <jakub@redhat.com>
22310 PR tree-optimization/85467
22311 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
22312 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
22313 VECTOR_CST element to type.
22315 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
22318 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
22319 * config/i386/i386.md (builtin_setjmp_setup): Removed.
22320 (builtin_longjmp): Likewise.
22321 (save_stack_nonlocal): New pattern.
22322 (restore_stack_nonlocal): Likewise.
22324 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
22327 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
22328 Replace ASM_OUTPUT_LABEL with fprintf.
22330 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
22333 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
22334 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
22335 * config/i386/i386-c.c (ix86_target_macros_internal): Also
22336 define __IBT__ and __SHSTK__ for -fcf-protection.
22337 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
22339 (ix86_trampoline_init): Likewise.
22340 (x86_output_mi_thunk): Likewise.
22341 (ix86_notrack_prefixed_insn_p): Likewise.
22342 (ix86_option_override_internal): Don't disallow -fcf-protection.
22343 * config/i386/i386.md (rdssp<mode>): Also enable for
22345 (incssp<mode>): Likewise.
22346 (nop_endbr): Likewise.
22347 * config/i386/i386.opt (mcet): Change help message to built-in
22350 (mshstk): Likewise.
22351 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
22352 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
22353 enable CET built-in functions.
22355 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
22357 * common/config/i386/i386-common.c
22358 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
22359 OPTION_MASK_ISA_MOVDIRI_UNSET,
22360 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
22361 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
22362 * config.gcc (movdirintrin.h): New header.
22363 * config/i386/cpuid.h (bit_MOVDIRI,
22364 bit_MOVDIR64B): New bits.
22365 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
22367 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
22368 (VOID, PVOID, PCVOID)): New function types.
22369 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
22370 __builtin_ia32_directstoreu_u64,
22371 __builtin_ia32_movdir64b): New builtins.
22372 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
22373 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
22375 (ix86_valid_target_attribute_inner_p): Ditto.
22376 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
22377 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
22378 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
22379 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
22380 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
22381 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
22382 (movdiri<mode>, movdir64b_<mode>): New.
22383 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
22384 * config/i386/immintrin.h: Include movdirintrin.h.
22385 * config/i386/movdirintrin.h: New file.
22386 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
22388 2018-04-19 Richard Biener <rguenther@suse.de>
22390 PR middle-end/85455
22391 * cfg.c (clear_bb_flags): When loop state says we have
22392 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
22394 2018-04-19 Richard Biener <rguenther@suse.de>
22396 PR tree-optimization/84737
22397 * tree-vect-data-refs.c (vect_copy_ref_info): New function
22398 copying restrict info.
22399 (vect_setup_realignment): Use it.
22400 * tree-vectorizer.h (vect_copy_ref_info): Declare.
22401 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
22402 the first DR to all generated stores.
22403 (vectorizable_load): Likewise for loads.
22405 2018-04-19 Jakub Jelinek <jakub@redhat.com>
22407 PR tree-optimization/85446
22408 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
22409 the integral and pointer types to have the same precision.
22411 * doc/install.texi: Document --disable-cet being the default and
22414 2018-04-18 Martin Liska <mliska@suse.cz>
22416 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
22419 2018-04-18 Martin Liska <mliska@suse.cz>
22422 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
22425 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
22426 arguments if they are comparable.
22428 2018-04-18 Martin Liska <mliska@suse.cz>
22431 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
22434 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
22437 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
22440 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
22441 ENDBR after calling __morestack.
22443 2018-04-18 David Malcolm <dmalcolm@redhat.com>
22446 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
22447 by using gcc_base_ver to generate a gcc_driver_version, and use
22448 it when generating GCC_DRIVER_NAME.
22449 * configure: Regenerate.
22451 2018-04-18 Jakub Jelinek <jakub@redhat.com>
22454 * config.gcc: Obsolete powerpc*-*-*spe*.
22456 2018-04-17 Jakub Jelinek <jakub@redhat.com>
22459 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
22460 (stabstr_D): Change type of unum from unsigned int to
22461 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
22464 2018-04-17 Jim Wilson <jimw@sifive.com>
22467 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
22468 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
22469 Set arg_pointer_offset after using pretend_args_size.
22471 2018-04-17 Jakub Jelinek <jakub@redhat.com>
22473 PR rtl-optimization/85431
22474 * dse.c (record_store): Ignore zero width stores.
22477 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
22478 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
22479 __builtin_stack_restore rather than after it.
22480 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
22481 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
22482 argument instead of virtual_dynamic_stack_rtx.
22484 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
22486 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
22488 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22489 Add note to error message to explain internal mapping of overloaded
22490 built-in function name to non-overloaded built-in function name.
22491 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
22494 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
22497 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
22498 where the inputs overlap with the output.
22500 2018-04-17 Jakub Jelinek <jakub@redhat.com>
22502 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
22503 (=v, v) alternative and explicit "memory" attribute.
22504 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
22505 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
22507 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
22508 "sselog1" type instead of "sselog".
22509 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
22510 "sselog". Remove explicit "memory" attribute.
22511 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
22512 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
22514 (vec_extract_hi_v32hi): Merge all alternatives into one, use
22515 "sselog1" type instead of "sselog". Remove explicit "memory"
22517 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
22518 use "sselog1" type instead of "sselog". Remove explicit "memory"
22520 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
22521 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
22523 (vec_extract_hi_v64qi): Merge all alternatives into one, use
22524 "sselog1" type instead of "sselog". Remove explicit "memory"
22526 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
22527 use "sselog1" type instead of "sselog". Remove explicit "memory"
22531 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
22533 PR middle-end/85414
22534 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
22537 2018-04-17 Martin Jambor <mjambor@suse.cz>
22540 * ipa-cp.c (create_specialized_node): Call
22541 expand_all_artificial_thunks if necessary.
22543 2018-04-17 Martin Liska <mliska@suse.cz>
22546 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
22547 in message, remote space in between '_G' and '('.
22549 2018-04-17 Jakub Jelinek <jakub@redhat.com>
22552 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
22553 avx512f_vmcmp<mode>3<round_saeonly_name>,
22554 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
22555 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
22556 avx512f_rndscale<mode><round_saeonly_name>,
22557 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
22558 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
22559 Use %<iptr>2 instead of %2 for -masm=intel.
22560 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
22561 avx512f_vcvttss2usi<round_saeonly_name>,
22562 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
22564 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
22565 avx512f_vcvttsd2usi<round_saeonly_name>,
22566 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
22567 Use %q1 instead of %1 for -masm=intel.
22568 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
22569 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
22570 of %3 for -masm=intel.
22571 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
22573 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
22575 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
22577 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
22578 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
22580 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
22582 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
22583 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
22584 %g1 and one with %0 and %1.
22585 (avx512er_vmrcp28<mode><round_saeonly_name>,
22586 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
22587 %1 for -masm=intel.
22588 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
22589 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
22590 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
22591 of %0 and %{%4%} for -masm=intel.
22592 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
22593 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
22594 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
22595 order of %0 and %{%5%}%{z%} for -masm=intel.
22597 2018-04-17 Jan Hubicka <jh@suse.cz>
22600 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
22602 2018-04-17 Martin Liska <mliska@suse.cz>
22605 * multiple_target.c (create_dispatcher_calls): Set apostrophes
22606 for target_clone error message. Make default implementation
22607 clone to be a local declaration.
22608 (separate_attrs): Add new argument and check for an empty
22610 (expand_target_clones): Handle it.
22611 (ipa_target_clone): Make redirection just for target_clones
22614 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
22615 Tom de Vries <tom@codesourcery.com>
22617 PR middle-end/84955
22618 * omp-expand.c (expand_oacc_for): Add dummy false branch for
22619 tiled basic blocks without omp continue statements.
22621 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
22624 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
22625 vec_extract expression as having side effects to make sure it gets
22628 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
22631 * config/i386/i386.c (get_builtin_code_for_version): Check
22634 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
22637 * config.gcc: Support "skylake".
22638 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
22640 * config/i386/i386.c (m_SKYLAKE): Define.
22641 (processor_target_table): Add "skylake".
22642 (ix86_option_override_internal): Add "skylake".
22643 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
22644 PROCESSOR_CANNONLAKE.
22645 (get_builtin_code_for_version): Fix priority for
22646 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
22647 PROCESSOR_SKYLAKE-AVX512.
22648 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
22649 (processor_type): Add PROCESSOR_SKYLAKE.
22651 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
22652 Jason Merrill <jason@redhat.com>
22655 * convert.c (convert_to_integer_1): Use direct recursion for
22656 enumeral types and types with a precision less than the number
22657 of bits in their mode.
22659 2018-04-16 Julia Koval <julia.koval@intel.com>
22662 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
22663 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
22665 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
22668 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
22669 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
22670 and -mno-direct-move.
22672 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
22675 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
22676 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
22677 Ensure negative shifts result in {0}.
22679 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
22681 PR rtl-optimization/79916
22682 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
22683 regs (if any) to define how to gnerate SD moves when LRA is in
22686 2018-04-13 Jakub Jelinek <jakub@redhat.com>
22688 PR rtl-optimization/85393
22689 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
22690 * except.c (expand_dw2_landing_pad_for_region): Make static.
22691 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
22692 a label and unconditional jump to old_bb, rather than
22693 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
22696 PR rtl-optimization/85376
22697 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
22698 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
22699 instead of a specific value.
22701 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
22702 Bin Cheng <bin.cheng@arm.com>
22704 PR tree-optimization/82965
22705 PR tree-optimization/83991
22706 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
22707 by_profile_only parameter.
22708 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
22709 information if the loop was predicted to iterate too many times.
22710 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
22712 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
22715 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
22718 2018-04-13 Martin Liska <mliska@suse.cz>
22719 Jakub Jelinek <jakub@redhat.com>
22721 PR middle-end/81657
22722 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
22723 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
22724 * builtins.c (expand_builtin_memory_copy_args): Use
22725 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
22726 handle dest_addr == pc_rtx.
22728 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
22731 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
22732 asked to not generate direct moves.
22733 (fix_trunc<mode>si2_stfiwx): Similar.
22734 (fix_trunc<mode>si2_internal): Similar.
22736 2018-04-12 Jakub Jelinek <jakub@redhat.com>
22739 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
22740 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
22741 lookup if dest in some wider mode is known to be const0_rtx and
22742 if so, record permanent equivalence for it to be ZERO_EXTEND of
22743 the narrower mode destination.
22745 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
22747 * lto-streamer-out.c (output_function): Revert 259346.
22748 * omp-expand.c (expand_oacc_for): Likewise.
22750 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
22752 PR rtl-optimization/85354
22753 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
22754 * sel-sched.c (sel_global_init): ... here.
22756 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
22759 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
22760 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
22761 mode for PE-COFF targets.
22762 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
22763 (i386_pe_asm_lto_end): Likewise.
22764 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
22765 (TARGET_ASM_LTO_END): Likewise.
22766 * config/i386/winnt.c (saved_debug_info_level): New static variable.
22767 (i386_pe_asm_lto_start): New function.
22768 (i386_pe_asm_lto_end): Likewise.
22770 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
22771 Richard Biener <rguenther@suse.de>
22773 PR middle-end/84955
22774 * lto-streamer-out.c (output_function): Fix CFG loop state before
22776 * omp-expand.c (expand_oacc_for): Handle calls to internal
22777 functions like regular functions.
22779 2018-04-12 Richard Biener <rguenther@suse.de>
22782 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
22783 for the early LTO debug to properly generate references to it
22784 during DIE emission. Do not re-use that for the skeleton for
22786 (dwarf2out_early_finish): Likewise.
22788 2018-04-12 Jakub Jelinek <jakub@redhat.com>
22791 * config/i386/sse.md
22792 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
22793 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
22794 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
22795 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
22796 and output is a reg, avoid creating invalid lowpart subreg, but
22797 instead split into a 512-bit move. Don't split if not AVX512VL,
22798 input is xmm16+ reg and output is a mem.
22799 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
22800 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
22801 xmm16+ reg and output is a mem.
22803 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22805 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
22806 also for flag_dwarf2_cfi_asm.
22808 2018-04-12 Jakub Jelinek <jakub@redhat.com>
22810 PR rtl-optimization/85342
22811 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
22812 a bool scalar var inside of the loop instead. Don't try to update
22813 recog_data.operand after failed apply_change_group.
22815 2018-04-12 Tom de Vries <tom@codesourcery.com>
22818 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
22819 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
22820 array with flexible array member as array without given dimension.
22821 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
22822 argument for undefined param to true.
22824 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
22827 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
22828 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
22829 from PowerPC section.
22830 * config/rs6000/sysv4.opt (mcall-): Improve help text.
22831 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
22832 help text that is too long.
22833 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
22834 help text that is too long.
22835 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
22836 help text that is too long.
22838 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
22840 * config/alpha/alpha.md (stack_probe_internal): Rename
22841 from "probe_stack". Update all callers.
22843 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
22845 PR rtl-optimization/84566
22846 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
22847 sched_macro_fuse_insns.
22849 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
22852 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
22853 (compute_block_dependences): ... from here.
22855 2018-04-11 Jakub Jelinek <jakub@redhat.com>
22857 PR tree-optimization/85331
22858 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
22859 from int to HOST_WIDE_INT.
22861 2018-04-11 Martin Jambor <mjambor@suse.cz>
22864 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
22865 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
22866 not the same as the source val.
22867 (cgraph_edge_brings_value_p): New parameter.
22868 (gather_edges_for_value): Pass destination value to
22869 cgraph_edge_brings_value_p.
22870 (perhaps_add_new_callers): Likewise.
22871 (get_info_about_necessary_edges): Likewise and exclude values brought
22872 only by self-recursive edges.
22873 (create_specialized_node): Redirect only clones of self-calling edges.
22874 (+self_recursive_pass_through_p): New function.
22875 (find_more_scalar_values_for_callers_subset): Use it.
22876 (find_aggregate_values_for_callers_subset): Likewise.
22877 (known_aggs_to_agg_replacement_list): Removed.
22878 (decide_whether_version_node): Re-calculate known constants for all
22879 remaining context clones.
22881 2018-04-11 Richard Biener <rguenther@suse.de>
22884 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
22885 from early DWARF output.
22886 (dwarf2out_early_finish): Output line info unconditionally into
22887 early DWARF and add reference to it.
22889 2018-04-11 Jakub Jelinek <jakub@redhat.com>
22892 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
22893 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
22894 other than V2DFmode using iptr mode attribute.
22895 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
22897 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
22899 PR rtl-optimization/84659
22900 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
22902 2018-04-11 Jakub Jelinek <jakub@redhat.com>
22905 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
22907 (output_loc_list): Pass address of a dummy size variable even in the
22908 locview handling loop.
22909 (index_location_lists): Add comment on why skip_loc_list_entry can't
22912 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
22915 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
22918 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
22921 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
22922 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
22923 and -mstring-compare-inline-limit.
22925 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
22928 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
22929 for stack clash protection in a register whenever we need it to be in
22932 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
22934 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
22935 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
22937 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
22940 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
22942 (mlong-double-): Ditto.
22943 * config/rs6000/sysv4.opt (msdata=): Ditto.
22944 (mtls-size=): Ditto.
22946 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22948 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
22949 erroneous entries for
22950 "vector int vec_ldl (int, long int *)", and
22951 "vector unsigned int vec_ldl (int, unsigned long int *)".
22952 Add comments and entries for
22953 "vector bool char vec_ldl (int, bool char *)",
22954 "vector bool short vec_ldl (int, bool short *)",
22955 "vector bool int vec_ldl (int, bool int *)",
22956 "vector bool long long vec_ldl (int, bool long long *)",
22957 "vector pixel vec_ldl (int, pixel *)",
22958 "vector long long vec_ldl (int, long long *)",
22959 "vector unsigned long long vec_ldl (int, unsigned long long *)".
22960 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
22961 type tree bool_long_long_type_node and correct definition of
22962 bool_V2DI_type_node to make reference to this new type tree.
22963 (rs6000_mangle_type): Replace erroneous reference to
22964 bool_long_type_node with bool_long_long_type_node.
22965 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
22966 comments to emphasize sign distinctions for char and int types and
22967 replace RS6000_BTI_bool_long constant with
22968 RS6000_BTI_bool_long_long constant. Also add comment to restrict
22969 use of RS6000_BTI_pixel.
22970 (bool_long_type_node): Remove this macro definition.
22971 (bool_long_long_type_node): New macro definition
22973 2018-04-10 Jakub Jelinek <jakub@redhat.com>
22975 PR rtl-optimization/85300
22976 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
22977 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
22978 simplify_unary_operation fails.
22980 2018-04-10 Martin Liska <mliska@suse.cz>
22982 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
22983 cgraph_edge and ipa_ref.
22985 2018-04-10 Jakub Jelinek <jakub@redhat.com>
22989 * config/i386/sse.md
22990 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
22991 computation of the VEC_MERGE selector from mask.
22992 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
22993 Fix decoding of the VEC_MERGE selector into mask.
22995 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
22997 PR tree-optimization/85286
22998 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
23000 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
23002 * final.c (final_1): Set insn_last_address as well as
23003 insn_current_address.
23005 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23008 * explow.c (emit_stack_probe): Call validize_mem on memory location
23009 before passing it to gen_probe_stack. Create address operand and
23010 legitimize it for the probe_stack_address case.
23012 2018-04-09 Jan Hubicka <jh@suse.cz>
23015 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
23016 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
23017 * tree.c (free_lang_data_in_type): Fix handling of binfos;
23019 (free_lang_data): Rebuild type inheritance graph.
23021 2018-04-09 Martin Sebor <msebor@redhat.com>
23023 * invoke.texi (-finline-small-functions): Mention other optimization
23025 (-findirect-inlining, -fpartial-inlining): Same.
23026 (-finline-functions-called-once): Same.
23027 (-freorder-blocks-and-partition): Same.
23029 2018-04-09 Jan Hubicka <jh@suse.cz>
23032 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
23033 jumps; choose last target that matches the criteria (i.e.
23034 no partition changes for non-crossing jumps).
23035 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
23036 support for redirecting crossing jumps to non-crossing.
23038 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
23040 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
23041 also for naked functions.
23043 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
23045 * config/arc/arc.md (add_shift): New pattern.
23046 (add_shift2): Likewise.
23047 (sub_shift): Likewise.
23048 (sub_shift_cmp0_noout): Likewise.
23049 (compare_si_ashiftsi): Likewise.
23050 (xbfu_cmp0_noout): New combine pattern.
23051 (xbfu_cmp0"): Likewise.
23052 (movsi_set_cc_insn): Place the predicable variant first.
23053 (commutative_binary_cmp0_noout): Remove clobber.
23054 (commutative_binary_cmp0): New pattern.
23055 (noncommutative_binary_cmp0): Likewise.
23056 (noncommutative_binary_cmp0_noout): Likewise.
23057 (noncommutative_binary_comparison_result_used): Removed.
23058 (rsub_cmp0): New pattern.
23059 (rsub_cmp0_noout): Likewise.
23060 (extzvsi): Changed, keep only meaningful variants.
23061 (SQH, SEZ): New iterators.
23062 (SQH_postfix): New mode attribute.
23063 (SEZ_prefix): New code attribute.
23064 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
23065 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
23066 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
23067 of numerical value.
23068 (noncommutative_operator): Check the availability of barrel
23071 2018-04-09 Richard Biener <rguenther@suse.de>
23073 PR tree-optimization/85284
23074 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
23075 Only use the niter constraining form of simple_iv when the exit
23076 is always executed.
23078 2018-04-09 Tom de Vries <tom@codesourcery.com>
23081 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
23082 (define_expand "*memory_barrier"): New define_expand.
23083 (define_insn "memory_barrier"): New insn.
23085 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
23087 PR rtl-optimization/80463
23088 PR rtl-optimization/83972
23089 PR rtl-optimization/83480
23091 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
23092 correct producer for the insn.
23093 (tidy_control_flow): Fixup seqnos in case of debug insns.
23095 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
23097 PR rtl-optimization/83913
23099 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
23100 different sched-times when merging exprs.
23102 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
23104 PR rtl-optimization/83962
23106 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
23107 tidy_fallthru_edge and tidy_control_flow.
23109 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
23111 PR rtl-optimization/83530
23113 * sel-sched.c (force_next_insn): New global variable.
23114 (remove_insn_for_debug): When force_next_insn is true, also leave only
23115 next insn in the ready list.
23116 (sel_sched_region): When the region wasn't scheduled, make another pass
23117 over it with force_next_insn set to 1.
23119 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
23121 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
23123 * config/nds32/constants.md (unspec_volatile_element): Add enum values
23124 for interrupt control.
23125 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
23126 functions for interrupt control.
23127 * config/nds32/nds32-intrinsic.md: Likewise.
23128 * config/nds32/nds32_intrinsic.h: Likewise.
23129 * config/nds32/nds32.h (nds32_builtins): Likewise.
23131 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
23133 * config/nds32/nds32.c (nds32_init_machine_status,
23134 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
23135 strict_aligned_p field.
23136 (nds32_expand_to_rtl_hook): New function.
23137 (TARGET_EXPAND_TO_RTL_HOOK): Define.
23138 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
23140 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
23141 Chung-Ju Wu <jasonwucj@gmail.com>
23143 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
23144 * config/nds32/nds32-n7.md: New file.
23145 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
23146 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
23148 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
23149 * config/nds32/nds32.md (pipeline_model): Add n7.
23150 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
23151 * config/nds32/pipelines.md: Include n7 settings.
23153 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
23154 Chung-Ju Wu <jasonwucj@gmail.com>
23156 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
23157 * config/nds32/nds32-e8.md: New file.
23158 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
23159 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
23161 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
23162 * config/nds32/nds32.md (pipeline_model): Add e8.
23163 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
23164 * config/nds32/pipelines.md: Include e8 settings.
23166 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
23167 Chung-Ju Wu <jasonwucj@gmail.com>
23169 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
23170 * config/nds32/nds32-n8.md: New file.
23171 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
23172 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
23174 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
23175 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
23176 * config/nds32/nds32.md (pipeline_model): Add n8.
23177 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
23178 * config/nds32/pipelines.md: Include n8 settings.
23180 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
23181 Chung-Ju Wu <jasonwucj@gmail.com>
23183 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
23184 * config/nds32/nds32-n9-2r1w.md: New file.
23185 * config/nds32/nds32-n9-3r2w.md: New file.
23186 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
23187 nds32_register_ports): New or modify for cpu n9.
23188 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
23190 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
23191 * config/nds32/nds32-utils.c: New file.
23192 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
23193 TARGET_MUL_SLOW): Define.
23194 * config/nds32/nds32.md (pipeline_model): New attribute.
23195 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
23196 New options that support cpu n9.
23197 * config/nds32/pipelines.md: Include n9 settings.
23198 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
23200 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
23202 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
23203 information if necessary.
23204 (output_cond_branch_compare_zero): Likewise.
23205 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
23206 (nds32_target_alignment): Refine for alignment.
23207 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
23208 (FUNCTION_BOUNDARY): Modify.
23209 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
23211 * config/nds32/nds32.opt (malways-align, malign-functions): New.
23213 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
23215 * config/nds32/constants.md (unspec_volatile_element): Add values for
23216 TLB operation and data prefetch.
23217 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
23218 functions for TLB operation and data prefetch.
23219 * config/nds32/nds32-intrinsic.md: Likewise.
23220 * config/nds32/nds32_intrinsic.h: Likewise.
23221 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
23222 (nds32_print_operand): Likewise.
23223 * config/nds32/nds32.h (nds32_builtins): Likewise.
23225 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
23226 Andrew Pinski <pinsika@gcc.gnu.org>
23228 PR middle-end/82976
23229 * match.pd: Use constant_boolean_node of correct type instead of
23230 boolean_true_node or boolean_false_node for simplifying
23231 pointer comparisons to zero.
23233 2018-04-07 Jakub Jelinek <jakub@redhat.com>
23235 PR tree-optimization/80021
23236 * tree.c (verify_type_variant): Make error call in verify_variant_match
23237 translatable and remove final full stop.
23239 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
23241 * config/nds32/constants.md (unspec_volatile_element): Add
23242 UNSPEC_VOLATILE_EH_RETURN.
23243 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
23244 nds32_output_stack_pop): Support dwarf exception handling process.
23245 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
23246 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
23247 exception handling process.
23248 (nds32_compute_stack_frame): Likewise.
23249 (nds32_return_addr_rtx): Likewise.
23250 (nds32_initial_elimination_offset): Likewise.
23251 (nds32_expand_prologue): Likewise.
23252 (nds32_expand_epilogue): Likewise.
23253 (nds32_dynamic_chain_address): New function.
23254 * config/nds32/nds32.h (machine_function): Add fields for dwarf
23255 exception handling.
23256 (DYNAMIC_CHAIN_ADDRESS): Define.
23257 (EH_RETURN_DATA_REGNO): Define.
23258 (EH_RETURN_STACKADJ_RTX): Define.
23259 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
23260 patterns for dwarf exception handling.
23262 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
23264 * config/nds32/nds32.h: Clean up obsolete macros.
23266 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
23268 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
23269 Add enum values for particular instructions.
23270 * config/nds32/nds32-intrinsic.c: Implementation of expanding
23271 particular intrinsic functions.
23272 * config/nds32/nds32-intrinsic.md: Likewise.
23273 * config/nds32/nds32_intrinsic.h: Likewise.
23274 * config/nds32/nds32.h (nds32_builtins): Likewise.
23275 * config/nds32/nds32.md (type): Add pbsad and pbsada.
23276 (btst, ave): New patterns for particular instructions.
23278 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
23280 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
23281 Add enum values for atomic load/store and memory sync.
23282 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
23284 * config/nds32/nds32-intrinsic.md: Likewise.
23285 * config/nds32/nds32_intrinsic.h: Likewise.
23286 * config/nds32/nds32.h (nds32_builtins): Likewise.
23288 2018-04-07 Jakub Jelinek <jakub@redhat.com>
23290 PR tree-optimization/85257
23291 * fold-const.c (native_encode_vector): If not all elts could fit
23292 and off is -1, return 0 rather than offset.
23293 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
23294 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
23295 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
23296 adjust buffer in native_interpret_expr call.
23298 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
23300 * config/nds32/constants.md (unspec_volatile_element): Add cache
23301 control enum values.
23302 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
23303 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
23304 * config/nds32/nds32.c (nds32_cctl_names): New.
23305 (nds32_print_operand): Handle cache control register names.
23306 * config/nds32/nds32.h (nds32_builtins): New enum values.
23307 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
23309 * config/nds32/nds32.md (type): Add mmu.
23310 * config/nds32/pipelines.md (simple_insn): Add mmu.
23312 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
23314 * config/nds32/nds32.md (type): Remove call.
23315 * config/nds32/pipelines.md (simple_insn): Likewise.
23317 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
23319 * config/nds32/constants.md (unspec_volatile_element): Add
23320 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
23321 UNSPEC_VOLATILE_FMFCFG.
23322 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
23323 description for fmfcfg and fmfcsr.
23324 (bdesc_1arg): Add fmtcsr.
23325 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
23326 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
23327 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
23328 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
23329 unspec_fmfcfg): New patterns.
23330 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
23331 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
23332 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
23333 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
23334 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
23335 __nds32__fmfcfg): Define.
23337 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
23339 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
23340 intrinsic register names.
23341 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
23342 intrinsic register enum values and macros.
23344 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
23346 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
23347 for load/store addressing form.
23348 (nds32_print_operand_address): Likewise.
23350 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
23353 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
23354 based on LABEL_REF. Remove useless assertion.
23355 (pic_address_needs_scratch): Fix formatting.
23356 (sparc_legitimize_pic_address): Minor tweaks.
23357 (sparc_delegitimize_address): Adjust assertion accordingly.
23358 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
23359 into symbolic_operand.
23360 (movsi_high_pic_label_ref): Likewise.
23361 (movsi_lo_sum_pic_label_ref): Likewise.
23362 (movdi_pic_label_ref): Likewise.
23363 (movdi_high_pic_label_ref): Likewise.
23364 (movdi_lo_sum_pic_label_ref): Likewise.
23366 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
23368 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
23369 custom LIB_SPEC setup.
23371 2018-04-06 Ruslan Bukin <br@bsdpad.com>
23372 Kito Cheng <kito.cheng@gmail.com>
23374 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
23375 * config/riscv/freebsd.h: New.
23377 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
23379 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
23380 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
23383 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
23384 Kito Cheng <kito.cheng@gmail.com>
23386 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
23387 nds32_output_call, nds32_symbol_binds_local_p): New functions.
23388 * config/nds32/nds32-protos.h (nds32_output_call,
23389 nds32_output_return): Declare.
23390 * config/nds32/nds32.md: Refine all the call and return patterns.
23392 2018-04-06 Jakub Jelinek <jakub@redhat.com>
23395 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
23396 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
23398 PR rtl-optimization/84872
23399 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
23400 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
23401 EDGE_CROSSING edge.
23403 2018-04-06 Tamar Christina <tamar.christina@arm.com>
23405 * expr.c (copy_blkmode_to_reg): Revert 254862.
23406 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
23408 2018-04-06 Richard Biener <rguenther@suse.de>
23410 PR middle-end/85244
23411 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
23412 after seeing a component reference with an adjacent field. Treat
23413 refs to arrays at struct end of external decls similar to
23414 refs to unconstrained commons.
23416 2018-04-06 Jakub Jelinek <jakub@redhat.com>
23419 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
23420 look through SAVE_EXPRs with non-side-effects argument. Adjust
23422 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
23425 2018-04-06 Richard Biener <rguenther@suse.de>
23427 PR middle-end/85180
23428 * alias.c (find_base_term): New wrapper around find_base_term
23429 unwinding CSELIB_VAL_PTR changes.
23430 (find_base_term): Do not restore CSELIB_VAL_PTR during the
23433 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23435 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
23437 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
23438 constant definitions.
23439 ("nop"): lr 0,0 -> nopr r0
23440 ("nop_lr0", "nop_lr1"): New insn definitions.
23442 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
23444 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
23445 NDS32_V3PUSH_AVAILABLE_P macro.
23447 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
23448 Chung-Ju Wu <jasonwucj@gmail.com>
23450 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
23451 (nds32*-*-*): Add float and fpu_config into supported_defaults.
23452 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
23453 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
23454 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
23455 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
23456 * config/nds32/constraints.md: New constraints and checking for hard
23457 float configuration.
23458 * config/nds32/iterators.md: New mode iterator and attribute for hard
23459 float configuration.
23460 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
23462 * config/nds32/nds32-fpu.md: New file.
23463 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
23464 deal with hard float code generation.
23465 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
23467 (abi_type, float_reg_number): New enum type.
23468 * config/nds32/nds32-predicates.c: New predicates for hard float.
23469 * config/nds32/nds32-protos.h: Declare functions for hard float.
23470 * config/nds32/nds32.c: Implementation for hard float configuration.
23471 * config/nds32/nds32.h: Definitions for hard float configuration.
23472 * config/nds32/nds32.md: Include hard float machine description and
23473 modify patterns for hard float configuration.
23474 * config/nds32/nds32.opt: New options for hard float configuration.
23475 * config/nds32/predicates.md: New predicates for hard float
23478 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
23480 * common/config/nds32/nds32-common.c
23481 (nds32_option_optimization_table): Enable -mreleax-hint by default.
23483 2018-04-05 Jakub Jelinek <jakub@redhat.com>
23485 PR middle-end/85195
23486 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
23487 CONSTRUCTOR_ELT (ctor, ...)->value.
23489 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
23492 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
23494 2018-04-05 Tom de Vries <tom@codesourcery.com>
23497 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
23500 2018-04-05 Shiva Chen <shiva0217@gmail.com>
23501 Kito Cheng <kito.cheng@gmail.com>
23503 * config/nds32/constraints.md (U33): Fine-tune checking condition.
23504 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
23505 * config/nds32/nds32.h (nds32_16bit_address_type): Add
23506 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
23508 2018-04-05 Shiva Chen <shiva0217@gmail.com>
23509 Kito Cheng <kito.cheng@gmail.com>
23511 * config/nds32/constraints.md (Ufe): New memory constraint.
23512 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
23513 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
23514 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
23516 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
23517 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
23519 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
23521 * config/nds32/nds32.md: Use optimize_size in the condition for
23522 alu-shift instructions.
23524 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
23526 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
23528 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
23530 * config/nds32/nds32.md (negsi2): Refine pattern.
23532 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
23533 Chung-Ju Wu <jasonwucj@gmail.com>
23535 * config/nds32/iterators.md (shift_rotate): New code iterator.
23536 (shift): New code attribute.
23537 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
23538 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
23539 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
23540 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
23541 bit-wise operations.
23542 (andsi3, *andsi3): Ditto.
23543 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
23544 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
23545 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
23546 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
23547 nds32_ior_operand, nds32_xor_operand): New predicates.
23549 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
23551 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
23552 (addsi3, subsi3): ... this.
23554 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
23556 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
23558 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
23560 * config/nds32/nds32.md: Adjust indention.
23562 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
23564 * config/nds32/nds32.md (feature): New attribute.
23566 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
23568 * config/nds32/nds32.md (subtype): New attribute.
23570 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23573 * config/arm/arm-builtins.c (arm_expand_builtin): Change
23574 expansion to perform a bitwise AND of the argument followed by a
23575 boolean negation of the result.
23577 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
23579 PR rtl-optimization/84878
23580 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
23581 the basic block. Assert the use reference is not artificial and that
23582 it has an associated insn.
23584 2018-04-04 Michael Matz <matz@suse.de>
23586 * builtins.c (compute_objsize): Pass correct operand
23587 to array_at_struct_end_p.
23589 2018-04-04 Richard Biener <rguenther@suse.de>
23592 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
23593 from contexts for DINFO_LEVEL_TERSE and below.
23595 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
23597 * config/nds32/nds32-doubleword.md (move_<mode>): Require
23598 resiter_operand condition.
23599 * config/nds32/nds32.md (*move<mode>): Ditto.
23601 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
23602 Monk Chiang <sh.chiang04@gmail.com>
23604 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
23606 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
23608 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
23610 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
23611 Kito Cheng <kito.cheng@gmail.com>
23613 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
23614 nds32_cond_code_str, output_cond_branch,
23615 output_cond_branch_compare_zero, nds32_expand_cbranch,
23616 nds32_expand_cstore, nds32_expand_movcc,
23617 nds32_output_cbranchsi4_equality_zero,
23618 nds32_output_cbranchsi4_equality_reg,
23619 nds32_output_cbranchsi4_equality_reg_or_const_int,
23620 nds32_output_cbranchsi4_greater_less_zero: New functions.
23621 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
23622 nds32_expand_cstore, nds32_expand_movcc,
23623 nds32_output_cbranchsi4_equality_zero,
23624 nds32_output_cbranchsi4_equality_reg,
23625 nds32_output_cbranchsi4_equality_reg_or_const_int,
23626 nds32_output_cbranchsi4_greater_less_zero): Declare.
23627 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
23628 nds32_rimm11s_operand): New predicates.
23629 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
23630 * config/nds32/nds32.md: Rewrite all the branch and conditional move
23633 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
23635 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
23636 * config/nds32/nds32.md: Ditto.
23637 * config/nds32/pipelines.md: Ditto.
23639 2018-04-04 Richard Biener <rguenther@suse.de>
23641 PR tree-optimization/85168
23642 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
23643 propagating abnormals.
23645 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
23647 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
23649 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
23650 Kito Cheng <kito.cheng@gmail.com>
23652 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
23653 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
23654 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
23655 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
23656 * config/nds32/nds32.md (sibcall_internal): New.
23657 (sibcall_register): Remove.
23658 (sibcall_immediate): Remove.
23659 (sibcall_value_internal): New.
23660 (sibcall_value_register): Remove.
23661 (sibcall_value_immediate): Remove.
23662 * config/nds32/predicates.md (nds32_general_register_operand): New.
23663 (nds32_call_address_operand): New.
23665 2018-04-03 Jakub Jelinek <jakub@redhat.com>
23667 PR rtl-optimization/85167
23668 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
23669 bb_defs if *split_p, instead preinitialize it to NULL.
23671 PR tree-optimization/85156
23672 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
23673 evaluating the argument multiple times.
23675 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
23677 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
23679 (_mm_cvtpd_ps): Likewise.
23680 (_mm_cvttpd_epi32): Likewise.
23681 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
23682 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
23683 vector, pixel, and bool following altivec.h include.
23685 2018-04-03 Martin Sebor <msebor@redhat.com>
23687 * doc/extend.texi (Common Function Attributes): Clarify.
23688 (const attribute): Likewise.
23689 (pure attribute): Likewise.
23691 2018-04-03 Jakub Jelinek <jakub@redhat.com>
23694 * config/i386/i386.c (ix86_expand_vector_set): Use
23695 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
23697 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
23699 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
23700 instructions when changing rounding bits to preserve precision bits
23701 in the x87 control word.
23703 2018-04-03 Martin Liska <mliska@suse.cz>
23705 PR tree-optimization/82491
23706 * rtl.h (strip_offset_and_add): Replace += suboffset with
23707 poly_uint64 () + suboffset.
23709 2018-03-29 Martin Liska <mliska@suse.cz>
23710 Martin Jambor <mjambor@suse.cz>
23713 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
23714 param_type is not an integral or pointer type.
23716 2018-04-03 Richard Biener <rguenther@suse.de>
23718 * sese.h (recompute_all_dominators): Remove.
23720 2018-04-02 Martin Sebor <msebor@redhat.com>
23722 * doc/invoke.texi (-Wrestrict): Fix typos.
23724 2018-04-02 Jim Wilson <jimw@sifive.com>
23726 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
23727 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
23728 (<optab>di3, <optab>si3_extend): Likewise.
23729 (<optab>si3_mask, <optab>si3_mask_1): New.
23730 (<optab>di3_mask, <optab>di3_mask_1): New.
23731 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
23732 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
23733 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
23735 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
23737 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
23740 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
23742 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
23743 (nds32_canonicalize_comparison): New function.
23745 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
23746 Kito Cheng <kito.cheng@gmail.com>
23747 Kuan-Lin Chen <kuanlinchentw@gmail.com>
23749 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
23750 * config/nds32/constants.md (unspec_volatile_element): Add
23751 UNSPEC_VOLATILE_RELAX_GROUP.
23752 * config/nds32/nds32-relax-opt.c: New file.
23753 * config/nds32/nds32-predicates.c
23754 (nds32_symbol_load_store_p): New function.
23755 * config/nds32/nds32-protos.h
23756 (nds32_symbol_load_store_p): Declare function.
23757 (make_pass_nds32_relax_opt): Declare new rtl pass function.
23758 * config/nds32/nds32.c
23759 (nds32_register_pass): New function to register pass.
23760 (nds32_register_passes): New function to register passes.
23761 * config/nds32/nds32.md (relax_group): New pattern.
23762 * config/nds32/nds32.opt (mrelax-hint): New option.
23763 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
23765 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
23767 * config/nds32/t-nds32: Modify files dependency.
23769 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
23771 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
23772 (PROFILE_HOOK): Define its implementation.
23774 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
23776 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
23777 type and 32-bit size.
23779 2018-04-01 Jakub Jelinek <jakub@redhat.com>
23781 PR middle-end/85090
23782 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
23783 (V_128_256): New mode iterator.
23784 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
23785 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
23786 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
23788 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
23789 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
23791 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
23794 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
23795 NaN inputs correctly.
23797 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
23800 * config/rs6000/vsx.md (??r): New mode attribute.
23801 (*vsx_mov<mode>_64bit): Use it.
23802 (*vsx_mov<mode>_32bit): Likewise.
23804 2018-03-30 Martin Sebor <msebor@redhat.com>
23806 PR tree-optimization/84818
23807 * builtins.c (check_access): Use warning_n.
23809 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23812 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
23814 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
23817 2018-03-30 Julia Koval <julia.koval@intel.com>
23820 * x86-tune.def (movx, partial_reg_dependency): Enable for
23823 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
23825 PR inline-asm/84985
23826 * lra-constraints.c (process_alt_operands): Move setting
23827 this_alternative_matches below.
23829 2018-03-29 Martin Liska <mliska@suse.cz>
23832 * doc/invoke.texi: Document how LTO works with debug info.
23833 Describe auto-load support of binutils. Mention 'x86-64'
23834 as valid option value of -march option.
23836 2018-03-29 Jakub Jelinek <jakub@redhat.com>
23838 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
23841 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
23842 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
23843 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
23846 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
23849 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
23850 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
23851 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
23852 for RS6000_BTM_POWERPC64.
23853 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
23854 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
23855 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
23858 (DIVDEU): Likewise.
23860 2018-03-28 Carl Love <cel@us.ibm.com>
23863 2017-09-27 Carl Love <cel@us.ibm.com>
23865 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
23866 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
23867 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
23870 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23872 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
23873 instead of __vector bool.
23874 (_mm_max_pu8): Likewise.
23875 (_mm_min_pi16): Likewise.
23877 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
23880 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
23881 (DIVWEUO): Likewise.
23882 (DIVDEO): Likewise.
23883 (DIVDEUO): Likewise.
23884 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
23885 DIVWEUO and DIVDEUO.
23886 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
23887 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
23888 (div_extend): Likewise.
23889 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
23891 (__builtin_divweuo): Likewise.
23892 (__builtin_divdeo): Likewise.
23893 (__builtin_divdeuo): Likewise.
23895 2018-03-28 Jakub Jelinek <jakub@redhat.com>
23898 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
23899 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
23901 PR tree-optimization/82004
23902 * gimple-match-head.c (optimize_pow_to_exp): New function.
23903 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
23904 Don't fold to exp if optimize_pow_to_exp is false.
23906 2018-03-28 Martin Liska <mliska@suse.cz>
23909 * calls.c (initialize_argument_information): Fix trailing space.
23910 * common.opt: Fix typo and provide better explanation for
23911 -fsanitize-coverage option.
23912 * config/i386/i386.opt: Fix typo.
23914 2018-03-28 Jakub Jelinek <jakub@redhat.com>
23915 Martin Liska <mliska@suse.cz>
23918 * gimplify.c (asan_poison_variable): Don't do the check for
23919 gimplify_omp_ctxp here.
23920 (gimplify_decl_expr): Do it here.
23921 (gimplify_target_expr): Likewise.
23923 2018-03-28 Martin Liska <mliska@suse.cz>
23926 * config/i386/i386.c (ix86_function_arg_advance): Do not call
23927 chkp_type_bounds_count if MPX is not enabled.
23929 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
23931 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
23933 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
23936 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
23937 function to create the function decl for complex long double
23938 multiply and divide for -mabi=ieeelongdouble.
23939 (init_float128_ieee): Call it.
23941 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
23944 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
23945 -fcf-protection=branch -mibt.
23946 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
23948 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23951 * config/arm/arm.c (arm_valid_symbolic_address): Handle
23952 arm_word_relocations.
23954 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
23957 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
23958 extern array declarations.
23960 2018-03-27 Richard Biener <rguenther@suse.de>
23962 PR middle-end/84067
23963 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
23964 explicit single_use checks.
23966 2018-03-27 Richard Biener <rguenther@suse.de>
23968 PR tree-optimization/85082
23969 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
23972 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23974 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
23975 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
23976 Turn on fasynchronous-unwind-tables and funwind-tables.
23978 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
23981 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
23982 (*bmi_blsr_<mode>_ccz): Ditto.
23984 2018-03-26 Tom de Vries <tom@codesourcery.com>
23986 PR tree-optimization/85063
23987 * omp-general.c (offloading_function_p): New function. Factor out
23989 * omp-offload.c (pass_omp_target_link::gate): ... here.
23990 * omp-general.h (offloading_function_p): Declare.
23991 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
23992 with attribute omp declare target for offloading functions.
23994 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
23996 PR tree-optimization/84005
23997 * tree-data-ref.h (get_base_for_alignment): Declare.
23998 * tree-data-ref.c (get_base_for_alignment_1): New function.
23999 (get_base_for_alignment): Likewise.
24000 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24001 get_base_for_alignment to find a suitable base object, instead
24002 of always using drb->base_address.
24004 2018-03-23 Jakub Jelinek <jakub@redhat.com>
24006 PR inline-asm/85022
24007 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
24008 known size by default.
24010 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
24012 PR inline-asm/85030
24013 * lra-constraints.c (process_alt_operands): Don't match BLKmode
24014 and non BLKmode operands.
24016 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24019 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
24020 Clean up attributes.
24022 2018-03-23 Richard Biener <rguenther@suse.de>
24025 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
24026 we are going to emit early debug for LTO.
24028 2018-03-23 Jakub Jelinek <jakub@redhat.com>
24030 PR inline-asm/85034
24031 * function.c (match_asm_constraints_1): Don't optimize if input
24032 doesn't satisfy general_operand predicate for output's mode.
24034 PR inline-asm/85022
24035 * alias.c (write_dependence_p): Don't require for x_canonicalized
24036 non-VOIDmode if x has VOIDmode.
24039 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
24040 just don't try to optimize it rather than assert it never happens.
24042 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
24044 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
24045 macro expansions for definition of ST_INTERNAL_<mode> and
24046 LD_INTERNAL_<mode> builtins.
24047 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
24049 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
24051 (altivec_expand_st_builtin): Likewise.
24052 (altivec_expand_builtin): Remove calls to deleted functions.
24053 (rs6000_address_for_altivec): Delete this function.
24054 * config/rs6000/vector.md: Remove expands for
24055 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
24057 2018-03-22 Sudakshina Das <sudi.das@arm.com>
24060 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
24061 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
24062 re-computing once computed.
24063 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
24064 (arm_init_machine_status): Initialize
24065 machine->static_chain_stack_bytes.
24067 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
24070 * doc/extend.texi: Add four new prototypes for vec_ld.
24071 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
24072 definitions for more logical presentation.
24073 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
24074 entries for V1TI variants of __builtin_altivec_ld builtin.
24075 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
24076 handling of V1TI variant of LVX icode pattern.
24077 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
24078 (rs6000_gimple_fold_builtin): Likewise.
24079 (altivec_init_builtins): Add code to define
24080 __builtin_altivec_lvx_v1ti function.
24082 2018-03-22 Jakub Jelinek <jakub@redhat.com>
24084 PR inline-asm/84941
24085 * function.c (match_asm_constraints_1): Don't do the optimization
24086 if input isn't a REG, SUBREG, MEM or constant.
24088 2018-03-22 Tom de Vries <tom@codesourcery.com>
24090 PR tree-optimization/84956
24091 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
24092 bb_has_abnormal_pred.
24094 2018-03-22 Jakub Jelinek <jakub@redhat.com>
24097 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
24098 DECL_INITIAL (decl) to decl at the end.
24099 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
24100 adjust the comment.
24102 2018-03-21 Joseph Myers <joseph@codesourcery.com>
24104 * doc/extend.texi (__builtin_tgmath): Document when complex
24105 integer types are treated as _Complex _Float64.
24107 2018-03-21 Tom de Vries <tom@codesourcery.com>
24109 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
24111 2018-03-21 Jakub Jelinek <jakub@redhat.com>
24113 PR tree-optimization/84960
24114 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
24115 if it is ENTRY block, move them into single succ of ENTRY in that case.
24117 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
24119 PR tree-optimization/84811
24120 * poly-int.h (poly_span_traits): Remove the T3 parameter and
24121 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
24122 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
24123 (known_subrange_p): Update accordingly. Cast each value involved
24124 in the size comparison, rather than casting the result of the
24127 2018-03-21 Jakub Jelinek <jakub@redhat.com>
24129 PR tree-optimization/84982
24130 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
24131 by flipping the least significant bit rather than all bits from
24132 bitpos to bitpos + bitsize - 1.
24134 2018-03-21 Nathan Sidwell <nathan@acm.org>
24136 * doc/extend.texi (Deprecated Features): Remove mention of
24137 long-deleted deprecations.
24139 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24142 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
24143 * configure: Regenerate.
24145 2018-03-21 Tom de Vries <tom@codesourcery.com>
24147 PR tree-optimization/83126
24148 * tree-parloops.c (num_phis): New function.
24149 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
24151 2018-03-21 Nathan Sidwell <nathan@acm.org>
24153 * doc/extend.texi (Deprecated Features): Update deprecated flags,
24154 mention anon-struct/union members and trailing attributes.
24156 2018-03-21 Bin Cheng <bin.cheng@arm.com>
24158 PR tree-optimization/84969
24159 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
24160 builtin memset partitions if they set different rhs values.
24162 2018-03-21 Jakub Jelinek <jakub@redhat.com>
24164 PR rtl-optimization/84989
24165 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
24166 VEC_DUPLICATE with scalar result mode.
24168 2018-03-21 Martin Liska <mliska@suse.cz>
24171 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
24172 not intended return statement.
24174 2018-03-21 Martin Liska <mliska@suse.cz>
24177 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
24178 (chkp_find_bound_slots_1): Limit number of iterations.
24180 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
24183 * Minor grammar fixes for x86 options.
24185 2018-03-20 Jakub Jelinek <jakub@redhat.com>
24188 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
24189 holding REG_CFA_RESTORE notes, instead turn them into a USE.
24191 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
24194 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
24195 (altivec_lvx_<mode>_1op): Likewise.
24196 (altivec_stvx_<mode>_2op): Likewise.
24197 (altivec_stvx_<mode>_1op): Likewise.
24198 (altivec_lvx_<VM2:mode>): New define_expand.
24199 (altivec_stvx_<VM2:mode>): Likewise.
24200 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
24201 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
24202 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
24203 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
24204 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
24205 (rs6000_gen_lvx): Likewise.
24206 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
24207 (altivec_expand_stv_builtin): Likewise.
24208 (altivec_expand_builtin): Likewise.
24209 * config/rs6000/vector.md: Likewise.
24211 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24214 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
24217 2018-03-20 Richard Biener <rguenther@suse.de>
24220 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
24221 sign-conversions as zero, fall back to standard scalar_stmt
24224 2018-03-20 Martin Liska <mliska@suse.cz>
24227 * predict.c (rebuild_frequencies): Handle case when we have
24228 PROFILE_ABSENT, but flag_guess_branch_prob is false.
24230 2018-03-20 Jakub Jelinek <jakub@redhat.com>
24233 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
24234 flag_section_anchors.
24235 * varasm.c (use_blocks_for_decl_p): Remove hack for
24236 dw2_force_const_mem.
24239 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
24241 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
24242 be created, use lowpart_subreg of operands[0] rather than operands[0]
24244 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
24245 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
24246 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
24247 and n constraint instead of aarch64_shift_imm_di and Usd.
24248 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
24249 (*aarch64_<optab>_reg_minus<mode>3): ... this.
24251 2018-03-20 Sudakshina Das <sudi.das@arm.com>
24254 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
24255 to favor GPR over NEON registers.
24256 (<shift>di3_neon): Likewise.
24258 2018-03-20 Tom de Vries <tom@codesourcery.com>
24261 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
24262 (nvptx_process_pars): Emit bar.sync asap and alap.
24264 2018-03-20 Tom de Vries <tom@codesourcery.com>
24267 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
24268 seen_label if seen_label is already set.
24270 2018-03-20 Jakub Jelinek <jakub@redhat.com>
24273 * config/i386/i386.c (fold_builtin_cpu): For features above 31
24274 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
24275 Use 1U instead of 1. Formatting fixes.
24278 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
24279 instead of TREE_TYPE (s1) for the return value.
24281 2018-03-19 Jakub Jelinek <jakub@redhat.com>
24283 PR tree-optimization/84946
24284 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
24285 bitsize + bitsize in poly_uint64 rather than poly_int64.
24288 * dwarf2asm.c: Include fold-const.c.
24289 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
24290 of decl rather than decl itself.
24292 PR rtl-optimization/84643
24293 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
24295 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
24298 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
24299 calling assemble_variable.
24301 2018-03-19 Sudakshina Das <sudi.das@arm.com>
24304 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
24305 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
24307 2018-03-19 Jim Wilson <jimw@sifive.com>
24310 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
24311 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
24312 (riscv_first_stack_step): Likewise.
24313 (riscv_option_override): Use STACK_BOUNDARY instead of
24314 MIN_STACK_BOUNDARY.
24315 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
24316 MIN_STACK_BOUNDARY.
24317 (BIGGEST_ALIGNMENT): Set to 128.
24318 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
24319 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
24322 2018-03-19 Richard Biener <rguenther@suse.de>
24324 PR tree-optimization/84933
24325 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
24326 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
24328 2018-03-19 Richard Biener <rguenther@suse.de>
24330 PR tree-optimization/84859
24331 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
24332 (cond_if_else_store_replacement): Perform sinking operation on
24333 single-store BBs regardless of MAX_STORES_TO_SINK setting.
24334 Generalize what a BB with a single eligible store is.
24336 2018-03-19 Richard Biener <rguenther@suse.de>
24338 PR tree-optimization/84929
24339 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
24340 chrec_is_positive against non-chrec arg.
24342 2018-03-19 Tamar Christina <tamar.christina@arm.com>
24345 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
24347 2018-03-18 Martin Liska <mliska@suse.cz>
24349 PR rtl-optimization/84635
24350 * regrename.c (build_def_use): Use matches_mode only when
24353 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
24355 PR tree-optimization/84913
24356 * tree-vect-loop.c (vectorizable_reduction): Don't try to
24357 vectorize chains of COND_EXPRs.
24359 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
24361 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
24363 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
24365 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
24367 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
24369 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
24371 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
24372 Kito Cheng <kito.cheng@gmail.com>
24374 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
24375 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
24376 (nds32_adjust_reg_alloc_order): New function.
24377 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
24379 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
24381 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
24382 nds32_print_operand, nds32_print_operand_address): Use
24383 HOST_WIDE_INT_PRINT_DEC instead.
24385 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
24387 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
24389 2018-03-17 Jakub Jelinek <jakub@redhat.com>
24392 * config/i386/i386.c (initial_ix86_tune_features,
24393 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
24394 unsigned long long.
24395 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
24396 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
24397 rather than 1u << ix86_tune. Formatting fix.
24398 (ix86_option_override_internal): Change ix86_arch_mask from
24399 unsigned int to unsigned HOST_WIDE_INT, initialize to
24400 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
24401 (ix86_function_specific_restore): Likewise.
24403 2018-03-16 Jakub Jelinek <jakub@redhat.com>
24406 * postreload.c (reload_combine_recognize_pattern): Perform
24407 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
24408 truncate_int_for_mode the result for the destination's mode.
24411 * hsa-gen.c (mem_type_for_type): Fix comment typo.
24412 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
24414 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
24417 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
24420 * lra-assigns.c (lra_split_hard_reg_for): Don't use
24421 regno_allocno_class_array and sorted_pseudos.
24422 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
24423 insns where regno is used.
24425 2018-03-16 Martin Liska <mliska@suse.cz>
24428 * multiple_target.c (create_dispatcher_calls): Redirect
24429 reference in the symbol table.
24431 2018-03-16 Martin Liska <mliska@suse.cz>
24434 * multiple_target.c (create_dispatcher_calls): Redirect also
24437 2018-03-16 Jakub Jelinek <jakub@redhat.com>
24441 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
24442 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
24443 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
24445 2018-03-16 Julia Koval <julia.koval@intel.com>
24447 * doc/invoke.texi (Skylake Server): Add CLWB.
24448 Cannonlake): Remove CLWB.
24450 2018-03-16 Jakub Jelinek <jakub@redhat.com>
24452 PR tree-optimization/84841
24453 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
24455 (FLOAT_ONE_CONST_TYPE): Define.
24456 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
24457 (sort_by_operand_rank): Put entries with higher constant_type last
24458 rather than first to match comments.
24460 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
24462 * config/nios2/nios2.md (movsi_internal): Fix thinko in
24465 2018-03-15 Jakub Jelinek <jakub@redhat.com>
24468 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
24469 check and use address of target always.
24471 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
24474 * config/i386/i386.c (indirect_thunk_needed): Update comments.
24475 (indirect_thunk_bnd_needed): Likewise.
24476 (indirect_thunks_used): Likewise.
24477 (indirect_thunks_bnd_used): Likewise.
24478 (indirect_return_needed): New.
24479 (indirect_return_bnd_needed): Likewise.
24480 (output_indirect_thunk_function): Add a bool argument for
24482 (output_indirect_thunk_function): Don't generate alias for
24483 function return thunk.
24484 (ix86_code_end): Call output_indirect_thunk_function to generate
24485 function return thunks.
24486 (ix86_output_function_return): Set indirect_return_bnd_needed
24487 and indirect_return_needed instead of indirect_thunk_bnd_needed
24488 and indirect_thunk_needed.
24490 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
24492 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
24493 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
24494 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
24496 2018-03-15 David Malcolm <dmalcolm@redhat.com>
24497 Paul Hua <paul.hua.gm@gmail.com>
24500 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
24502 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
24504 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
24505 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
24506 resp. SFmode cases.
24508 2018-03-15 Tamar Christina <tamar.christina@arm.com>
24511 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
24512 instead of GET_MODE_SIZE when comparing Units.
24514 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
24517 * varasm.c (hash_section): Return an unchangeble hash value
24518 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
24519 Return !aarch64_can_use_per_function_literal_pools_p ().
24521 2018-03-15 Jakub Jelinek <jakub@redhat.com>
24524 * optabs.c (emit_conditional_move): Pass address of cmode's copy
24525 rather than address of cmode as last argument to prepare_cmp_insn.
24527 2018-03-15 Julia Koval <julia.koval@intel.com>
24529 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
24530 F_AVX512VNNI, F_AVX512BITALG): New.
24532 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
24535 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
24536 insn for floating-point loads and stores.
24538 2018-03-14 Carl Love <cel@us.ibm.com>
24540 * config/rs6000/rs6000-c.c: Add macro definitions for
24541 ALTIVEC_BUILTIN_VEC_PERMXOR.
24542 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
24543 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
24544 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
24545 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
24547 * config/doc/extend.texi: Add prototypes for vec_permxor.
24549 2018-03-14 David Malcolm <dmalcolm@redhat.com>
24552 * diagnostic-show-locus.c (class layout_point): Convert m_line
24553 from int to linenum_type.
24554 (line_span::comparator): Use linenum "compare" function when
24555 comparing line numbers.
24556 (test_line_span): New function.
24557 (layout_range::contains_point): Convert param "row" from int to
24559 (layout_range::intersects_line_p): Likewise.
24560 (layout::will_show_line_p): Likewise.
24561 (layout::print_source_line): Likewise.
24562 (layout::should_print_annotation_line_p): Likewise.
24563 (layout::print_annotation_line): Likewise.
24564 (layout::print_leading_fixits): Likewise.
24565 (layout::annotation_line_showed_range_p): Likewise.
24566 (struct line_corrections): Likewise for field m_row.
24567 (line_corrections::line_corrections): Likewise for param "row".
24568 (layout::print_trailing_fixits): Likewise.
24569 (layout::get_state_at_point): Likewise.
24570 (layout::get_x_bound_for_row): Likewise.
24571 (layout::print_line): Likewise.
24572 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
24573 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
24574 * input.c (selftest::test_linenum_comparisons): New function.
24575 (selftest::input_c_tests): Call it.
24576 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
24577 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
24578 * selftest.h (ASSERT_GT): New macro.
24579 (ASSERT_GT_AT): New macro.
24580 (ASSERT_LT): New macro.
24581 (ASSERT_LT_AT): New macro.
24583 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
24585 PR rtl-optimization/84780
24586 * combine.c (distribute_links): Don't make a link based on pc_rtx.
24588 2018-03-14 Martin Liska <mliska@suse.cz>
24590 * tree.c (record_node_allocation_statistics): Use
24591 get_stats_node_kind.
24592 (get_stats_node_kind): New function extracted from
24593 record_node_allocation_statistics.
24594 (free_node): Use get_stats_node_kind.
24596 2018-03-14 Richard Biener <rguenther@suse.de>
24598 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
24599 that the value-set of ANTIC_IN doesn't grow.
24602 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
24604 (BB_VISITED_WITH_VISITED_SUCCS): New define.
24605 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
24607 2018-03-14 Julia Koval <julia.koval@intel.com>
24609 * config.gcc (icelake-client, icelake-server): New.
24611 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
24612 (initial_ix86_arch_features): Ditto.
24613 (PTA_SKYLAKE): Add SGX.
24614 (PTA_ICELAKE): Remove.
24615 (PTA_ICELAKE_CLIENT): New.
24616 (PTA_ICELAKE_SERVER): New.
24617 (ix86_option_override_internal): Split up icelake on icelake client and
24619 (get_builtin_code_for_version): Ditto.
24620 (fold_builtin_cpu): Ditto.
24621 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
24622 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
24623 * config/i386/i386.h (processor_type): Ditto.
24624 * doc/invoke.texi: Ditto.
24626 2018-03-14 Jakub Jelinek <jakub@redhat.com>
24629 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
24630 INTEGER_CST offset, add it together with bitpos / 8 and
24631 sign extend based on POINTER_SIZE.
24635 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
24638 * config/i386/constraints.md (Yc): New register constraint.
24639 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
24640 Use Yc constraint for alternative 2 of operand 0. Remove
24641 preferred_for_speed attribute.
24643 2018-03-14 Richard Biener <rguenther@suse.de>
24645 PR tree-optimization/84830
24646 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
24647 with the old one to avoid oscillations.
24649 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
24652 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
24654 (assign_by_spills): Return a flag of reload assignment failure.
24655 Do not process the reload assignment failures. Do not spill other
24656 reload pseudos if they has the same reg class. Update n if
24658 (lra_assign): Add a return arg. Set up from the result of
24659 assign_by_spills call.
24660 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
24661 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
24662 usage_insns if it is not NULL.
24663 (spill_hard_reg_in_range): New function.
24664 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
24665 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
24666 function prototypes.
24667 (lra_assign): Change prototype.
24668 * lra.c (lra): Add code to deal with fails by splitting hard reg
24671 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
24673 * config/riscv/riscv.opt (mrelax): New option.
24674 * config/riscv/riscv.c (riscv_file_start): Emit ".option
24675 "norelax" when riscv_mrelax is disabled.
24676 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
24678 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24681 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
24682 reassociation for int modes.
24684 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
24686 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
24687 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
24689 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
24690 * config/aarch64/aarch64-sve.md
24691 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
24692 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
24693 (*extend<mode><Vwide>2): Rename to...
24694 (aarch64_sve_extend<mode><Vwide>2): ...this.
24695 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
24696 renaming the old pattern to...
24697 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
24699 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
24700 define_expand, renaming the old pattern to...
24701 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
24702 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
24703 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
24704 account when deciding which SVE instruction the optab should use.
24705 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
24707 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
24709 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
24710 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
24711 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
24712 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
24713 (tlsdesc_small_<mode>): Turn a define_expand and use
24714 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
24715 (tlsdesc_small_advsimd_<mode>): ...this.
24716 (tlsdesc_small_sve_<mode>): New pattern.
24718 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
24720 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
24721 (UNSPEC_UMUL_HIGHPART): New constants.
24722 (MUL_HIGHPART): New int iteraor.
24723 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
24724 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
24726 (*<su>mul<mode>3_highpart): New define_insn.
24728 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
24731 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
24734 2018-03-13 Martin Liska <mliska@suse.cz>
24737 * (sem_item_optimizer::sem_item_optimizer): Initialize new
24739 (sem_item_optimizer::~sem_item_optimizer): Release it.
24740 (sem_item_optimizer::merge_classes): Register variable aliases.
24741 (sem_item_optimizer::fixup_pt_set): New function.
24742 (sem_item_optimizer::fixup_points_to_sets): Likewise.
24743 * ipa-icf.h: Declare new variables and functions.
24745 2018-03-13 Jakub Jelinek <jakub@redhat.com>
24747 PR middle-end/84834
24748 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
24749 integer_pow2p@2 and test integer_pow2p in condition.
24750 (A < 0 ? C : 0): Similarly for @1.
24752 PR middle-end/84831
24753 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
24754 characters starting at p contain '\0' character, don't look beyond
24758 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
24759 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
24762 * reg-stack.c (change_stack): Change update_end var from int to
24763 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
24764 also call set_block_for_insn on the newly added insns and rescan.
24767 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
24768 on the last operand.
24771 * tree.c (stabilize_reference_1): Return save_expr (e) for
24772 STATEMENT_LIST even if it doesn't have side-effects.
24774 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
24776 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
24778 2018-03-12 Renlin Li <renlin.li@arm.com>
24780 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
24781 aarch64_output_scalar_simd_mov_immediate.
24783 2018-03-12 Martin Sebor <msebor@redhat.com>
24785 PR tree-optimization/83456
24786 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
24787 for perfectly overlapping calls to memcpy.
24788 (gimple_fold_builtin_memory_chk): Same.
24789 (gimple_fold_builtin_strcpy): Handle no-warning.
24790 (gimple_fold_builtin_stxcpy_chk): Same.
24791 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
24793 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
24795 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
24796 parameter. Use it for SFmode.
24797 (rs6000_function_arg_advance_1): Adjust.
24798 (rs6000_function_arg): Adjust.
24799 (rs6000_gimplify_va_arg): Pass false for that new parameter.
24801 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
24803 PR rtl-optimization/84169
24804 PR rtl-optimization/84780
24805 * combine.c (can_combine_p): Check for a 2-insn combination whether
24806 the destination register is used between the two insns, too.
24808 2018-03-12 Richard Biener <rguenther@suse.de>
24810 PR tree-optimization/84803
24811 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
24812 for refs DR analysis didn't process.
24814 2018-03-12 Richard Biener <rguenther@suse.de>
24816 PR tree-optimization/84777
24817 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
24818 force-vectorize loops ignore whether we are optimizing for size.
24820 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
24822 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
24823 (TARGET_MD_ASM_ADJUST): Define.
24825 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
24826 Kito Cheng <kito.cheng@gmail.com>
24827 Chung-Ju Wu <jasonwucj@gmail.com>
24829 * config/nds32/nds32.c (nds32_compute_stack_frame,
24830 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
24831 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
24832 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
24833 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
24834 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
24835 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
24836 * config/nds32/nds32.md (prologue, epilogue): Use macro
24837 NDS32_V3PUSH_AVAILABLE_P to do checking.
24839 2018-03-11 Jakub Jelinek <jakub@redhat.com>
24842 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
24843 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
24844 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
24845 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
24846 addition of most attributes on !orig_type_die or the attribute not
24847 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
24849 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
24850 Chung-Ju Wu <jasonwucj@gmail.com>
24852 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
24853 __NDS32_VH__ macro.
24854 * config/nds32/nds32.opt (mvh): New option.
24856 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
24857 Chung-Ju Wu <jasonwucj@gmail.com>
24859 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
24861 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
24862 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
24865 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
24866 Chung-Ju Wu <jasonwucj@gmail.com>
24868 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
24870 * config/nds32/nds32-multiple.md (strlensi): New pattern.
24871 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
24873 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
24874 Kito Cheng <kito.cheng@gmail.com>
24875 Chung-Ju Wu <jasonwucj@gmail.com>
24877 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
24878 UNSPEC_FFMISM and UNSPEC_FLMISM.
24879 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
24880 for ffb, ffmism and flmism.
24881 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
24882 (unspec_ffmism): Ditto.
24883 (unspec_flmism): Ditto.
24884 (nds32_expand_builtin_impl): Check if string extension is available.
24885 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
24886 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
24888 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
24891 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
24894 * lra-assigns.c (assign_by_spills): Return a flag of reload
24895 assignment failure. Do not process the reload assignment
24896 failures. Do not spill other reload pseudos if they has the same
24898 (lra_assign): Add a return arg. Set up from the result of
24899 assign_by_spills call.
24900 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
24901 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
24902 usage_insns if it is not NULL.
24903 (spill_hard_reg_in_range): New function.
24904 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
24905 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
24906 function prototypes.
24907 (lra_assign): Change prototype.
24908 * lra.c (lra): Add code to deal with fails by splitting hard reg
24911 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
24914 * config/i386/i386.opt: Replace Enforcment with Enforcement.
24916 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
24919 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
24920 (dw_val_node): Add val_symbolic_view.
24921 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
24922 (symview_upper_bound): New.
24923 (new_line_info_table): Initialize symviews_since_reset.
24924 (dwarf2out_source_line): Count symviews_since_reset and set
24925 symview_upper_bound.
24926 (dw_val_equal_p): Handle symview.
24927 (add_AT_symview): New.
24928 (print_dw_val): Handle symview.
24929 (attr_checksum, attr_checksum_ordered): Likewise.
24930 (same_dw_val_p, size_of_die): Likewise.
24931 (value_format, output_die): Likewise.
24932 (add_high_low_attributes): Use add_AT_symview for entry_view.
24933 (dwarf2out_finish): Reset symview_upper_bound, clear
24936 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
24939 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
24940 Add strict argument and use it.
24941 (rs6000_split_multireg_move): Update for new strict argument.
24942 (mem_operand_gpr): Disallow all non-offsettable addresses.
24943 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
24945 2018-03-09 Jakub Jelinek <jakub@redhat.com>
24948 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
24949 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
24950 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
24953 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
24954 decl, use remap_type if we want to use the type.
24956 2018-03-09 Martin Sebor <msebor@redhat.com>
24958 PR tree-optimization/84526
24959 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
24961 (builtin_access::generic_overlap): Be prepared to handle non-array
24964 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
24966 PR rtl-optimization/84682
24967 * lra-constraints.c (process_address_1): Check is_address flag
24968 for address constraints.
24969 (process_alt_operands): Likewise.
24970 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
24971 preprocess_constraints.
24972 * recog.h (preprocess_constraints): Add oploc parameter.
24974 * recog.c (preprocess_constraints): Test address_operand for
24975 CT_ADDRESS constraints.
24977 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
24980 * lra-assigns.c (assign_by_spills): Return a flag of reload
24981 assignment failure. Do not process the reload assignment
24982 failures. Do not spill other reload pseudos if they has the same
24984 (lra_assign): Add a return arg. Set up from the result of
24985 assign_by_spills call.
24986 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
24987 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
24988 usage_insns if it is not NULL.
24989 (spill_hard_reg_in_range): New function.
24990 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
24991 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
24992 function prototypes.
24993 (lra_assign): Change prototype.
24994 * lra.c (lra): Add code to deal with fails by splitting hard reg
24997 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25000 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
25001 Accept complain bool parameter. Only emit errors if it is true.
25002 (arm_parse_cpu_option_name): Likewise.
25003 (arm_target_thumb_only): Adjust callers of the above.
25004 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
25005 prototype to take a default true bool parameter.
25006 (arm_parse_arch_option_name): Likewise.
25008 2018-03-09 David Malcolm <dmalcolm@redhat.com>
25009 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
25013 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
25014 * configure: Regenerate.
25015 * configure.ac ("linker --version-script option"): New.
25016 ("linker soname option"): New.
25018 2018-03-09 Richard Biener <rguenther@suse.de>
25020 PR tree-optimization/84775
25021 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
25022 immediate uses of predicate stmts and mark them modified.
25025 PR tree-optimization/84178
25026 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
25028 (version_loop_for_if_conversion): Delay update_ssa call.
25029 (tree_if_conversion): Delay update_ssa until after predicate
25032 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
25035 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
25036 when the function accesses prior frames.
25038 2018-03-08 Jakub Jelinek <jakub@redhat.com>
25041 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
25042 gen_llsym, otherwise call maybe_gen_llsym.
25044 PR inline-asm/84742
25045 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
25046 has ',' character inside of it.
25048 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25051 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
25052 as clobbering CC_REGNUM.
25054 2018-03-08 Richard Biener <rguenther@suse.de>
25056 PR middle-end/84552
25057 * tree-scalar-evolution.c: Include tree-into-ssa.h.
25058 (follow_copies_to_constant): Do not follow SSA names registered
25061 2018-03-08 Richard Biener <rguenther@suse.de>
25063 PR tree-optimization/84178
25064 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
25066 (version_loop_for_if_conversion): Delay update_ssa call.
25067 (tree_if_conversion): Delay update_ssa until after predicate
25070 2018-03-08 David Malcolm <dmalcolm@redhat.com>
25072 PR tree-optimization/84178
25073 * tree-if-conv.c (release_bb_predicate): Remove the
25074 the assertion that the stmts have NULL use_ops.
25075 Discard the statements, asserting that they haven't
25076 yet been added to a BB.
25078 2018-03-08 Richard Biener <rguenther@suse.de>
25080 PR tree-optimization/84746
25081 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
25082 (phi_translate): Pass in destination ANTIC_OUT set.
25083 (phi_translate_1): Likewise. For a simplified result lookup
25084 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
25085 (phi_translate_set): Adjust.
25086 (do_pre_regular_insertion): Likewise.
25087 (do_pre_partial_partial_insertion): Likewise.
25089 2018-03-08 Martin Liska <mliska@suse.cz>
25091 PR gcov-profile/84735
25092 * doc/gcov.texi: Document usage of profile files.
25093 * gcov-io.h: Document changes in the format.
25095 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
25099 * dwarf2out.c (struct dw_line_info_table): Update comments for
25101 (FORCE_RESET_NEXT_VIEW): New.
25102 (FORCE_RESETTING_VIEW_P): New.
25103 (RESETTING_VIEW_P): Check for -1 too.
25104 (ZERO_VIEW_P): Likewise.
25105 (new_line_info_table): Force-reset next view.
25106 (dwarf2out_begin_function): Likewise.
25107 (dwarf2out_source_line): Simplify zero_view_p initialization.
25108 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
25109 view directly. Omit view when omitting .loc at line 0.
25111 2018-03-08 Jakub Jelinek <jakub@redhat.com>
25113 PR tree-optimization/84740
25114 * tree-switch-conversion.c (process_switch): Call build_constructors
25115 only if info.phi_count is non-zero.
25117 PR tree-optimization/84739
25118 * tree-tailcall.c (find_tail_calls): Check call arguments against
25119 DECL_ARGUMENTS (current_function_decl) rather than
25120 DECL_ARGUMENTS (func) when checking for tail recursion.
25122 2018-03-07 Jakub Jelinek <jakub@redhat.com>
25124 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
25125 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
25126 Volker Reichelt's entry and add entries for people that perform
25127 GCC fuzzy testing and report numerous bugs.
25129 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
25132 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
25133 readonly data in sdata, if that is disabled.
25134 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
25135 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
25136 -mreadonly-in-sdata option.
25138 2018-03-07 Martin Sebor <msebor@redhat.com>
25140 PR tree-optimization/84468
25141 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
25142 basic block when looking for nul assignment.
25144 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
25147 * except.h (output_function_exception_table): Adjust prototype.
25148 * except.c (output_function_exception_table): Remove FNNAME parameter
25149 and add SECTION parameter. Ouput one part of the table at a time.
25150 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
25151 the first part of the exception table and emit unwind directives.
25152 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
25153 (i386_pe_seh_cold_init): Likewise.
25154 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
25155 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
25156 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
25157 (ix86_output_call_insn): Emit a nop in one more case for SEH.
25158 * config/i386/winnt.c: Include except.h.
25159 (struct seh_frame_state): Add reg_offset, after_prologue and
25160 in_cold_section fields.
25161 (i386_pe_seh_end_prologue): Set seh->after_prologue.
25162 (i386_pe_seh_cold_init): New function.
25163 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
25164 to seh->in_cold_section.
25165 (seh_emit_push): Record the offset of the push.
25166 (seh_emit_save): Record the offet of the save.
25167 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
25168 Test seh->after_prologue to disregard the epilogue.
25169 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
25170 (i386_pe_end_cold_function): New function.
25172 2018-03-07 Jakub Jelinek <jakub@redhat.com>
25175 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
25176 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
25179 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
25181 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
25182 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
25184 PR middle-end/84723
25185 * multiple_target.c: Include tree-inline.h and intl.h.
25186 (expand_target_clones): Diagnose and fail if node->definition and
25187 !tree_versionable_function_p (node->decl).
25189 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
25191 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
25193 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
25194 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
25195 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
25197 2018-03-06 Jakub Jelinek <jakub@redhat.com>
25200 * combine.c (try_combine): Use reg_or_subregno instead of handling
25201 just paradoxical SUBREGs and REGs.
25203 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
25205 * config/arc/arc.c (arc_finalize_pic): Remove function.
25206 (arc_must_save_register): We use single base PIC register, remove
25207 checks to save/restore the PIC register.
25208 (arc_expand_prologue): Likewise.
25209 * config/arc/arc-protos.h (arc_set_default_type_attributes):
25211 (arc_verify_short): Likewise.
25212 (arc_attr_type): Likewise.
25213 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
25214 (walk_stores): Likewise.
25215 (arc_address_cost): Make it static.
25216 (arc_verify_short): Likewise.
25217 (branch_dest): Likewise.
25218 (arc_attr_type): Likewise.
25219 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
25220 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
25221 (arc_final_prescan_insn): Remove inserting the nops due to
25222 hardware hazards. It is done in reorg step.
25223 (insn_length_variant_t): Remove.
25224 (insn_length_parameters_t): Likewise.
25225 (arc_insn_length_parameters): Likewise.
25226 (arc_get_insn_variants): Likewise.
25227 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
25229 2018-03-06 Jakub Jelinek <jakub@redhat.com>
25231 PR inline-asm/84683
25232 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
25235 PR tree-optimization/84687
25236 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
25238 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
25240 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25242 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
25243 Rename to ppc_speculation_barrier.
25244 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
25245 __builtin_ppc_speculation_barrier.
25247 2018-03-05 Jakub Jelinek <jakub@redhat.com>
25250 * combine.c (combine_simplify_rtx): Don't try to simplify if
25251 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
25254 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
25256 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
25257 to 32 bytes when compiling for POWER9.
25259 2018-03-05 Jakub Jelinek <jakub@redhat.com>
25262 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
25263 regparm >= 3 with no arg reg available also for calls with
25264 flag_force_indirect_call. Pass decl to ix86_function_regparm.
25267 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
25269 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
25271 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
25274 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
25276 2018-03-05 Richard Biener <rguenther@suse.de>
25278 PR tree-optimization/84486
25279 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
25280 When inserting a __builtin_assume_aligned call set the LHS
25281 SSA name alignment info accordingly.
25283 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
25285 PR tree-optimization/84114
25286 * config/aarch64/aarch64.c (aarch64_reassociation_width)
25287 Avoid reassociation of FLOAT_MODE addition.
25289 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
25291 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
25292 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
25293 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
25294 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
25295 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
25296 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
25297 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
25299 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
25300 __builtin_ia32_wbinvd): New builtins.
25301 (SPECIAL_ARGS2): New.
25302 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
25303 (SPECIAL_ARGS2): New.
25304 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
25305 (ix86_valid_target_attribute_inner_p): Ditto.
25306 (ix86_init_mmx_sse_builtins): Add special_args2.
25307 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
25308 TARGET_WBNOINVD_P): New.
25309 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
25310 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
25311 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
25312 * config/i386/immintrin.h (_wbinvd): New intrinsic.
25313 * config/i386/pconfigintrin.h: New file.
25314 * config/i386/wbnoinvdintrin.h: Ditto.
25315 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
25317 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
25319 2018-03-05 Richard Biener <rguenther@suse.de>
25321 PR tree-optimization/84670
25322 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
25324 (BB_VISITED_WITH_VISITED_SUCCS): New define.
25325 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
25326 (compute_antic_aux): Only assert the number of values in ANTIC_IN
25327 doesn't grow if all successors (recursively) were visited at least
25330 2018-03-05 Richard Biener <rguenther@suse.de>
25332 PR tree-optimization/84650
25333 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
25334 if executed in the loop pipeline.
25336 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
25338 * doc/configfiles.texi (Configuration Files): Move info about
25339 conditionalizing $target-protos.h to...
25340 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
25341 differs from $target-protos.h.
25343 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
25344 Chung-Ju Wu <jasonwucj@gmail.com>
25346 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
25347 * config/nds32/nds32-multiple.md (setmemsi): Define.
25348 * config/nds32/nds32-memory-manipulation.c
25349 (nds32_gen_dup_4_byte_to_word_value): New.
25350 (emit_setmem_word_loop): New.
25351 (emit_setmem_byte_loop): New.
25352 (nds32_expand_setmem_loop): New.
25353 (nds32_expand_setmem_loop_v3m): New.
25354 (nds32_expand_setmem_unroll): New.
25355 (nds32_expand_setmem): New.
25357 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
25358 Chung-Ju Wu <jasonwucj@gmail.com>
25360 * config/nds32/nds32-memory-manipulation.c
25361 (nds32_emit_load_store): New.
25362 (nds32_emit_post_inc_load_store): New.
25363 (nds32_emit_mem_move): New.
25364 (nds32_emit_mem_move_block): New.
25365 (nds32_expand_movmemsi_loop_unknown_size): New.
25366 (nds32_expand_movmemsi_loop_known_size): New.
25367 (nds32_expand_movmemsi_loop): New.
25368 (nds32_expand_movmemsi_unroll): New.
25369 (nds32_expand_movmemqi): Rename ...
25370 (nds32_expand_movmemsi): ... to this.
25371 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
25372 (movmemsi): ... to this.
25373 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
25374 (nds32_expand_movmemsi): ... to this.
25376 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
25377 Monk Chiang <sh.chiang04@gmail.com>
25378 Chung-Ju Wu <jasonwucj@gmail.com>
25380 * config/nds32/nds32-protos.h
25381 (nds32_expand_load_multiple): New arguments.
25382 (nds32_expand_store_multiple): Ditto.
25383 (nds32_valid_multiple_load_store): Rename ...
25384 (nds32_valid_multiple_load_store_p): ... to this.
25385 * config/nds32/nds32-memory-manipulation.c
25386 (nds32_expand_load_multiple): Refine implementation.
25387 (nds32_expand_store_multiple): Ditto.
25388 * config/nds32/nds32-multiple.md
25389 (load_multiple): Update nds32_expand_load_multiple interface.
25390 (store_multiple): Update nds32_expand_store_multiple interface.
25391 * config/nds32/nds32-predicates.c
25392 (nds32_valid_multiple_load_store): Rename ...
25393 (nds32_valid_multiple_load_store_p): ... to this and refine
25395 * config/nds32/predicates.md
25396 (nds32_load_multiple_and_update_address_operation): New predicate.
25397 (nds32_store_multiple_and_update_address_operation): New predicate.
25399 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
25400 Chung-Ju Wu <jasonwucj@gmail.com>
25402 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
25403 (combo): New attribute.
25404 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
25406 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
25408 * config/nds32/nds32.opt: Change -mcmodel= default value.
25410 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
25411 Monk Chiang <sh.chiang04@gmail.com>
25412 Chung-Ju Wu <jasonwucj@gmail.com>
25414 * config/nds32/constants.md (unspec_element): New enum.
25415 * config/nds32/constraints.md (Umw): New constraint.
25416 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
25417 * config/nds32/nds32-intrinsic.md: Likewise.
25418 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
25419 (nds32_valid_smw_lwm_base_p): New.
25420 (nds32_output_smw_single_word): New.
25421 (nds32_output_lmw_single_word): New.
25422 (nds32_expand_unaligned_load): New.
25423 (nds32_expand_unaligned_store): New.
25424 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
25425 (nds32_output_smw_single_word): Declare.
25426 (nds32_output_lmw_single_word): Declare.
25427 (nds32_expand_unaligned_load): Declare.
25428 (nds32_expand_unaligned_store): Declare.
25429 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
25430 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
25431 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
25432 NDS32_BUILTIN_UASTORE_DW.
25433 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
25436 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
25437 Kito Cheng <kito.cheng@gmail.com>
25438 Chung-Ju Wu <jasonwucj@gmail.com>
25440 * config/nds32/nds32-intrinsic.c
25441 (nds32_expand_builtin_null_ftype_reg): Delete.
25442 (nds32_expand_builtin_reg_ftype_imm): Ditto.
25443 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
25444 (nds32_read_argument): New.
25445 (nds32_legitimize_target): Ditto.
25446 (nds32_legitimize_argument): Ditto.
25447 (nds32_check_constant_argument): Ditto.
25448 (nds32_expand_unop_builtin): Ditto.
25449 (nds32_expand_unopimm_builtin): Ditto.
25450 (nds32_expand_binop_builtin): Ditto.
25451 (nds32_builtin_decl_impl): Ditto.
25452 (builtin_description): Ditto.
25453 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
25454 (nds32_init_builtins_impl): Ditto.
25455 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
25456 (nds32_builtin_decl): New.
25457 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
25458 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
25460 2018-03-02 Jeff Law <law@redhat.com>
25462 * reorg.c (stop_search_p): Handle DEBUG_INSN.
25463 (redundant_insn, fill_simple_delay_slots): Likewise.
25464 (fill_slots_from_thread): Likewise.
25465 * resource.c (mark_referenced_resources): Likewise.
25466 (mark_set_resources, find_dead_or_set_registers): Likewise.
25468 2018-03-02 Jakub Jelinek <jakub@redhat.com>
25470 * substring-locations.h (format_warning_va): Formatting fix for
25471 ATTRIBUTE_GCC_DIAG.
25472 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
25474 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
25475 * substring-locations.c: Include intl.h.
25476 (format_warning_va): Turned into small wrapper around
25477 format_warning_n_va, renamed to ...
25478 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
25479 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
25481 (format_warning_at_substring_n): New function.
25482 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
25483 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
25484 format_warning_at_substring with just a shorter name instead of
25485 const function pointer.
25486 (fmtwarn_n): New function.
25487 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
25488 appropriate, get rid of all the fmtstr temporaries, move conditionals
25489 with G_() wrapped string literals directly into fmtwarn arguments,
25490 cast dir.len to (int), formatting fixes.
25492 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
25494 * doc/invoke.texi: Remove "Cilk Plus" references.
25496 2018-03-02 Jakub Jelinek <jakub@redhat.com>
25497 Richard Biener <rguenther@suse.de>
25500 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
25501 for error or warning attributes if CALL_FROM_THUNK_P is set.
25504 2018-03-02 Jakub Jelinek <jakub@redhat.com>
25507 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
25508 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
25511 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
25512 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
25514 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
25515 instead of -1U in last predictors element's probability member.
25517 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
25520 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
25521 arguments if they are comparable.
25523 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
25525 PR tree-optimization/84634
25526 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
25527 masks and masked_loop_p with a single loop_masks, making sure it's
25528 null for bb vectorization.
25530 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
25532 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
25533 (vect_analyze_data_ref_access): Use loop->safe_len rather than
25534 loop->force_vectorize to check whether there is no alias.
25536 2018-03-02 Jakub Jelinek <jakub@redhat.com>
25539 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
25541 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
25543 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
25544 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
25545 instead of a loop around prev_real_insn.
25546 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
25549 PR inline-asm/84625
25550 * config/i386/i386.c (ix86_print_operand): Use conditional
25551 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
25554 2018-03-02 Richard Biener <rguenther@suse.de>
25556 PR tree-optimization/84427
25557 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
25558 (bitmap_set_subtract_values): Rewrite to handle multiple
25561 (prune_clobbered_mems): Likewise.
25562 (phi_translate): Take edge instead of pred/phiblock.
25563 (phi_translate_1): Likewise.
25564 (phi_translate_set): Likewise. Insert all translated
25565 exprs for a value into the set, keeping possibly multiple
25566 expressions per value.
25567 (compute_antic_aux): Adjust for phi_translate changes.
25568 When intersecting union the expressions and prune those
25569 not in the final value set, keeping possibly multiple
25570 expressions per value. Do not use value-insertion
25571 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
25572 all expressions. Add verification that the value-sets
25573 only shrink during iteration.
25574 (compute_partial_antic_aux): Adjust for the phi_translate changes.
25575 (do_pre_regular_insertion): Likewise.
25576 (do_pre_partial_partial_insertion): Likewise.
25578 2018-03-02 Richard Biener <rguenther@suse.de>
25581 * config/darwin.c (saved_debug_info_level): New static global.
25582 (darwin_asm_lto_start): Disable debug info generation for LTO out.
25583 (darwin_asm_lto_end): Restore debug info generation settings.
25585 2018-03-01 Martin Liska <mliska@suse.cz>
25588 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
25589 volatile arguments.
25591 2018-03-01 Richard Biener <rguenther@suse.de>
25594 * dwarf2out.c (gen_variable_die): Properly handle late VLA
25595 type annotation with LTO when debug was disabled at compile-time.
25597 2018-03-01 Matthew Fortune <mfortune@gmail.com>
25599 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
25601 (mips_final_postscan_insn): Likewise.
25603 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
25605 PR rtl-optimization/84528
25606 * alias.c (init_alias_target): Add commentary.
25607 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
25608 a unique base value if the frame pointer is not eliminated
25609 to the stack pointer.
25611 2018-03-01 Tom de Vries <tom@codesourcery.com>
25613 PR rtl-optimization/83327
25614 * lra-int.h (hard_regs_spilled_into): Declare.
25615 * lra.c (hard_regs_spilled_into): Define.
25616 (init_reg_info): Init hard_regs_spilled_into.
25617 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
25618 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
25619 (process_bb_lives): Handle hard_regs_spilled_into.
25620 (lra_create_live_ranges_1): Before doing liveness propagation, clear
25621 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
25623 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
25625 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
25626 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
25627 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
25628 * config/rs6000/aix72.h: New file.
25630 2018-02-28 Jakub Jelinek <jakub@redhat.com>
25632 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
25633 instead of warning_at with conditional singular and plural messages
25637 * stor-layout.c (update_alignment_for_field): For
25638 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
25639 && !DECL_PACKED (field), do the alignment update, just use
25640 only desired_align instead of MAX (type_align, desired_align)
25642 (place_field): Don't do known_align < desired_align handling
25643 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
25644 is non-NULL, instead do it after rli->prev_field handling and
25645 only if not within a bitfield word. For DECL_PACKED (field)
25646 use type_align of BITS_PER_UNIT.
25648 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
25650 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
25651 superfluous parentheses and trailing spaces.
25653 2018-02-28 Richard Biener <rguenther@suse.de>
25655 PR tree-optimization/84584
25656 * graphite-scop-detection.c (scop_detection::add_scop): Discard
25657 SCoPs with fake exit edge.
25659 2018-02-28 Martin Liska <mliska@suse.cz>
25662 * timevar.c (timer::print): Fix format to properly print 100%
25665 2018-02-28 Richard Biener <rguenther@suse.de>
25667 PR middle-end/84607
25668 * genmatch.c (capture_info::walk_match): Do not mark
25669 captured expressions without operands as expr_p given
25670 they act more like predicates and should be subject to
25671 "lost tail" side-effect preserving.
25673 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
25675 PR rtl-optimization/81611
25676 * auto-inc-dec.c (attempt_change): Move dead note from
25677 mem_insn if it's the next use of regno
25678 (find_address): Take address use of reg holding
25679 non-incremented value. Add parm to limit search to the named
25681 (merge_in_block): Attempt to use a mem insn that is the next
25682 use of the original regno.
25684 2018-02-27 Martin Sebor <msebor@redhat.com>
25687 * doc/invoke.texi (-Wmissing-attributes): New option.
25688 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
25690 2018-02-27 Martin Sebor <msebor@redhat.com>
25692 PR translation/84207
25693 * diagnostic-core.h (warning_n, error_n, inform_n): Change
25694 n argument to unsigned HOST_WIDE_INT.
25695 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
25696 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
25697 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
25698 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
25700 2018-02-27 Richard Biener <rguenther@suse.de>
25702 PR tree-optimization/84512
25703 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
25704 Do not use the estimate returned from record_stmt_cost for
25705 the scalar iteration cost but sum properly using add_stmt_cost.
25707 2018-02-27 Richard Biener <rguenther@suse.de>
25709 PR tree-optimization/84466
25710 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
25711 Adjust last change to less strictly validate use operands.
25713 2018-02-27 Martin Liska <mliska@suse.cz>
25715 PR gcov-profile/84548
25716 * gcov.c (process_file): Allow partial overlap and consider it
25717 also as group functions.
25718 (output_lines): Properly calculate range of lines for a group.
25720 2018-02-27 Martin Liska <mliska@suse.cz>
25722 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
25723 'ggc' suffixes. Change first column width.
25724 (timer::print): Fix formatting of the column.
25726 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
25728 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
25729 preserve inline entry blocks for the sake of debug inline
25730 entry point markers alone.
25731 (remove_unused_locals): Suggest in comments a better place to
25732 force the preservation of inline entry blocks that are
25733 otherwise unused, but do not preserve them.
25735 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
25737 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
25739 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
25742 * config/i386/constraints.md (Bs): Replace
25743 ix86_indirect_branch_register with
25744 TARGET_INDIRECT_BRANCH_REGISTER.
25746 * config/i386/i386.md (indirect_jump): Likewise.
25747 (tablejump): Likewise.
25748 (*sibcall_memory): Likewise.
25749 (*sibcall_value_memory): Likewise.
25750 Peepholes of indirect call and jump via memory: Likewise.
25751 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
25752 (*sibcall_value_GOT_32): Likewise.
25753 * config/i386/predicates.md (indirect_branch_operand): Likewise.
25754 (GOT_memory_operand): Likewise.
25755 (call_insn_operand): Likewise.
25756 (sibcall_insn_operand): Likewise.
25757 (GOT32_symbol_operand): Likewise.
25758 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
25760 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
25762 PR rtl-optimization/83496
25763 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
25764 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
25765 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
25766 redundant insn, if any.
25767 (relax_delay_slots): Likewise.
25768 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
25770 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
25772 PR tree-optimization/83965
25773 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
25774 that grouped statements are part of a reduction chain. Return
25775 true if the statement is not marked as a reduction itself but
25776 is part of a group.
25777 (vect_recog_dot_prod_pattern): Don't check whether the statement
25778 is part of a group here.
25779 (vect_recog_sad_pattern): Likewise.
25780 (vect_recog_widen_sum_pattern): Likewise.
25782 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
25785 * final.c (rest_of_clean_state): Also look for calls inside sequences.
25787 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
25790 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
25792 (ix86_output_indirect_function_return): New prototype.
25793 (ix86_split_simple_return_pop_internal): Likewise.
25794 * config/i386/i386.c (indirect_return_via_cx): New.
25795 (indirect_return_via_cx_bnd): Likewise.
25796 (indirect_thunk_name): Handle return va CX_REG.
25797 (output_indirect_thunk_function): Create alias for
25798 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
25799 (ix86_output_indirect_jmp): Remove the bool argument.
25800 (ix86_output_indirect_function_return): New function.
25801 (ix86_split_simple_return_pop_internal): Likewise.
25802 * config/i386/i386.md (*indirect_jump): Don't pass false
25803 to ix86_output_indirect_jmp.
25804 (*tablejump_1): Likewise.
25805 (simple_return_pop_internal): Change it to define_insn_and_split.
25806 Call ix86_split_simple_return_pop_internal to split it for
25807 -mfunction-return=.
25808 (simple_return_indirect_internal): Call
25809 ix86_output_indirect_function_return instead of
25810 ix86_output_indirect_jmp.
25812 2018-02-26 Jakub Jelinek <jakub@redhat.com>
25815 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
25816 memset and value initialization afterwards.
25818 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
25820 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
25822 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25825 * common/config/aarch64/aarch64-common.c
25826 (aarch_option_optimization_table[]): Switch
25827 off fomit-frame-pointer
25829 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
25830 Chung-Ju Wu <jasonwucj@gmail.com>
25832 * config/nds32/nds32-multiple.md (load_multiple): Disallow
25834 (store_multiple): Ditto.
25836 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
25838 * config.gcc: Add --with-cpu support for nds32 target.
25839 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
25840 * config/nds32/nds32.opt: Add -mcpu= option.
25842 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
25844 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
25845 isel=yes): Warn for these deprecated options.
25847 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
25849 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
25850 ISA_2_5_MASKS_EMBEDDED.
25852 2018-02-23 Jakub Jelinek <jakub@redhat.com>
25854 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
25855 p->max as pointers rather than using iterative_hash_expr.
25857 2018-02-23 Carl Love <cel@us.ibm.com>
25859 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
25860 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
25862 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
25863 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
25864 P8V_BUILTIN_VEC_VUNSIGNED2.
25866 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
25869 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
25870 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
25872 * lra-constraints.c (curr_insn_transform): Set up
25873 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
25875 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
25876 * lra-eliminations.c (spill_pseudos): Ditto.
25877 (process_insn_for_elimination): Ditto.
25878 * lra-lives.c (reg_early_clobber_p): Use the new macros.
25879 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
25880 LRA_NON_CLOBBERED_ALT.
25882 2018-02-22 Martin Sebor <msebor@redhat.com>
25884 PR tree-optimization/84480
25885 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
25886 to maybe_diag_stxncpy_trunc. Call it.
25887 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
25888 from gimple_fold_builtin_strcpy. Print inlining stack.
25889 (handle_builtin_stxncpy): Print inlining stack.
25890 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
25892 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
25895 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
25896 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
25897 and -fcheck-pointer-bounds are used together.
25898 (indirect_thunk_prefix): New enum.
25899 (indirect_thunk_need_prefix): New function.
25900 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
25901 "_nt" instead of "_bnd" for NOTRACK prefix.
25902 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
25903 (output_indirect_thunk_function): Likewise.
25905 (ix86_code_end): Update output_indirect_thunk_function calls.
25906 (ix86_output_indirect_branch_via_reg): Replace
25907 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
25908 (ix86_output_indirect_branch_via_push): Likewise.
25909 (ix86_output_function_return): Likewise.
25910 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
25911 incompatible with -fcf-protection=branch and
25912 -fcheck-pointer-bounds.
25914 2018-02-22 Steve Ellcey <sellcey@cavium.com>
25917 * config/aarch64/aarch64.c (aarch64_print_address_internal):
25918 Change gcc_assert call to output_operand_lossage.
25920 2018-02-22 Steve Ellcey <sellcey@cavium.com>
25922 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
25924 2018-02-22 DJ Delorie <dj@redhat.com>
25925 Sebastian Perta <sebastian.perta@renesas.com>
25926 Oleg Endo <olegendo@gcc.gnu.org>
25928 * config/rx/rx.c (rx_rtx_costs): New function.
25929 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
25931 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
25933 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
25935 2018-02-22 Martin Liska <mliska@suse.cz>
25938 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
25939 Add "native" as a possible value.
25941 2018-02-22 Martin Liska <mliska@suse.cz>
25944 * config/i386/i386.c (ix86_option_override_internal):
25945 Add "native" as a possible value for -march and -mtune.
25947 2018-02-22 Jakub Jelinek <jakub@redhat.com>
25950 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
25951 to all type variants.
25953 PR tree-optimization/84503
25954 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
25955 width as info->bitpos + info->bitsize - start.
25956 (merged_store_group::merge_overlapping): Simplify width computation.
25957 (check_no_overlap): New function.
25958 (imm_store_chain_info::try_coalesce_bswap): Compute expected
25959 start + width and last_order of the group, fail if check_no_overlap
25961 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
25962 to group if check_no_overlap fails.
25964 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
25966 * config/rs6000/altivec.md: Delete contraint arguments to
25967 define_expand, define_split, and define_peephole2, and in
25968 define_insn_and_split if always unused.
25969 * config/rs6000/darwin.md: Ditto.
25970 * config/rs6000/dfp.md: Ditto.
25971 * config/rs6000/rs6000.md: Ditto.
25972 * config/rs6000/sync.md: Ditto.
25973 * config/rs6000/vector.md: Ditto.
25974 * config/rs6000/vsx.md: Ditto.
25976 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
25978 * config/rs6000/altivec.md: Write output control strings as braced
25979 blocks instead of double-quoted strings.
25980 * config/rs6000/darwin.md: Ditto.
25981 * config/rs6000/rs6000.md: Ditto.
25982 * config/rs6000/vector.md: Ditto.
25983 * config/rs6000/vsx.md: Ditto.
25985 2018-02-21 Jason Merrill <jason@redhat.com>
25987 PR c++/84314 - ICE with templates and fastcall attribute.
25988 * attribs.c (build_type_attribute_qual_variant): Remove assert.
25990 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
25992 * ipa-cp.c (determine_versionability): Fix comment typos.
25994 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
25997 * ipa-cp.c (determine_versionability): Do not version functions caling
26000 2018-02-21 Martin Liska <mliska@suse.cz>
26003 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
26004 Add "native" as a possible value.
26005 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
26006 the macro when native cpu detection is available.
26008 2018-02-21 Martin Liska <mliska@suse.cz>
26011 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
26012 Add "native" as a possible value.
26013 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
26014 when native cpu detection is available.
26016 2018-02-21 Jakub Jelinek <jakub@redhat.com>
26017 Martin Sebor <msebor@redhat.com>
26019 PR tree-optimization/84478
26020 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
26022 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
26023 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
26024 support which is conservatively correct, for 2 only stay conservative
26025 for maxlen. Formatting and comment capitalization fixes. Add STRICT
26026 argument to the 2 argument get_range_strlen, adjust 6 arg
26027 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
26029 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
26030 (gimple_fold_builtin_strlen): Pass true as last argument to
26033 2018-02-20 Martin Sebor <msebor@redhat.com>
26035 PR middle-end/84095
26036 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
26037 (builtin_memref::set_base_and_offset): Same. Handle inner references.
26038 (builtin_memref::builtin_memref): Factor out parts into
26039 set_base_and_offset and call it.
26041 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
26043 PR middle-end/84406
26044 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
26045 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
26046 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
26047 search at the associated MODE_INT.
26049 2018-02-20 Jeff Law <law@redhat.com>
26051 PR middle-end/82123
26052 PR tree-optimization/81592
26053 PR middle-end/79257
26054 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
26055 for range data rather than using global data.
26056 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
26057 range data rather than using global data.
26058 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
26059 pass it to children as needed.
26060 (struct directive::fmtresult): Similarly.
26061 (struct directive::set_width): Similarly.
26062 (struct directive::set_precision): Similarly.
26063 (format_integer, format_directive, parse_directive): Similarly.
26064 (format_none): Accept unnamed vr_values parameter.
26065 (format_percent, format_floating, format_character): Similarly.
26066 (format_string, format_plain): Similarly.
26067 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
26068 the EVRP range analyzer for range data rather than using global data.
26069 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
26070 gimple-ssa-evrp-analyze.h
26071 (class sprintf_dom_walker): Add after_dom_children member function.
26072 Add evrp_range_analyzer member.
26073 (sprintf_dom_walker::before_dom_children): Call into the EVRP
26074 range analyzer as needed.
26075 (sprintf_dom_walker::after_dom_children): New member function.
26076 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
26078 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
26079 (evrp_range_analyzer::pop_to_marker): Likewise.
26081 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
26083 PR tree-optimization/84419
26084 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
26085 with the required type if its current type is compatible but
26088 2018-02-20 Jakub Jelinek <jakub@redhat.com>
26090 PR middle-end/82004
26091 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
26092 after vectorization.
26094 2018-02-20 Martin Liska <mliska@suse.cz>
26097 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
26098 possible values if we don't have a hint.
26100 2018-02-20 Martin Liska <mliska@suse.cz>
26104 * final.c (shorten_branches): Build align_tab array with one
26106 * opts.c (finish_options): Add alignment option limit check.
26107 (MAX_CODE_ALIGN): Likewise.
26108 (MAX_CODE_ALIGN_VALUE): Likewise.
26109 * doc/invoke.texi: Document maximum allowed option value for
26110 all -falign-* options.
26112 2018-02-19 Jakub Jelinek <jakub@redhat.com>
26115 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
26116 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
26117 * var-tracking.c (emit_note_insn_var_location): Remove all references
26118 to NOTE_INSN_CALL_ARG_LOCATION.
26119 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
26120 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
26121 Use copy_rtx_if_shared.
26122 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
26123 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
26124 (dwarf2out_var_location): Remove handling of
26125 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
26127 * final.c (final_scan_insn): Remove all references to
26128 NOTE_INSN_CALL_ARG_LOCATION.
26129 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
26130 before dumping final insns.
26131 * except.c (emit_note_eh_region_end): Remove all references to
26132 NOTE_INSN_CALL_ARG_LOCATION.
26133 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
26134 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
26135 * config/arc/arc.c (hwloop_optimize): Likewise.
26136 * config/arm/arm.c (create_fix_barrier): Likewise.
26137 * config/s390/s390.c (s390_chunkify_start): Likewise.
26138 * config/sh/sh.c (find_barrier): Likewise.
26139 * config/i386/i386.c (rest_of_insert_endbranch,
26140 ix86_seh_fixup_eh_fallthru): Likewise.
26141 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
26142 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
26143 * config/frv/frv.c (frv_function_prologue): Likewise.
26144 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
26146 (note_outside_basic_block_p): Remove all references to
26147 NOTE_INSN_CALL_ARG_LOCATION.
26148 * gengtype.c (adjust_field_rtx_def): Likewise.
26149 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
26151 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
26152 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
26155 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
26158 PR tree-optimization/84452
26159 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
26160 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
26163 2018-02-19 Martin Liska <mliska@suse.cz>
26166 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
26168 2018-02-19 Martin Liska <mliska@suse.cz>
26169 Richard Sandiford <richard.sandiford@linaro.org>
26171 PR tree-optimization/82491
26172 * gimple-fold.c (get_base_constructor): Make earlier bail out
26175 2018-02-19 Carl Love <cel@us.ibm.com>
26177 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
26178 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
26180 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
26181 P8V_BUILTIN_VEC_NEG.
26183 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
26185 * config/rl78/rl78.md (movdf): New define expand.
26187 2018-02-19 Martin Liska <mliska@suse.cz>
26190 * doc/invoke.texi: Fix typo.
26191 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
26193 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
26195 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
26196 handle rs6000_single_float and rs6000_double_float specially for
26199 2018-02-16 Jeff Law <law@redhat.com>
26201 * config/rx/rx.c (add_pop_cfi_notes): New function.;
26202 (pop_regs): Use it.
26204 2018-02-16 Jakub Jelinek <jakub@redhat.com>
26207 * ipa-inline.c (inline_small_functions): Fix a typo.
26209 2018-02-16 Nathan Sidwell <nathan@acm.org>
26211 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
26213 2018-02-16 Carl Love <cel@us.ibm.com>
26215 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
26216 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
26217 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
26218 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
26219 expansion to P8V_BUILTIN_VEC_FLOAT2.
26221 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
26223 PR rtl-optimization/70023
26224 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
26225 src_regno into account.
26227 2018-02-16 Carl Love <cel@us.ibm.com>
26229 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
26230 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
26231 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
26232 * config/rs6000/rs6000.c: Remove case statements for
26233 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
26234 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
26235 and P9V_BUILTIN_VEC_VINSERT4B.
26236 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
26237 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
26238 * config/rs6000/vsx.md:
26239 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
26242 2018-02-16 Carl Love <cel@us.ibm.com>
26244 * config/rs6000/altivec.h: Add builtin names vec_extract4b
26246 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
26248 * config/rs6000/rs6000-c.c: Add the definitions for
26249 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
26250 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
26251 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
26252 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
26253 definition for insert4b and define insn *insert3b_internal.
26254 * doc/extend.texi: Add documentation for vec_extract4b.
26256 2018-02-16 Nathan Sidwell <nathan@acm.org>
26258 * doc/extend.texi (Backwards Compatibility): Mention friend
26259 injection. Note for-scope is deprecated.
26260 * doc/invoke.texi (-ffriend-injection): Deprecate.
26262 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
26264 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
26265 that moved to I2, also allow destinations that are a paradoxical
26266 subreg (instead of a normal reg).
26268 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
26271 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
26274 2018-02-16 Richard Biener <rguenther@suse.de>
26276 PR tree-optimization/84037
26277 PR tree-optimization/84016
26279 * config/i386/i386.c (ix86_builtin_vectorization_cost):
26280 Adjust vec_construct for the fact we need additional higher latency
26281 128bit inserts for AVX256 and AVX512 vector builds.
26282 (ix86_add_stmt_cost): Scale vector construction cost for
26285 2018-02-16 Richard Biener <rguenther@suse.de>
26287 PR tree-optimization/84417
26288 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
26289 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
26290 (non_rewritable_lvalue_p): Likewise, use poly-ints.
26292 2018-02-16 Martin Liska <mliska@suse.cz>
26295 * internal-fn.def (ASAN_CHECK): Set proper flags.
26296 (ASAN_MARK): Likewise.
26298 2018-02-16 Julia Koval <julia.koval@intel.com>
26300 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
26301 from PTA_CANNONLAKE.
26303 2018-02-16 Jakub Jelinek <jakub@redhat.com>
26306 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
26307 Use ++iter rather than iter++ for std::list iterators.
26308 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
26309 defer deleting them until all nodes in the forest are processed. Do
26310 free even leaf nodes. Change to_process into auto_vec.
26313 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
26314 * vec.h (vec_default_construct): Use memset instead of placement new
26315 if BROKEN_VALUE_INITIALIZATION is defined.
26316 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
26317 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
26320 PR rtl-optimization/83723
26321 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
26322 * lra.c (lra_substitute_pseudo): Likewise. If true, use
26323 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
26325 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
26327 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
26329 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
26331 PR rtl-optimization/81443
26332 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
26333 from inner REGs to paradoxical SUBREGs.
26335 2018-02-16 Richard Biener <rguenther@suse.de>
26337 PR tree-optimization/84399
26338 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
26339 For operands we can analyze at their definition make sure we can
26340 analyze them at each use as well.
26342 2018-02-16 Richard Biener <rguenther@suse.de>
26344 PR tree-optimization/84190
26345 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
26346 volatile accesses if the decl isn't volatile.
26348 2018-02-15 Jason Merrill <jason@redhat.com>
26350 PR c++/84314 - ICE with templates and fastcall attribute.
26351 * attribs.c (build_type_attribute_qual_variant): Don't clobber
26352 TYPE_CANONICAL on an existing type.
26354 2018-02-15 Jakub Jelinek <jakub@redhat.com>
26356 PR tree-optimization/84383
26357 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
26358 dstoff nor call operand_equal_p if dstbase is NULL.
26360 PR tree-optimization/84334
26361 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
26362 also a CONSTANT_CLASS_P, punt.
26364 2018-02-14 Jim Wilson <jimw@sifive.com>
26366 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
26367 first SMALL_OPERAND check. New local min_second_step. Move assert
26368 to where locals are set. Add TARGET_RVC support.
26369 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
26371 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
26373 * doc/invoke.texi: Correct -Wformat-overflow code sample.
26375 2018-02-14 Martin Sebor <msebor@redhat.com>
26377 PR tree-optimization/83698
26378 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
26379 arrays constrain the offset range to their bounds.
26380 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
26381 (builtin_access::overlap): Avoid setting the size of overlap if it's
26383 (maybe_diag_overlap): Also consider arrays when deciding what values
26384 of offsets to include in diagnostics.
26386 2018-02-14 Martin Sebor <msebor@redhat.com>
26389 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
26390 that correspond to the kind of a declaration.
26392 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
26395 * config/pa/pa.md: Load address of PIC label using the linkage table
26396 if the label is nonlocal.
26398 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
26400 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
26401 warning message if user requests -maltivec=be.
26402 * doc/invoke.texi: Document deprecation of -maltivec=be.
26404 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
26407 * config/rs6000/rs6000-c.c: Update definitions for
26408 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
26409 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
26411 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
26414 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
26415 add _get_ssp intrinsics. Remove argument from
26416 __builtin_ia32_rdssp[d|q].
26417 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
26418 * config/i386/i386-builtin.def: Remove argument from
26419 __builtin_ia32_rdssp[d|q].
26420 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
26421 ix86_expand_special_args_builtin for _rdssp[d|q].
26422 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
26423 Clear register before usage.
26424 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
26425 Add documentation for new _get_ssp and _inc_ssp intrinsics.
26427 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
26429 PR tree-optimization/84357
26430 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
26431 operand 1 of an ARRAY_REF too.
26433 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
26436 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
26437 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
26439 (set_of_reg): New struct.
26440 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
26441 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
26442 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
26444 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
26445 Split into bitclr, bitset, bitinvert patterns if appropriate.
26446 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
26447 use rx_fuse_in_memory_bitop.
26448 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
26449 to named insn, correct maximum insn length.
26451 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
26454 * machmode.def: Define a complex mode for PARTIAL_INT.
26455 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
26457 * doc/rtl.texi: Document CSPImode.
26458 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
26460 (msp430_hard_regno_nregs_with_padding): Likewise.
26462 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
26465 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
26467 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
26469 PR rtl-optimization/84169
26470 * combine.c (try_combine): New variable split_i2i3. Set it to true if
26471 we generated a parallel as new i3 and we split that to new i2 and i3
26472 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
26473 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
26474 those to i2, not i1. Partially rewrite this scan code.
26476 2018-02-13 Jakub Jelinek <jakub@redhat.com>
26479 * stor-layout.c (place_field): For variable length fields, adjust
26480 offset_align afterwards not just based on the field's alignment,
26481 but also on the size.
26483 PR middle-end/84309
26484 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
26485 of exps and logs in the use_exp2 case.
26487 2018-02-13 Jeff Law <law@redhat.com>
26489 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
26490 entry for "vector".
26492 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
26495 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
26499 * common.opt (gas-loc-support, gas-locview-support): New.
26500 (ginline-points, ginternal-reset-location-views): New.
26501 * doc/invoke.texi: Document them. Use @itemx where intended.
26502 (gvariable-location-views): Adjust.
26503 * target.def (reset_location_view): New.
26504 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
26505 (TARGET_RESET_LOCATION_VIEW): New.
26506 * doc/tm.texi: Rebuilt.
26507 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
26508 (dwarf2out_default_as_locview_support): New.
26509 (output_asm_line_debug_info): Use option variables.
26510 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
26511 (output_loc_list): Likewise.
26512 (add_high_low_attributes): Check option variables.
26513 Don't output entry view attribute in strict mode.
26514 (gen_inlined_subroutine_die): Check option variables.
26515 (dwarf2out_inline_entry): Likewise.
26516 (init_sections_and_labels): Likewise.
26517 (dwarf2out_early_finish): Likewise.
26518 (maybe_reset_location_view): New, from...
26519 (dwarf2out_var_location): ... here. Call it.
26520 * debug.h (dwarf2out_default_as_loc_support): Declare.
26521 (dwarf2out_default_as_locview_support): Declare.
26522 * hooks.c (hook_int_rtx_insn_0): New.
26523 * hooks.h (hook_int_rtx_insn_0): Declare.
26524 * toplev.c (process_options): Take -gas-loc-support and
26525 -gas-locview-support from dwarf2out. Enable
26526 -gvariable-location-views by default only with locview
26527 assembler support. Enable -ginternal-reset-location-views by
26528 default only if the target defines the corresponding hook.
26529 Enable -ginline-points by default if location views are
26530 enabled; force it disabled if statement frontiers are
26532 * tree-inline.c (expand_call_inline): Check option variables.
26533 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
26535 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
26537 PR tree-optimization/84321
26538 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
26539 handling. Also check whether the anti-range contains any values
26540 that satisfy the mask; switch to a VR_RANGE if not.
26542 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
26545 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
26547 2018-02-13 Martin Jambor <mjambor@suse.cz>
26550 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
26551 of call statements, also set location of a load to a temporary.
26553 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
26555 * config/rl78/rl78.c (add_vector_labels): New function.
26556 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
26557 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
26558 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
26559 which checks that no arguments are passed.
26560 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
26561 * doc/extend.texi: Documentation for the new attribute.
26563 2018-02-13 Andreas Schwab <schwab@suse.de>
26565 * config/riscv/linux.h (CPP_SPEC): Define.
26567 2018-02-13 Jakub Jelinek <jakub@redhat.com>
26570 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
26571 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
26572 OPTION_MASK_ISA_AES as first argument to def_builtin_const
26573 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
26574 instead of OPTION_MASK_ISA_PCLMUL as first argument to
26575 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
26576 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
26577 temporarily for AES and PCLMUL builtins.
26579 PR tree-optimization/84339
26580 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
26581 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
26584 PR middle-end/84309
26585 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
26586 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
26587 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
26589 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
26591 * omp-simd-clone.h: New file.
26592 * omp-simd-clone.c: Include omp-simd-clone.h.
26593 (expand_simd_clones): No longer static.
26594 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
26595 cgraph.h and omp-simd-clone.h.
26596 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
26597 (vect_recog_widen_shift_pattern): Formatting fix.
26598 (vect_pattern_recog_1): Don't check optab for calls.
26601 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
26602 operands[2] into a REG before using gen_lowpart on it.
26604 2018-02-12 Jeff Law <law@redhat.com>
26607 * config/sh/sh.c (find_barrier): Consider a sibling call
26610 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
26611 successfully back substituting a reg.
26613 2018-02-12 Richard Biener <rguenther@suse.de>
26615 PR tree-optimization/84037
26616 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
26617 parameter, move visited init to caller.
26618 (vect_slp_analyze_operations): Separate cost from validity
26619 check, initialize visited once for all instances.
26620 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
26622 * tree-vect-stmts.c (vect_model_simple_cost): Make early
26624 (vect_model_promotion_demotion_cost): Likewise.
26625 (vectorizable_bswap): Guard cost modeling with !slp_node
26626 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
26628 (vectorizable_call): Likewise.
26629 (vectorizable_conversion): Likewise.
26630 (vectorizable_assignment): Likewise.
26631 (vectorizable_shift): Likewise.
26632 (vectorizable_operation): Likewise.
26633 (vectorizable_store): Likewise.
26634 (vectorizable_load): Likewise.
26635 (vectorizable_condition): Likewise.
26636 (vectorizable_comparison): Likewise.
26638 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
26641 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
26642 (ASAN_MARK): Fix fnspec to account for return value, change pointer
26643 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
26645 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
26647 PR middle-end/83665
26648 * params.def (inline-min-speedup): Increase from 8 to 15.
26649 (max-inline-insns-auto): Decrease from 40 to 30.
26650 * ipa-split.c (consider_split): Add some buffer for function to
26651 be considered inlining candidate.
26652 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
26655 2018-02-12 Richard Biener <rguenther@suse.de>
26657 PR tree-optimization/84037
26658 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
26659 matched stmts if we cannot swap the non-matched ones.
26661 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
26663 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
26664 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
26665 _mm_maskz_scalef_round_ss): New intrinsics.
26666 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
26667 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
26668 __builtin_ia32_scalefss_round): Remove.
26669 (__builtin_ia32_scalefsd_mask_round,
26670 __builtin_ia32_scalefss_mask_round): New intrinsics.
26671 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
26672 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
26673 ((match_operand:VF_128 2 "<round_nimm_predicate>"
26674 "<round_constraint>")): Changed to ...
26675 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
26676 "<round_scalar_constraint>")): ... this.
26677 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
26678 %0, %1, %2<round_op3>}"): Changed to ...
26679 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
26680 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
26681 %2<round_scalar_mask_op3>}"): ... this.
26682 * config/i386/subst.md (round_scalar_nimm_predicate): New.
26684 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
26686 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
26687 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
26688 (_mm_maskz_sqrt_round_ss): New intrinsics.
26689 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
26690 (__builtin_ia32_sqrtsd_mask_round)
26691 (__builtin_ia32_sqrtss_mask_round): New builtins.
26692 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
26693 (__builtin_ia32_sqrtss_round): Remove.
26694 (__builtin_ia32_sqrtsd_mask_round)
26695 (__builtin_ia32_sqrtss_mask_round): New builtins.
26696 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
26697 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
26698 ((match_operand:VF_128 1 "vector_operand"
26699 "xBm,<round_constraint>")): Changed to ...
26700 ((match_operand:VF_128 1 "vector_operand"
26701 "xBm,<round_scalar_constraint>")): ... this.
26702 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
26703 %0, %2, %<iptr>1<round_op3>}): Changed to ...
26704 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
26705 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
26706 %<iptr>1<round_scalar_mask_op3>}): ... this.
26707 ((set_attr "prefix" "<round_prefix>")): Changed to ...
26708 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
26710 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
26713 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
26714 Cast vec_cmpeq result to correct type.
26715 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
26716 Cast vec_cmpgt result to correct type.
26718 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
26720 * final.c (final_scan_insn_1): Renamed from...
26721 (final_scan_insn): ... this. New wrapper, to recover
26722 seen from the outermost call in recursive ones.
26723 * config/sparc/sparc.c (output_return): Drop seen from call.
26724 (output_sibcall): Likewise.
26725 * config/visium/visium.c (output_branch): Likewise.
26727 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
26729 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
26732 2018-02-10 Alan Modra <amodra@gmail.com>
26735 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
26736 Specify LR as an input.
26738 2018-02-10 Jakub Jelinek <jakub@redhat.com>
26741 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
26742 remove_member_access_dummy_vars): New functions.
26743 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
26744 lower_omp_1, execute_lower_omp): Use them.
26746 PR rtl-optimization/84308
26747 * shrink-wrap.c (spread_components): Release todo vector.
26749 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
26751 PR rtl-optimization/57193
26752 * ira-color.c (struct allocno_color_data): Add member
26753 conflict_allocno_hard_prefs.
26754 (update_conflict_allocno_hard_prefs): New.
26755 (bucket_allocno_compare_func): Add a preference based on
26756 conflict_allocno_hard_prefs.
26757 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
26758 (color_allocnos): Remove a dead code. Initiate
26759 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
26761 2018-02-09 Jakub Jelinek <jakub@redhat.com>
26764 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
26765 constraint from =wa to wa. Avoid a subreg on the output operand,
26766 instead use a pseudo and subreg it in a move.
26767 (p9_xxbrd_<mode>): Changed to ...
26768 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
26769 (p9_xxbrd_v2df): New expander.
26770 (p9_xxbrw_<mode>): Changed to ...
26771 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
26772 (p9_xxbrw_v4sf): New expander.
26774 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
26776 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
26778 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
26781 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
26782 multiply in 32-bit mode.
26783 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
26784 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
26787 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
26789 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
26790 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
26791 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
26792 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
26794 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
26797 * dwarf2out.c (is_trivial_indirect_ref): New function.
26798 (dwarf2out_late_global_decl): Do not generate a location
26799 attribute for variables that have a non-trivial DECL_VALUE_EXPR
26800 and that are not defined in the current unit.
26802 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
26804 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
26805 instead of a libcall for UNORDERED.
26807 2018-02-09 Tamar Christina <tamar.christina@arm.com>
26810 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
26811 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
26813 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26816 * config/s390/s390.c (s390_set_current_function): Invoke
26817 s390_indirect_branch_settings also if fndecl didn't change.
26819 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
26821 * config/rs6000/rs6000.md (blockage): Set length to zero.
26823 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
26825 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
26827 2018-02-09 Jakub Jelinek <jakub@redhat.com>
26830 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
26831 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
26835 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
26836 PARALLEL incoming that failed vt_get_decl_and_offset check.
26838 PR middle-end/84237
26839 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
26840 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
26842 (get_variable_section): For decls in named .bss* sections pass true as
26843 second argument to bss_initializer_p.
26845 2018-02-09 Marek Polacek <polacek@redhat.com>
26846 Jakub Jelinek <jakub@redhat.com>
26849 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
26850 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
26851 Sync some changes from cxx_fold_indirect_ref.
26853 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
26855 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
26857 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
26858 (BLOCK_INLINE_ENTRY_LABEL): New.
26859 (dwarf2out_var_location): Disregard inline entry markers.
26860 (inline_entry_data): New struct.
26861 (inline_entry_data_hasher): New hashtable type.
26862 (inline_entry_data_hasher::hash): New.
26863 (inline_entry_data_hasher::equal): New.
26864 (inline_entry_data_table): New variable.
26865 (add_high_low_attributes): Add DW_AT_entry_pc and
26866 DW_AT_GNU_entry_view attributes if a pending entry is found
26867 in inline_entry_data_table. Add old entry_pc attribute only
26868 if debug nonbinding markers are disabled.
26869 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
26870 markers are enabled.
26871 (block_within_block_p, dwarf2out_inline_entry): New.
26872 (dwarf2out_finish): Check that no entries remained in
26873 inline_entry_data_table.
26874 * final.c (reemit_insn_block_notes): Handle inline entry notes.
26875 (final_scan_insn, notice_source_line): Likewise.
26876 (rest_of_clean_state): Skip inline entry markers.
26877 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
26879 * gimple.c (gimple_build_debug_inline_entry): New.
26880 * gimple.h (enum gimple_debug_subcode): Add
26881 GIMPLE_DEBUG_INLINE_ENTRY.
26882 (gimple_build_debug_inline_entry): Declare.
26883 (gimple_debug_inline_entry_p): New.
26884 (gimple_debug_nonbind_marker_p): Adjust.
26885 * insn-notes.def (INLINE_ENTRY): New.
26886 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
26887 inline entry marker notes.
26888 (print_insn): Likewise.
26889 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
26890 (INSN_DEBUG_MARKER_KIND): Likewise.
26891 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
26892 * tree-inline.c (expand_call_inline): Build and insert
26893 debug_inline_entry stmt.
26894 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
26895 inline entry blocks early, if nonbind markers are enabled.
26896 (dump_scope_block): Dump fragment info.
26897 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
26898 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
26899 (gimple_build_debug_inline_entry): New.
26900 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
26901 Enable/disable inline entry points too.
26902 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
26903 (DEBUG_INSN): Describe inline entry markers.
26905 * common.opt (gvariable-location-views): New.
26906 (gvariable-location-views=incompat5): New.
26907 * config.in: Rebuilt.
26908 * configure: Rebuilt.
26909 * configure.ac: Test assembler for view support.
26910 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
26911 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
26912 * dwarf2out.c (var_loc_view): New typedef.
26913 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
26914 (dwarf2out_locviews_in_attribute): New.
26915 (dwarf2out_locviews_in_loclist): New.
26916 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
26917 (enum dw_line_info_opcode): Add LI_adv_address.
26918 (struct dw_line_info_table): Add view.
26919 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
26920 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
26921 (zero_view_p): New variable.
26922 (ZERO_VIEW_P): New macro.
26923 (output_asm_line_debug_info): New.
26924 (struct var_loc_node): Add view.
26925 (add_AT_view_list, AT_loc_list): New.
26926 (add_var_loc_to_decl): Add view param. Test it against last.
26927 (new_loc_list): Add view params. Record them.
26928 (AT_loc_list_ptr): Handle loc and view lists.
26929 (view_list_to_loc_list_val_node): New.
26930 (print_dw_val): Handle dw_val_class_view_list.
26931 (size_of_die): Likewise.
26932 (value_format): Likewise.
26933 (loc_list_has_views): New.
26934 (gen_llsym): Set vl_symbol too.
26935 (maybe_gen_llsym, skip_loc_list_entry): New.
26936 (dwarf2out_maybe_output_loclist_view_pair): New.
26937 (output_loc_list): Output view list or entries too.
26938 (output_view_list_offset): New.
26939 (output_die): Handle dw_val_class_view_list.
26940 (output_dwarf_version): New.
26941 (output_compilation_unit_header): Use it.
26942 (output_skeleton_debug_sections): Likewise.
26943 (output_rnglists, output_line_info): Likewise.
26944 (output_pubnames, output_aranges): Update version comments.
26945 (output_one_line_info_table): Output view numbers in asm comments.
26946 (dw_loc_list): Determine current endview, pass it to new_loc_list.
26947 Call maybe_gen_llsym.
26948 (loc_list_from_tree_1): Adjust.
26949 (add_AT_location_description): Create view list attribute if
26950 needed, check it's absent otherwise.
26951 (convert_cfa_to_fb_loc_list): Adjust.
26952 (maybe_emit_file): Call output_asm_line_debug_info for test.
26953 (dwarf2out_var_location): Reset views as needed. Precompute
26954 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
26955 attribute. Set view.
26956 (new_line_info_table): Reset next view.
26957 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
26958 (dwarf2out_source_line): Likewise. Output view resets and labels to
26959 the assembler, or select appropriate line info opcodes.
26960 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
26961 (optimize_string_length): Catch it. Adjust.
26962 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
26963 dw_val_class_view_list, and remove it if no longer needed.
26964 (hash_loc_list): Hash view numbers.
26965 (loc_list_hasher::equal): Compare them.
26966 (optimize_location_lists): Check whether a view list symbol is
26967 needed, and whether the locview attribute is present, and
26968 whether they match. Remove the locview attribute if no longer
26970 (index_location_lists): Call skip_loc_list_entry for test.
26971 (dwarf2out_finish): Call output_asm_line_debug_info for test.
26972 Use output_dwarf_version.
26973 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
26974 (struct dw_val_node): Add val_view_list.
26975 * final.c (SEEN_NEXT_VIEW): New.
26976 (set_next_view_needed): New.
26977 (clear_next_view_needed): New.
26978 (maybe_output_next_view): New.
26979 (final_start_function): Rename to...
26980 (final_start_function_1): ... this. Take pointer to FIRST,
26981 add SEEN parameter. Emit param bindings in the initial view.
26982 (final_start_function): Reintroduce SEEN-less interface.
26983 (final): Rename to...
26984 (final_1): ... this. Take SEEN parameter. Output final pending
26985 next view at the end.
26986 (final): Reintroduce seen-less interface.
26987 (final_scan_insn): Output pending next view before switching
26988 sections or ending a block. Mark the next view as needed when
26989 outputting variable locations. Notify debug backend of section
26990 changes, and of location view changes.
26991 (rest_of_handle_final): Adjust.
26992 * toplev.c (process_options): Autodetect value for debug variable
26993 location views option. Warn on incompat5 without -gdwarf-5.
26994 * doc/invoke.texi (gvariable-location-views): New.
26995 (gvariable-location-views=incompat5): New.
26996 (gno-variable-location-views): New.
26998 2018-02-08 David Malcolm <dmalcolm@redhat.com>
27000 PR tree-optimization/84136
27001 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
27002 that the result of find_edge is non-NULL.
27004 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
27007 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
27008 storing integer register in SImode. Fix cost of 256 and 512
27009 byte aligned SSE register store.
27011 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
27013 * config/i386/i386.c (ix86_multiplication_cost): Fix
27014 multiplication cost for TARGET_AVX512DQ.
27016 2018-02-08 Marek Polacek <polacek@redhat.com>
27018 PR tree-optimization/84238
27019 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
27022 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
27024 PR tree-optimization/84265
27025 * tree-vect-stmts.c (vectorizable_store): Don't treat
27026 VMAT_CONTIGUOUS accesses as grouped.
27027 (vectorizable_load): Likewise.
27029 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
27031 PR tree-optimization/81635
27032 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
27033 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
27034 (test_round_for_mask): New functions.
27035 (wide_int_cc_tests): Call test_round_for_mask.
27036 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
27037 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
27038 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
27039 range returned by get_range_info.
27041 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
27044 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
27045 * symtab.c: Include builtins.h
27046 (symtab_node::output_to_lto_symbol_table_p): Move here
27047 from lto-streamer-out.c:output_symbol_p.
27048 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
27049 (output_symbol_p): Move all logic to symtab.c
27050 (produce_symtab): Update.
27052 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27054 * config/s390/s390-opts.h (enum indirect_branch): Define.
27055 * config/s390/s390-protos.h (s390_return_addr_from_memory)
27056 (s390_indirect_branch_via_thunk)
27057 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
27058 (enum s390_indirect_branch_type): Define.
27059 * config/s390/s390.c (struct s390_frame_layout, struct
27060 machine_function): Remove.
27061 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
27062 (indirect_branch_table_label_no, indirect_branch_table_name):
27064 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
27065 (enum s390_indirect_branch_option): Define.
27066 (s390_return_addr_from_memory): New function.
27067 (s390_handle_string_attribute): New function.
27068 (s390_attribute_table): Add new attribute handler.
27069 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
27070 (s390_indirect_branch_via_thunk): New function.
27071 (s390_indirect_branch_via_inline_thunk): New function.
27072 (s390_function_ok_for_sibcall): When jumping via thunk disallow
27073 sibling call optimization for non z10 compiles.
27074 (s390_emit_call): Force indirect branch target to be a single
27075 register. Add r1 clobber for non-z10 compiles.
27076 (s390_emit_epilogue): Emit return jump via return_use expander.
27077 (s390_reorg): Handle JUMP_INSNs as execute targets.
27078 (s390_option_override_internal): Perform validity checks for the
27079 new command line options.
27080 (s390_indirect_branch_attrvalue): New function.
27081 (s390_indirect_branch_settings): New function.
27082 (s390_set_current_function): Invoke s390_indirect_branch_settings.
27083 (s390_output_indirect_thunk_function): New function.
27084 (s390_code_end): Implement target hook.
27085 (s390_case_values_threshold): Implement target hook.
27086 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
27088 * config/s390/s390.h (struct s390_frame_layout)
27089 (struct machine_function): Move here from s390.c.
27090 (TARGET_INDIRECT_BRANCH_NOBP_RET)
27091 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
27092 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
27093 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
27094 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
27095 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
27096 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
27097 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
27098 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
27099 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
27100 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
27101 (mnemonic attribute): Add values which aren't recognized
27103 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
27104 pattern for branch conversion. Fix mnemonic attribute.
27105 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
27106 indirect branch via thunk if requested.
27107 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
27108 ("*indirect_jump"): Disable for branch conversion using out of
27110 ("indirect_jump_via_thunk<mode>_z10")
27111 ("indirect_jump_via_thunk<mode>")
27112 ("indirect_jump_via_inlinethunk<mode>_z10")
27113 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
27114 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
27115 ("casesi_jump_via_inlinethunk<mode>_z10")
27116 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
27117 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
27118 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
27119 ("*indirect2_jump"): Disable for branch conversion.
27120 ("casesi_jump"): Turn into expander and expand patterns for branch
27122 ("return_use"): New expander.
27123 ("*return"): Emit return via thunk and rename it to ...
27124 ("*return<mode>"): ... this one.
27125 * config/s390/s390.opt: Add new options and and enum for the
27128 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
27130 * lra-constraints.c (match_reload): Unconditionally use
27131 gen_lowpart_SUBREG, rather than selecting between that
27132 and equivalent gen_rtx_SUBREG code.
27134 2018-02-08 Richard Biener <rguenther@suse.de>
27136 PR tree-optimization/84233
27137 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
27138 changed flag instead of boguously re-using phi_inserted.
27140 2018-02-08 Martin Jambor <mjambor@suse.cz>
27142 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
27143 static local variables.
27145 2018-02-08 Richard Biener <rguenther@suse.de>
27147 PR tree-optimization/84278
27148 * tree-vect-stmts.c (vectorizable_store): When looking for
27149 smaller vector types to perform grouped strided loads/stores
27150 make sure the mode is supported by the target.
27151 (vectorizable_load): Likewise.
27153 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
27155 * config/aarch64/aarch64.c (aarch64_components_for_bb):
27156 Increase LDP/STP opportunities by adding adjacent callee-saves.
27158 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
27160 PR rtl-optimization/84068
27161 PR rtl-optimization/83459
27162 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
27164 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
27166 PR tree-optimization/84224
27167 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
27168 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
27169 non-zero arguments.
27171 2018-02-07 Iain Sandoe <iain@codesourcery.com>
27174 * config/rs6000/altivec.md (*restore_world): Remove LR use.
27175 * config/rs6000/predicates.md (restore_world_operation): Adjust op
27176 count, remove one USE.
27178 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
27180 * doc/install.texi (Configuration): Document the
27181 --with-long-double-format={ibm,ieee} PowerPC configuration
27185 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
27186 Convert from define_expand to be define_insn_and_split. Rework
27187 float/double/_Float128 conversions to QI/HI/SImode to work with
27188 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
27189 conversions to QI/HImode types did a store and then a load to
27190 truncate the value. For conversions to VSX registers, don't split
27191 the insn, instead emit the code directly. Use the code iterator
27192 any_fix to combine signed and unsigned conversions.
27193 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
27194 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
27195 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
27196 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
27197 (fix_<mode>di2_hw): Likewise.
27198 (fixuns_<mode>di2_hw): Likewise.
27199 (fix_<mode>si2_hw): Likewise.
27200 (fixuns_<mode>si2_hw): Likewise.
27201 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
27202 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
27203 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
27204 fix<uns>_trunc<SFDF:mode>si2_p8.
27205 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
27206 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
27207 (fix<uns>_<mode>_mem): Likewise.
27208 (fctiw<u>z_<mode>_mem): Likewise.
27209 (fix<uns>_<mode>_mem): Likewise.
27210 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
27211 the register allocator from doing a direct move to the GPRs to do
27212 a store, and instead use the ISA 3.0 store byte/half-word from
27213 vector register instruction. For IEEE 128-bit floating point,
27214 also optimize stores of 32-bit ints.
27215 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
27217 2018-02-07 Alan Hayward <alan.hayward@arm.com>
27219 * genextract.c (push_pathstr_operand): New function to support
27221 (walk_rtx): Call push_pathstr_operand.
27222 (print_path): Support [a-zA-Z].
27224 2018-02-07 Richard Biener <rguenther@suse.de>
27226 PR tree-optimization/84037
27227 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
27228 (cse_and_gimplify_to_preheader): Declare.
27229 (vect_get_place_in_interleaving_chain): Likewise.
27230 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
27232 (_loop_vec_info::~_loop_vec_info): Delete it.
27233 (cse_and_gimplify_to_preheader): New function.
27234 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
27235 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
27236 (vectorizable_load): Likewise. For grouped stores always base
27237 the IV on the first element.
27238 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
27239 condition before gimplifying.
27241 2018-02-07 Jakub Jelinek <jakub@redhat.com>
27243 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
27244 *DIV_EXPR and *MOD_EXPR.
27246 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
27249 * config/i386/i386.c (ix86_option_override_internal): Mask out
27250 the CF_SET bit when checking -fcf-protection.
27252 2018-02-07 Tom de Vries <tom@codesourcery.com>
27255 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
27258 2018-02-07 Richard Biener <rguenther@suse.de>
27260 PR tree-optimization/84204
27261 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
27264 PR tree-optimization/84205
27265 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
27266 special-case isl_ast_op_zdiv_r.
27268 PR tree-optimization/84223
27269 * graphite-scop-detection.c (gather_bbs::before_dom_children):
27270 Only add conditions from within the region.
27271 (gather_bbs::after_dom_children): Adjust.
27273 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
27276 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
27277 * config/avr/avr.md: Only post-reload split REG-REG moves if
27278 either register is GENERAL_REG_P.
27280 2018-02-07 Jakub Jelinek <jakub@redhat.com>
27282 PR tree-optimization/84235
27283 * tree-ssa-scopedtables.c
27284 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
27285 if the subtraction is performed in floating point type where NaNs are
27286 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
27287 build 1. Formatting fix.
27289 2018-02-06 Jakub Jelinek <jakub@redhat.com>
27292 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
27293 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
27294 and skip it regardless of bb boundaries. Use CALL_P macro,
27295 don't test INSN_P (insn) together with CALL_P or JUMP_P check
27296 unnecessarily, formatting fix.
27298 2018-02-06 Michael Collison <michael.collison@arm.com>
27300 * config/arm/thumb2.md:
27301 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
27302 (*thumb_mov_notscc): Ditto.
27304 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
27307 * config/rs6000/rs6000.md (su code attribute): Use "u" for
27308 unsigned_fix, not "s".
27310 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27312 * configure.ac (gcc_fn_eh_frame_ro): New function.
27313 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
27314 correct .eh_frame permissions.
27315 * configure: Regenerate.
27317 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
27319 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
27320 irrelevant options.
27322 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27324 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27325 Display warning message for -mno-speculate-indirect-jumps.
27327 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
27329 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
27331 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
27333 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
27335 PR tree-optimization/84225
27336 * tree-eh.c (find_trapping_overflow): Only call
27337 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
27339 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
27342 * config/i386/i386.c: Reimplement the check of possible options
27343 -mibt/-mshstk conbination. Change error messages.
27344 * doc/invoke.texi: Fix a typo: remove extra '='.
27346 2018-02-06 Marek Polacek <polacek@redhat.com>
27348 PR tree-optimization/84228
27349 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
27351 2018-02-06 Tamar Christina <tamar.christina@arm.com>
27354 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
27355 emitted arch directives.
27356 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
27357 __ARM_FEATURE_COPROC before changing architectures.
27359 2018-02-06 Richard Biener <rguenther@suse.de>
27361 * config/i386/i386.c (print_reg): Fix typo.
27362 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
27364 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
27366 * configure: Regenerate.
27368 2018-02-05 Martin Sebor <msebor@redhat.com>
27370 PR tree-optimization/83369
27371 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
27374 2018-02-05 Martin Liska <mliska@suse.cz>
27376 * doc/invoke.texi: Cherry-pick upstream r323995.
27378 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
27380 * ira.c (ira_init_register_move_cost): Adjust comment.
27382 2018-02-05 Martin Liska <mliska@suse.cz>
27384 PR gcov-profile/84137
27385 * doc/gcov.texi: Fix typo in documentation.
27387 2018-02-05 Martin Liska <mliska@suse.cz>
27389 PR gcov-profile/83879
27390 * doc/gcov.texi: Document necessity of --dynamic-list-data when
27391 using dlopen functionality.
27393 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
27395 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
27396 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
27397 _mm_maskz_range_ss, _mm_mask_range_round_ss,
27398 _mm_maskz_range_round_ss): New intrinsics.
27399 (__builtin_ia32_rangesd128_round)
27400 (__builtin_ia32_rangess128_round): Remove.
27401 (__builtin_ia32_rangesd128_mask_round,
27402 __builtin_ia32_rangess128_mask_round): New builtins.
27403 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
27404 __builtin_ia32_rangess128_round): Remove.
27405 (__builtin_ia32_rangesd128_mask_round,
27406 __builtin_ia32_rangess128_mask_round): New builtins.
27407 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
27408 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
27409 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
27410 "<round_saeonly_constraint>")): Changed to ...
27411 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
27412 "<round_saeonly_scalar_constraint>")): ... this.
27413 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
27414 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
27415 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
27416 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
27417 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
27419 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
27421 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
27423 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
27424 Remove all values except native, 8540 and 8548.
27426 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
27428 * config/i386/i386.c (ix86_output_function_return): Pass
27429 INVALID_REGNUM, instead of -1, as invalid register number to
27430 indirect_thunk_name and output_indirect_thunk.
27432 2018-02-02 Julia Koval <julia.koval@intel.com>
27434 * config.gcc: Add -march=icelake.
27435 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
27436 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
27437 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
27438 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
27439 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
27440 (processor_target_table): Add icelake.
27441 (ix86_option_override_internal): Handle new PTAs.
27442 (get_builtin_code_for_version): Handle icelake.
27443 (M_INTEL_COREI7_ICELAKE): New.
27444 (fold_builtin_cpu): Handle icelake.
27445 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
27446 * doc/invoke.texi: Add -march=icelake.
27448 2018-02-02 Julia Koval <julia.koval@intel.com>
27450 * config/i386/i386.c (ix86_option_override_internal): Change flags type
27451 to wide_int_bitmask.
27452 * wide-int-bitmask.h: New.
27454 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
27457 * config/i386/i386.md: Replace Pmode with word_mode in
27458 builtin_setjmp_setup and builtin_longjmp to support x32.
27460 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
27464 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
27466 (rs6000_supported_cpu_names): New static variable.
27467 (linux_cpu_translation_table): Likewise.
27468 (elf_platform) <cpu>: Define new static variable and use it.
27469 Translate kernel AT_PLATFORM name to canonical name if needed.
27470 Error if platform name is unknown.
27472 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
27475 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
27477 2018-02-01 Jeff Law <law@redhat.com>
27480 * config/i386/i386.c (release_scratch_register_on_entry): Add new
27481 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
27482 the scratch if RELEASE_VIA_POP is false.
27483 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
27484 If we have to save a temporary register, decrement SIZE appropriately.
27485 Pass new arguments to release_scratch_register_on_entry.
27486 (ix86_adjust_stack_and_probe): Likewise.
27487 (ix86_emit_probe_stack_range): Pass new arguments to
27488 release_scratch_register_on_entry.
27490 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
27492 PR rtl-optimization/84157
27493 * combine.c (change_zero_ext): Use REG_P predicate in
27494 front of HARD_REGISTER_P predicate.
27496 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
27498 * config/avr/avr.c (avr_option_override): Move disabling of
27499 -fdelete-null-pointer-checks to...
27500 * common/config/avr/avr-common.c (avr_option_optimization_table):
27503 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
27505 PR tree-optimization/81635
27506 * tree-data-ref.c (split_constant_offset_1): For types that
27507 wrap on overflow, try to use range info to prove that wrapping
27510 2018-02-01 Renlin Li <renlin.li@arm.com>
27513 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
27514 TAILCALL_ADDR_REGS.
27515 (aarch64_register_move_cost): Likewise.
27516 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
27517 TAILCALL_ADDR_REGS.
27518 (REG_CLASS_NAMES): Likewise.
27519 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
27520 TAILCALL_ADDR_REGS. Remove IP registers.
27521 * config/aarch64/aarch64.md (Ucs): Update register constraint.
27523 2018-02-01 Richard Biener <rguenther@suse.de>
27525 * domwalk.h (dom_walker::dom_walker): Add additional constructor
27526 for specifying RPO order and allow NULL for that.
27527 * domwalk.c (dom_walker::dom_walker): Likewise.
27528 (dom_walker::walk): Handle NULL RPO order.
27529 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
27531 (rewrite_update_dom_walker): Likewise.
27532 (mark_def_dom_walker): Likewise.
27534 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
27536 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
27537 (aarch64_maybe_expand_sve_subreg_move): Declare.
27538 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
27539 * config/aarch64/predicates.md (aarch64_any_register_operand): New
27541 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
27542 that are semantically a reverse operation.
27543 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
27544 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
27545 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
27547 (aarch64_can_change_mode_class): For big-endian, forbid changes
27548 between two SVE modes if they have different element sizes.
27550 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
27552 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
27553 the TImode handling for big-endian targets.
27555 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
27557 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
27558 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
27560 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
27561 Remove BSWAP handing for big-endian targets and use the form of
27562 LD1RQ appropariate for the mode.
27564 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
27566 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
27567 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
27568 duplicated element.
27570 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
27573 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
27574 check for operands that need to go through aarch64_sve_reload_be.
27576 2018-02-01 Jakub Jelinek <jakub@redhat.com>
27578 PR tree-optimization/81661
27579 PR tree-optimization/84117
27580 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
27581 * tree-eh.c: Include gimplify.h.
27582 (find_trapping_overflow, replace_trapping_overflow,
27583 rewrite_to_non_trapping_overflow): New functions.
27584 * tree-vect-loop.c: Include tree-eh.h.
27585 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
27586 * tree-data-ref.c: Include tree-eh.h.
27587 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
27589 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
27591 PR rtl-optimization/84123
27592 * combine.c (change_zero_ext): Check if hard register satisfies
27593 can_change_dest_mode before calling gen_lowpart_SUBREG.
27595 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
27598 * ira.c (ira_init_register_move_cost): Remove assert.
27600 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
27602 PR rtl-optimization/84071
27603 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
27604 * doc/tm.texi: Regenerate.
27606 2018-01-31 Richard Biener <rguenther@suse.de>
27608 PR tree-optimization/84132
27609 * tree-data-ref.c (analyze_miv_subscript): Properly
27610 check whether evolution_function_is_affine_multivariate_p
27611 before calling gcd_of_steps_may_divide_p.
27613 2018-01-31 Julia Koval <julia.koval@intel.com>
27616 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
27617 * config/i386/i386.md (rdpid_rex64) New.
27618 (rdpid): Make 32bit only.
27620 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
27623 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
27624 an IDENTIFIER_NODE for FUNCTION_TYPE's.
27626 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
27629 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
27631 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
27633 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
27635 PR rtl-optimization/84071
27636 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
27637 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
27639 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
27641 * config/arc/arc.c (arc_handle_aux_attribute): New function.
27642 (arc_attribute_table): Add 'aux' attribute.
27643 (arc_in_small_data_p): Consider aux like variables.
27644 (arc_is_aux_reg_p): New function.
27645 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
27646 (arc_get_aux_arg): New function.
27647 (prepare_move_operands): Handle aux-register access.
27648 (arc_handle_aux_attribute): New function.
27649 * doc/extend.texi (ARC Variable attributes): Add subsection.
27651 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
27653 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
27654 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
27655 (arc_attribute_table): Add 'uncached' attribute.
27656 (arc_print_operand): Print '.di' flag for uncached memory
27658 (arc_in_small_data_p): Do not consider for small data the uncached
27660 (arc_is_uncached_mem_p): New function.
27661 * config/arc/predicates.md (compact_store_memory_operand): Check
27662 for uncached memory accesses.
27663 (nonvol_nonimm_operand): Likewise.
27664 * doc/extend.texi (ARC Type Attribute): New subsection.
27666 2018-01-31 Jakub Jelinek <jakub@redhat.com>
27669 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
27670 falign-loops=): Add Optimization flag.
27672 2018-01-30 Jeff Law <law@redhat.com>
27675 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
27676 INT_REGISTERS_SAVED. Check it prior to calling
27677 get_scratch_register_on_entry.
27678 (ix86_adjust_stack_and_probe): Similarly.
27679 (ix86_emit_probe_stack_range): Similarly.
27680 (ix86_expand_prologue): Corresponding changes.
27682 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27685 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
27686 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
27688 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
27691 * lra-constraints.c (curr_insn_transform): Process AND in the
27694 2018-01-30 Jakub Jelinek <jakub@redhat.com>
27696 PR rtl-optimization/83986
27697 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
27698 dependence against last_pending_memory_flush in addition to
27699 pending_jump_insns.
27701 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
27703 PR tree-optimization/81611
27704 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
27707 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27710 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
27713 2018-01-30 Richard Biener <rguenther@suse.de>
27714 Jakub Jelinek <jakub@redhat.com>
27716 PR tree-optimization/84111
27717 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
27718 inner loops added during recursion, as they don't have up-to-date
27721 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
27724 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
27725 (can_inline_edge_by_limits_p): ... here.
27726 (can_early_inline_edge_p, check_callers,
27727 update_caller_keys, update_callee_keys, recursive_inlining,
27728 add_new_edges_to_heap, speculation_useful_p,
27729 inline_small_functions,
27730 inline_small_functions, flatten_function,
27731 inline_to_all_callers_1): Update.
27733 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
27735 * profile-count.c (profile_count::combine_with_ipa_count): Handle
27738 2018-01-30 Richard Biener <rguenther@suse.de>
27740 PR tree-optimization/83008
27741 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
27742 invariant and constant vector uses in stmts when they need
27743 more than one stmt.
27745 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27748 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
27749 * configure: Regenerate.
27751 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
27753 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
27755 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
27756 Use gen_rtx_REG rather than gen_lowpart.
27758 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
27760 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
27761 rather than 0 when creating partial subregs.
27763 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
27765 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
27768 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
27771 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
27772 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
27773 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
27774 flags. This restores the settings used before the 2017-07-24.
27775 Turning off pre increment/decrement/modify allows IVOPTS to
27776 optimize DF/SF loops where the index is an int.
27778 2018-01-29 Richard Biener <rguenther@suse.de>
27779 Kelvin Nilsen <kelvin@gcc.gnu.org>
27782 * tree-vect-stmts.c (vectorizable_call): Don't call
27783 targetm.vectorize_builtin_md_vectorized_function if callee is
27786 2018-01-22 Carl Love <cel@us.ibm.com>
27788 * doc/extend.tex: Fix typo in second arg in
27789 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
27791 2018-01-29 Richard Biener <rguenther@suse.de>
27793 PR tree-optimization/84086
27794 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
27795 (flush_ssaname_freelist): When SSA names were released reset
27796 the SCEV hash table.
27798 2018-01-29 Richard Biener <rguenther@suse.de>
27800 PR tree-optimization/84057
27801 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
27802 removed paths when removing edges.
27804 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
27806 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
27807 -mfunction-return=@var{choice}.
27809 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
27811 PR diagnostic/84034
27812 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
27813 Handle CR like TAB.
27814 (layout::print_source_line): Likewise.
27815 (test_get_line_width_without_trailing_whitespace): Add test cases.
27817 2018-01-27 Jakub Jelinek <jakub@redhat.com>
27819 PR middle-end/84040
27820 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
27823 2018-01-26 Jim Wilson <jimw@sifive.com>
27825 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
27827 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
27830 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27832 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
27833 and CMP + SUB-immediate -> SUBS.
27835 2018-01-26 Martin Sebor <msebor@redhat.com>
27837 PR tree-optimization/83896
27838 * tree-ssa-strlen.c (get_string_len): Rename...
27839 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
27840 Avoid assuming length is constant.
27841 (handle_char_store): Use HOST_WIDE_INT for string length.
27843 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
27846 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
27847 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
27849 2018-01-26 Richard Biener <rguenther@suse.de>
27851 PR rtl-optimization/84003
27852 * dse.c (record_store): Only record redundant stores when
27853 the earlier store aliases at least all accesses the later one does.
27855 2018-01-26 Jakub Jelinek <jakub@redhat.com>
27857 PR rtl-optimization/83985
27858 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
27859 REG_CFA_RESTORE insns.
27860 (delete_unmarked_insns): Don't ignore separate shrink wrapping
27861 REG_CFA_RESTORE insns here.
27864 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
27865 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
27867 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
27869 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
27870 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
27871 (arc_init): Likewise.
27872 (arc_override_options): Likewise.
27873 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
27875 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
27877 * config/arc/arc.h (TARGET_DBNZ): Define.
27878 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
27879 properly set the tune attribute.
27880 (dbnz): Use TARGET_DBNZ guard.
27881 * config/arc/arc.opt (mtune): Add core3 option.
27883 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
27885 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
27886 recognize new pic like addresses.
27887 (arc_delegitimize_address): Clean up.
27889 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
27891 * config/arc/arc-arches.def: Option mrf16 valid for all
27893 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
27894 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
27895 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
27896 * config/arc/arc-tables.opt: Regenerate.
27897 * config/arc/arc.c (arc_conditional_register_usage): Handle
27898 reduced register file case.
27899 (arc_file_start): Set must have build attributes.
27900 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
27901 mrf16 option value.
27902 * config/arc/arc.opt (mrf16): Add new option.
27903 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
27904 * config/arc/genmultilib.awk: Handle new mrf16 option.
27905 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
27906 * config/arc/t-multilib: Regenerate.
27907 * doc/invoke.texi (ARC Options): Document mrf16 option.
27909 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
27911 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
27912 * config/arc/arc.c (arc_handle_secure_attribute): New function.
27913 (arc_attribute_table): Add 'secure_call' attribute.
27914 (arc_print_operand): Print secure call operand.
27915 (arc_function_ok_for_sibcall): Don't optimize tail calls when
27917 (arc_is_secure_call_p): New function. * config/arc/arc.md
27918 (call_i): Add support for sjli instruction.
27919 (call_value_i): Likewise.
27920 * config/arc/constraints.md (Csc): New constraint.
27922 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
27923 John Eric Martin <John.Martin@emmicro-us.com>
27925 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
27926 * config/arc/arc.c (_arc_jli_section): New struct.
27927 (arc_jli_section): New type.
27928 (rc_jli_sections): New static variable.
27929 (arc_handle_jli_attribute): New function.
27930 (arc_attribute_table): Add jli_always and jli_fixed attribute.
27931 (arc_file_end): New function.
27932 (TARGET_ASM_FILE_END): Define.
27933 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
27934 (arc_add_jli_section): New function.
27935 (jli_call_scan): Likewise.
27936 (arc_reorg): Call jli_call_scan.
27937 (arc_output_addsi): Remove 'S' from printing asm operand.
27938 (arc_is_jli_call_p): New function.
27939 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
27941 (movhi_insn): Likewise.
27942 (movsi_insn): Likewise.
27943 (movsi_set_cc_insn): Likewise.
27944 (loadqi_update): Likewise.
27945 (load_zeroextendqisi_update): Likewise.
27946 (load_signextendqisi_update): Likewise.
27947 (loadhi_update): Likewise.
27948 (load_zeroextendhisi_update): Likewise.
27949 (load_signextendhisi_update): Likewise.
27950 (loadsi_update): Likewise.
27951 (loadsf_update): Likewise.
27952 (movsicc_insn): Likewise.
27953 (bset_insn): Likewise.
27954 (bxor_insn): Likewise.
27955 (bclr_insn): Likewise.
27956 (bmsk_insn): Likewise.
27957 (bicsi3_insn): Likewise.
27958 (cmpsi_cc_c_insn): Likewise.
27959 (movsi_ne): Likewise.
27960 (movsi_cond_exec): Likewise.
27961 (clrsbsi2): Likewise.
27962 (norm_f): Likewise.
27972 (call_i): Remove 'S' asm letter, add jli instruction.
27973 (call_value_i): Likewise.
27974 * config/arc/arc.op (mjli-always): New option.
27975 * config/arc/constraints.md (Cji): New constraint.
27976 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
27978 (subsf3_fpx): Likewise.
27979 (mulsf3_fpx): Likewise.
27980 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
27982 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
27983 function attrbutes.
27984 * doc/invoke.texi (ARC): Document mjli-always option.
27986 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
27988 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
27989 avoid addition with 0 and use incw and decw where possible.
27991 2018-01-26 Richard Biener <rguenther@suse.de>
27993 PR tree-optimization/81082
27994 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
27995 association if it requires casting to unsigned.
27996 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
27997 from fold_plusminus_mult_expr to catch important cases late when
27998 range info is available.
28000 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28002 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
28003 * configure.ac (hidden_linkonce): New test.
28004 * configure: Regenerate.
28005 * config.in: Regenerate.
28007 2018-01-26 Julia Koval <julia.koval@intel.com>
28009 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
28010 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
28011 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
28012 _mm_mask_bitshuffle_epi64_mask): Fix type.
28013 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
28014 USI_FTYPE_V4DI_V4DI_USI): Remove.
28015 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
28016 __builtin_ia32_vpshufbitqmb256_mask,
28017 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
28018 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
28019 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
28021 2018-01-26 Alan Modra <amodra@gmail.com>
28024 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
28025 UNSPEC_VBPERMQ. Sort other unspecs.
28027 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
28029 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
28031 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
28033 PR middle-end/83055
28034 * predict.c (drop_profile): Do not push/pop cfun; update also
28036 (handle_missing_profiles): Fix logic looking for zero profiles.
28038 2018-01-25 Jakub Jelinek <jakub@redhat.com>
28040 PR middle-end/83977
28041 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
28042 on functions with #pragma omp declare simd or functions with simd
28044 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
28045 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
28046 Remove trailing \n from warning_at calls.
28048 2018-01-25 Tom de Vries <tom@codesourcery.com>
28051 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
28052 for neutered workers.
28054 2018-01-24 Joseph Myers <joseph@codesourcery.com>
28057 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
28058 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
28060 2018-01-24 Jeff Law <law@redhat.com>
28063 * i386.c (get_probe_interval): Move to earlier point.
28064 (ix86_compute_frame_layout): If -fstack-clash-protection and
28065 the frame is larger than the probe interval, then use pushes
28066 to save registers rather than reg->mem moves.
28067 (ix86_expand_prologue): Remove conditional for int_registers_saved
28070 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
28073 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
28074 min/max for never referenced object.
28076 2018-01-24 Jakub Jelinek <jakub@redhat.com>
28078 PR middle-end/83977
28079 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
28081 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
28082 attributes from DECL_ATTRIBUTES (decl) without affecting
28083 DECL_ATTRIBUTES (current_function_decl).
28084 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
28085 functions with non-NULL DECL_ABSTRACT_ORIGIN.
28087 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
28089 PR tree-optimization/83979
28090 * fold-const.c (fold_comparison): Use constant_boolean_node
28091 instead of boolean_{true,false}_node.
28093 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
28095 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
28098 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28100 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
28101 Simplify the clause that sets the length attribute.
28102 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
28103 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
28104 clause that sets the length attribute.
28105 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
28107 2018-01-24 Tom de Vries <tom@codesourcery.com>
28110 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
28111 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
28112 Add strict parameter.
28113 (prevent_branch_around_nothing): Insert dummy insn between branch to
28114 label and label with no ptx insn inbetween.
28115 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
28117 2018-01-24 Tom de Vries <tom@codesourcery.com>
28120 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
28121 for neutered threads in warp.
28122 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
28124 2018-01-24 Richard Biener <rguenther@suse.de>
28126 PR tree-optimization/83176
28127 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
28130 2018-01-24 Richard Biener <rguenther@suse.de>
28132 PR tree-optimization/82819
28133 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
28134 code generating pluses that are no-ops in the target precision.
28136 2018-01-24 Richard Biener <rguenther@suse.de>
28138 PR middle-end/84000
28139 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
28141 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
28143 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
28144 to merge probabilities.
28145 * predict.c (probably_never_executed): Also mark as cold functions
28146 with global 0 profile and guessed local profile.
28147 * profile-count.c (profile_probability::combine_with_count): New
28149 * profile-count.h (profile_probability::operator*,
28150 profile_probability::operator*=, profile_probability::operator/,
28151 profile_probability::operator/=): Reduce precision to adjusted
28152 and set value to guessed on contradictory divisions.
28153 (profile_probability::combine_with_freq): Remove.
28154 (profile_probability::combine_wiht_count): Declare.
28155 (profile_count::force_nonzero):: Set to adjusted.
28156 (profile_count::probability_in):: Set quality to adjusted.
28157 * tree-ssa-tail-merge.c (replace_block_by): Use
28158 combine_with_count.
28160 2018-01-23 Andrew Waterman <andrew@sifive.com>
28161 Jim Wilson <jimw@sifive.com>
28163 * config/riscv/riscv.c (riscv_stack_boundary): New.
28164 (riscv_option_override): Set riscv_stack_boundary. Handle
28165 riscv_preferred_stack_boundary_arg.
28166 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
28167 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
28168 (STACK_BOUNDARY): Set to riscv_stack_boundary.
28169 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
28170 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
28171 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
28173 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
28176 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
28177 of struct ix86_frame.
28178 (ix86_expand_epilogue): Likewise. Add a local variable for
28179 the reg_save_offset field in struct ix86_frame.
28181 2018-01-23 Bin Cheng <bin.cheng@arm.com>
28183 PR tree-optimization/82604
28184 * tree-loop-distribution.c (enum partition_kind): New enum item
28185 PKIND_PARTIAL_MEMSET.
28186 (partition_builtin_p): Support above new enum item.
28187 (generate_code_for_partition): Ditto.
28188 (compute_access_range): Differentiate cases that equality can be
28189 proven at all loops, the innermost loops or no loops.
28190 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
28191 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
28192 (finalize_partitions, distribute_loop): Don't fuse partition of
28193 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
28194 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
28195 parloop is enabled.
28197 2018-01-23 Martin Liska <mliska@suse.cz>
28199 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
28200 order to ignore the predictor.
28201 (PRED_POLYMORPHIC_CALL): Likewise.
28202 (PRED_RECURSIVE_CALL): Likewise.
28204 2018-01-23 Martin Liska <mliska@suse.cz>
28206 * tree-profile.c (tree_profiling): Print function header to
28207 aware reader which function we are working on.
28208 * value-prof.c (gimple_find_values_to_profile): Do not print
28209 not interesting value histograms.
28211 2018-01-23 Martin Liska <mliska@suse.cz>
28213 * profile-count.h (enum profile_quality): Add
28214 profile_uninitialized as the first value. Do not number values
28215 as they are zero based.
28216 (profile_count::verify): Update sanity check.
28217 (profile_probability::verify): Likewise.
28219 2018-01-23 Nathan Sidwell <nathan@acm.org>
28221 * doc/invoke.texi (ffor-scope): Deprecate.
28223 2018-01-23 David Malcolm <dmalcolm@redhat.com>
28225 PR tree-optimization/83510
28226 * domwalk.c (set_all_edges_as_executable): New function.
28227 (dom_walker::dom_walker): Convert bool param
28228 "skip_unreachable_blocks" to enum reachability. Move setup of
28229 edge flags to set_all_edges_as_executable and only do it when
28230 reachability is REACHABLE_BLOCKS.
28231 * domwalk.h (enum dom_walker::reachability): New enum.
28232 (dom_walker::dom_walker): Convert bool param
28233 "skip_unreachable_blocks" to enum reachability.
28234 (set_all_edges_as_executable): New decl.
28235 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
28236 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
28238 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
28239 but converting true to REACHABLE_BLOCKS.
28240 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
28242 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
28243 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
28244 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
28246 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
28247 if check_all_array_refs will be called.
28249 2018-01-23 David Malcolm <dmalcolm@redhat.com>
28251 * tree.c (selftest::test_location_wrappers): Add more test
28254 2018-01-23 David Malcolm <dmalcolm@redhat.com>
28256 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
28257 (selftest::test_bit_in_range): Likewise.
28259 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
28262 * doc/sourcebuild.texi (vect_float): Say that the selector
28263 only describes the situation when -funsafe-math-optimizations is on.
28264 (vect_float_strict): Document.
28266 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
28268 PR tree-optimization/83965
28269 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
28270 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
28271 instead of checking only for a reduction.
28272 (vect_recog_widen_sum_pattern): Likewise.
28274 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
28276 * predict.c (probably_never_executed): Only use precise profile info.
28277 (compute_function_frequency): Skip after inlining hack since we now
28278 have quality checking.
28280 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
28282 * profile-count.h (profile_probability::very_unlikely,
28283 profile_probability::unlikely, profile_probability::even): Set
28284 precision to guessed.
28286 2018-01-23 Richard Biener <rguenther@suse.de>
28288 PR tree-optimization/83963
28289 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
28290 Properly terminate dominator walk when crossing the exit edge not
28291 when visiting its source block.
28293 2018-01-23 Jakub Jelinek <jakub@redhat.com>
28296 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
28297 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
28299 2018-01-22 Jakub Jelinek <jakub@redhat.com>
28301 PR tree-optimization/83957
28302 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
28303 semicolon after for body surrounded by braces.
28305 PR tree-optimization/83081
28306 * profile-count.h (profile_probability::split): New method.
28307 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
28308 Use profile_probability::split.
28309 (do_compare_rtx_and_jump): Fix adjustment of probabilities
28310 when splitting a single conditional jump into 2.
28312 2018-01-22 David Malcolm <dmalcolm@redhat.com>
28314 PR tree-optimization/69452
28315 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
28318 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
28320 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
28321 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
28322 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
28324 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
28326 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
28328 * config/rl78/rl78.md (movdi): New define_expand.
28329 * config/rl78/rl78.c (rl78_split_movdi): New function.
28331 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
28334 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
28336 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
28337 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
28338 128-bit to produce an UNSPEC move to get the double word with the
28339 signbit and then a shift directly to do signbit.
28340 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
28341 implementation with a new version that just does either a direct
28342 move or a regular move. Move memory interface to separate insns.
28343 Move insns so they are next to the expander.
28344 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
28345 with signbit move. Split big and little endian case.
28346 (signbit<mode>2_dm_mem_le): Likewise.
28347 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
28348 (signbit<mode>2_dm2): Likewise.
28350 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
28352 * config/rl78/rl78.md (anddi3): New define_expand.
28354 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
28356 * config/rl78/rl78.md (umindi3): New define_expand.
28358 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
28360 * config/rl78/rl78.md (smindi3): New define_expand.
28362 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
28364 * config/rl78/rl78.md (smaxdi3): New define_expand.
28366 2018-01-22 Carl Love <cel@us.ibm.com>
28368 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
28369 LVX_V1TI): Add macro expansion.
28370 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
28371 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
28372 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
28373 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
28374 Change check to determine if the instruction is a byte reversing
28375 entry. Fix typo in comment.
28376 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
28377 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
28378 Add def_builtin calls for new builtins.
28379 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
28380 Add define_insn expansion.
28382 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
28384 * config/rl78/rl78.md (umaxdi3): New define_expand.
28386 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
28388 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
28389 for non-QImode registers.
28391 2018-01-22 Richard Biener <rguenther@suse.de>
28393 PR tree-optimization/83963
28394 * graphite-scop-detection.c (scop_detection::get_sese): Delay
28395 including the loop exit block.
28396 (scop_detection::merge_sese): Likewise.
28397 (scop_detection::add_scop): Do it here instead.
28399 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28401 * doc/sourcebuild.texi (arm_softfloat): Document.
28403 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
28406 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
28407 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
28408 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
28410 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28411 David Edelsohn <dje.gcc@gmail.com>
28414 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
28415 Change "crset eq" to "crset 2".
28416 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
28417 (*call_indirect_aix<mode>_nospec): Likewise.
28418 (*call_value_indirect_aix<mode>_nospec): Likewise.
28419 (*call_indirect_elfv2<mode>_nospec): Likewise.
28420 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
28421 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
28422 change assembly output from . to $.
28423 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
28424 (indirect_jump<mode>_nospec): Change assembly output from . to $.
28425 (*tablejump<mode>_internal1_nospec): Likewise.
28427 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
28430 * config/sh/sh_optimize_sett_clrt.cc:
28431 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
28433 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
28435 PR tree-optimization/83940
28436 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
28437 offset_dt to vect_constant_def rather than vect_unknown_def_type.
28438 (vect_check_load_store_mask): Add a mask_dt_out parameter and
28439 use it to pass back the definition type.
28440 (vect_check_store_rhs): Likewise rhs_dt_out.
28441 (vect_build_gather_load_calls): Add a mask_dt argument and use
28442 it instead of a call to vect_is_simple_use.
28443 (vectorizable_store): Update calls to vect_check_load_store_mask
28444 and vect_check_store_rhs. Use the dt returned by the latter instead
28445 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
28446 instead of calls to vect_is_simple_use. Pass the scalar rather
28447 than the vector operand to vect_is_simple_use when handling
28448 second and subsequent copies of an rhs value.
28449 (vectorizable_load): Update calls to vect_check_load_store_mask
28450 and vect_build_gather_load_calls. Use the cached mask_dt and
28451 gs_info.offset_dt instead of calls to vect_is_simple_use.
28453 2018-01-20 Jakub Jelinek <jakub@redhat.com>
28455 PR middle-end/83945
28456 * tree-emutls.c: Include gimplify.h.
28457 (lower_emutls_2): New function.
28458 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
28459 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
28460 it before further processing.
28463 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
28464 UINTVAL (trueop1) instead of INTVAL (op1).
28466 2018-01-19 Jakub Jelinek <jakub@redhat.com>
28470 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
28471 INCOMING_FRAME_SP_OFFSET if not defined.
28472 (scan_trace): Add ENTRY argument. If true and
28473 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
28474 emit a note to adjust the CFA offset.
28475 (create_cfi_notes): Adjust scan_trace callers.
28476 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
28477 INCOMING_FRAME_SP_OFFSET in the CIE.
28478 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
28479 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
28481 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
28482 * doc/tm.texi: Regenerated.
28484 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28486 PR rtl-optimization/83147
28487 * lra-constraints.c (remove_inheritance_pseudos): Use
28488 lra_substitute_pseudo_within_insn.
28490 2018-01-19 Tom de Vries <tom@codesourcery.com>
28491 Cesar Philippidis <cesar@codesourcery.com>
28494 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
28496 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
28499 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
28500 spaces for function labels.
28502 2018-01-19 Martin Liska <mliska@suse.cz>
28504 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
28505 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
28506 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
28507 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
28508 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
28509 (PRED_CONST_RETURN): Change from 69 to 65.
28510 (PRED_NULL_RETURN): Change from 91 to 71.
28511 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
28512 (PRED_LOOP_GUARD): Change from 66 to 73.
28514 2018-01-19 Martin Liska <mliska@suse.cz>
28516 * predict.c (predict_insn_def): Add new assert.
28517 (struct branch_predictor): Change type to signed integer.
28518 (test_prediction_value_range): Amend test to cover
28519 PROB_UNINITIALIZED.
28520 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
28521 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
28522 (PRED_LOOP_ITERATIONS_MAX): Likewise.
28523 (PRED_LOOP_IV_COMPARE): Likewise.
28524 * predict.h (PROB_UNINITIALIZED): Define new constant.
28526 2018-01-19 Martin Liska <mliska@suse.cz>
28528 * predict.c (dump_prediction): Add new format for
28529 analyze_brprob.py script which is enabled with -details
28531 * profile-count.h (precise_p): New function.
28533 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
28535 PR tree-optimization/83922
28536 * tree-vect-loop.c (vect_verify_full_masking): Return false if
28537 there are no statements that need masking.
28538 (vect_active_double_reduction_p): New function.
28539 (vect_analyze_loop_operations): Use it when handling phis that
28540 are not in the loop header.
28542 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
28544 PR tree-optimization/83914
28545 * tree-vect-loop.c (vectorizable_induction): Don't convert
28546 init_expr or apply the peeling adjustment for inductions
28547 that are nested within the vectorized loop.
28549 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28551 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
28554 2018-01-18 Jakub Jelinek <jakub@redhat.com>
28558 * function.h (gimplify_parameters): Add gimple_seq * argument.
28559 * function.c: Include gimple.h and options.h.
28560 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
28561 for the added local temporaries if needed.
28562 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
28563 if there are any parameter cleanups, wrap whole body into a
28564 try/finally with the cleanups.
28566 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
28569 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
28570 Use GET_MODE_CLASS for scalar floating point.
28572 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
28576 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
28577 Fix call of call_cgraph_insertion_hooks.
28579 2018-01-18 Martin Sebor <msebor@redhat.com>
28581 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
28583 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
28586 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
28589 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
28592 * common.opt: (-ffile-prefix-map): New option.
28593 * opts.c (common_handle_option): Defer it.
28594 * opts-global.c (handle_common_deferred_options): Handle it.
28595 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
28596 * file-prefix-map.h: New file.
28597 (remap_debug_filename, add_debug_prefix_map): ...here.
28598 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
28599 * final.c (debug_prefix_map, add_debug_prefix_map
28600 remap_debug_filename): Move to...
28601 * file-prefix-map.c: New file.
28602 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
28603 generalize, get rid of alloca(), use strrchr() instead of strchr().
28604 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
28605 Implement in terms of add_prefix_map().
28606 (remap_macro_filename, remap_debug_filename): Implement in term of
28608 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
28609 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
28610 * dbxout.c: Include file-prefix-map.h.
28611 * varasm.c: Likewise.
28612 * vmsdbgout.c: Likewise.
28613 * xcoffout.c: Likewise.
28614 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
28615 * doc/cppopts.texi (-fmacro-prefix-map): Document.
28616 * doc/invoke.texi (-ffile-prefix-map): Document.
28617 (-fdebug-prefix-map): Update description.
28619 2018-01-18 Martin Liska <mliska@suse.cz>
28621 * config/i386/i386.c (indirect_thunk_name): Document that also
28623 (output_indirect_thunk): Document why both instructions
28624 (pause and lfence) are generated.
28626 2018-01-18 Richard Biener <rguenther@suse.de>
28628 PR tree-optimization/83887
28629 * graphite-scop-detection.c
28630 (scop_detection::get_nearest_dom_with_single_entry): Remove.
28631 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
28632 (scop_detection::merge_sese): Re-implement with a flood-fill
28633 algorithm that properly finds a SESE region if it exists.
28635 2018-01-18 Jakub Jelinek <jakub@redhat.com>
28638 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
28639 pointer_diff optimizations use view_convert instead of convert.
28641 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28643 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
28644 Generate different code for -mno-speculate-indirect-jumps.
28645 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
28646 (*call_indirect_aix<mode>): Disable for
28647 -mno-speculate-indirect-jumps.
28648 (*call_indirect_aix<mode>_nospec): New define_insn.
28649 (*call_value_indirect_aix<mode>): Disable for
28650 -mno-speculate-indirect-jumps.
28651 (*call_value_indirect_aix<mode>_nospec): New define_insn.
28652 (*sibcall_nonlocal_sysv<mode>): Generate different code for
28653 -mno-speculate-indirect-jumps.
28654 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
28656 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
28658 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
28659 long double type, set the flags for noting the default long double
28660 type, even if we don't pass or return a long double type.
28662 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
28665 * ipa-inline.c (flatten_function): Do not overwrite final inlining
28668 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
28670 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
28671 support for merge[hl].
28672 (fold_mergehl_helper): New helper function.
28673 (tree-vector-builder.h): New #include for tree_vector_builder usage.
28674 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
28675 (altivec_vmrglw_direct): Add xxmrglw insn.
28677 2018-01-17 Andrew Waterman <andrew@sifive.com>
28679 * config/riscv/riscv.c (riscv_conditional_register_usage): If
28680 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
28682 2018-01-17 David Malcolm <dmalcolm@redhat.com>
28685 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
28686 call the lto_location_cache before reading the
28687 DECL_SOURCE_LOCATION of the types.
28689 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
28690 Richard Sandiford <richard.sandiford@linaro.org>
28692 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
28693 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
28694 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
28695 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
28696 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
28698 * config/aarch64/constraints.md (aarch64_movti_operand):
28700 * config/aarch64/predicates.md (Uti): Add new constraint.
28702 2018-01-17 Carl Love <cel@us.ibm.com>
28704 * config/rs6000/vsx.md (define_expand xl_len_r,
28705 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
28706 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
28708 (define_expand, define_insn): Move the shift left from the
28709 define_insn to the define_expand for lxvl and stxvl instructions.
28710 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
28711 and XL_LEN_R definitions to PURE.
28713 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
28715 * config/i386/i386.c (indirect_thunk_name): Declare regno
28716 as unsigned int. Compare regno with INVALID_REGNUM.
28717 (output_indirect_thunk): Ditto.
28718 (output_indirect_thunk_function): Ditto.
28719 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
28720 in the call to output_indirect_thunk_function.
28722 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
28724 PR middle-end/83884
28725 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
28726 rather than the size of inner_type to determine the stack slot size
28727 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
28729 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
28732 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
28735 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28737 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
28738 endian Linux systems to optionally enable multilibs for selecting
28739 the long double type if the user configured an explicit type.
28740 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
28741 have no long double multilibs if not defined.
28742 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
28743 warn if the user used -mabi={ieee,ibm}longdouble and we built
28744 multilibs for long double.
28745 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
28746 appropriate multilib option.
28747 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
28749 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
28750 for building long double multilibs.
28751 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
28753 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
28755 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
28758 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
28760 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
28763 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
28766 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
28769 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
28771 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
28772 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
28774 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
28776 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
28777 different rtl trees depending on TARGET_64BIT.
28778 (rs6000_gen_lvx): Likewise.
28780 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
28782 * config/visium/visium.md (nop): Tweak comment.
28783 (hazard_nop): Likewise.
28785 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28787 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
28788 -mspeculate-indirect-jumps.
28789 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
28790 for -mno-speculate-indirect-jumps.
28791 (*call_indirect_elfv2<mode>_nospec): New define_insn.
28792 (*call_value_indirect_elfv2<mode>): Disable for
28793 -mno-speculate-indirect-jumps.
28794 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
28795 (indirect_jump): Emit different RTL for
28796 -mno-speculate-indirect-jumps.
28797 (*indirect_jump<mode>): Disable for
28798 -mno-speculate-indirect-jumps.
28799 (*indirect_jump<mode>_nospec): New define_insn.
28800 (tablejump): Emit different RTL for
28801 -mno-speculate-indirect-jumps.
28802 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
28803 (tablejumpsi_nospec): New define_expand.
28804 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
28805 (tablejumpdi_nospec): New define_expand.
28806 (*tablejump<mode>_internal1): Disable for
28807 -mno-speculate-indirect-jumps.
28808 (*tablejump<mode>_internal1_nospec): New define_insn.
28809 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
28812 2018-01-16 Artyom Skrobov tyomitch@gmail.com
28814 * caller-save.c (insert_save): Drop unnecessary parameter. All
28817 2018-01-16 Jakub Jelinek <jakub@redhat.com>
28818 Richard Biener <rguenth@suse.de>
28821 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
28822 return early, inline manually is_gimple_sizepos. Make sure if we
28823 call gimplify_expr we don't end up with a gimple constant.
28824 * tree.c (variably_modified_type_p): Don't return true for
28825 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
28826 * gimplify.h (is_gimple_sizepos): Remove.
28828 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
28830 PR tree-optimization/83857
28831 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
28832 vectorizable_live_operation for pure SLP statements.
28833 (vectorizable_live_operation): Handle PHIs.
28835 2018-01-16 Richard Biener <rguenther@suse.de>
28837 PR tree-optimization/83867
28838 * tree-vect-stmts.c (vect_transform_stmt): Precompute
28839 nested_in_vect_loop_p since the scalar stmt may get invalidated.
28841 2018-01-16 Jakub Jelinek <jakub@redhat.com>
28844 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
28845 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
28846 If off is not INTEGER_CST, issue a may not be aligned warning
28847 rather than isn't aligned. Use isn%'t rather than isn't.
28848 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
28850 <case MULT_EXPR>: Improve the case when bottom and one of the
28851 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
28852 operand, in that case check if the other operand is multiple of
28853 bottom divided by the INTEGER_CST operand.
28855 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
28858 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
28859 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
28860 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
28861 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
28862 * config/pa/pa.c (pa_function_arg_advance): Likewise.
28863 (pa_function_arg, pa_arg_partial_bytes): Likewise.
28864 (pa_function_arg_size): New function.
28866 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
28868 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
28869 in a separate statement.
28871 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
28873 PR tree-optimization/83847
28874 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
28875 group gathers and scatters.
28877 2018-01-16 Jakub Jelinek <jakub@redhat.com>
28879 PR rtl-optimization/86620
28880 * params.def (max-sched-ready-insns): Bump minimum value to 1.
28882 PR rtl-optimization/83213
28883 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
28884 to last if both are JUMP_INSNs.
28886 PR tree-optimization/83843
28887 * gimple-ssa-store-merging.c
28888 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
28889 store_immediate_info for bswap/nop orig_stores.
28891 2018-01-15 Andrew Waterman <andrew@sifive.com>
28893 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
28895 <UDIV>: Increase cost if !TARGET_DIV.
28897 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28899 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
28900 (define_attr "cr_logical_3op"): New.
28901 (cceq_ior_compare): Adjust.
28902 (cceq_ior_compare_complement): Adjust.
28903 (*cceq_rev_compare): Adjust.
28904 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
28905 (is_cracked_insn): Adjust.
28906 (insn_must_be_first_in_group): Adjust.
28907 * config/rs6000/40x.md: Adjust.
28908 * config/rs6000/440.md: Adjust.
28909 * config/rs6000/476.md: Adjust.
28910 * config/rs6000/601.md: Adjust.
28911 * config/rs6000/603.md: Adjust.
28912 * config/rs6000/6xx.md: Adjust.
28913 * config/rs6000/7450.md: Adjust.
28914 * config/rs6000/7xx.md: Adjust.
28915 * config/rs6000/8540.md: Adjust.
28916 * config/rs6000/cell.md: Adjust.
28917 * config/rs6000/e300c2c3.md: Adjust.
28918 * config/rs6000/e500mc.md: Adjust.
28919 * config/rs6000/e500mc64.md: Adjust.
28920 * config/rs6000/e5500.md: Adjust.
28921 * config/rs6000/e6500.md: Adjust.
28922 * config/rs6000/mpc.md: Adjust.
28923 * config/rs6000/power4.md: Adjust.
28924 * config/rs6000/power5.md: Adjust.
28925 * config/rs6000/power6.md: Adjust.
28926 * config/rs6000/power7.md: Adjust.
28927 * config/rs6000/power8.md: Adjust.
28928 * config/rs6000/power9.md: Adjust.
28929 * config/rs6000/rs64.md: Adjust.
28930 * config/rs6000/titan.md: Adjust.
28932 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
28934 * config/i386/predicates.md (indirect_branch_operand): Rewrite
28935 ix86_indirect_branch_register logic.
28937 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
28939 * config/i386/constraints.md (Bs): Update
28940 ix86_indirect_branch_register check. Don't check
28941 ix86_indirect_branch_register with GOT_memory_operand.
28943 * config/i386/predicates.md (GOT_memory_operand): Don't check
28944 ix86_indirect_branch_register here.
28945 (GOT32_symbol_operand): Likewise.
28947 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
28949 * config/i386/predicates.md (constant_call_address_operand):
28950 Rewrite ix86_indirect_branch_register logic.
28951 (sibcall_insn_operand): Likewise.
28953 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
28955 * config/i386/constraints.md (Bs): Replace
28956 ix86_indirect_branch_thunk_register with
28957 ix86_indirect_branch_register.
28959 * config/i386/i386.md (indirect_jump): Likewise.
28960 (tablejump): Likewise.
28961 (*sibcall_memory): Likewise.
28962 (*sibcall_value_memory): Likewise.
28963 Peepholes of indirect call and jump via memory: Likewise.
28964 * config/i386/i386.opt: Likewise.
28965 * config/i386/predicates.md (indirect_branch_operand): Likewise.
28966 (GOT_memory_operand): Likewise.
28967 (call_insn_operand): Likewise.
28968 (sibcall_insn_operand): Likewise.
28969 (GOT32_symbol_operand): Likewise.
28971 2018-01-15 Jakub Jelinek <jakub@redhat.com>
28973 PR middle-end/83837
28974 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
28975 type rather than type addr's type points to.
28976 (expand_omp_atomic_mutex): Likewise.
28977 (expand_omp_atomic): Likewise.
28979 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
28982 * config/i386/i386.c (output_indirect_thunk_function): Use
28983 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
28984 for __x86_return_thunk.
28986 2018-01-15 Richard Biener <rguenther@suse.de>
28988 PR middle-end/83850
28989 * expmed.c (extract_bit_field_1): Fix typo.
28991 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28994 * config/arm/iterators.md (VF): New mode iterator.
28995 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
28996 Remove integer-related logic from pattern.
28997 (neon_vabd<mode>_3): Likewise.
28999 2018-01-15 Jakub Jelinek <jakub@redhat.com>
29001 PR middle-end/82694
29002 * common.opt (fstrict-overflow): No longer an alias.
29003 (fwrapv-pointer): New option.
29004 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
29005 also for pointer types based on flag_wrapv_pointer.
29006 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
29007 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
29008 opts->x_flag_wrapv got set.
29009 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
29010 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
29011 POINTER_TYPE_OVERFLOW_UNDEFINED.
29012 * match.pd: Likewise in address comparison pattern.
29013 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
29015 2018-01-15 Richard Biener <rguenther@suse.de>
29018 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
29019 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
29020 Reset type names to their identifier if their TYPE_DECL doesn't
29021 have linkage (and thus is used for ODR and devirt).
29022 (save_debug_info_for_decl): Remove.
29023 (save_debug_info_for_type): Likewise.
29024 (add_tree_to_fld_list): Adjust.
29025 * tree-pretty-print.c (dump_generic_node): Make dumping of
29026 type names more robust.
29028 2018-01-15 Richard Biener <rguenther@suse.de>
29030 * BASE-VER: Bump to 8.0.1.
29032 2018-01-14 Martin Sebor <msebor@redhat.com>
29035 * builtins.c (check_access): Avoid warning when the no-warning bit
29038 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
29040 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
29041 * ira-color (allocno_hard_regs_compare): Likewise.
29043 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
29046 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
29047 Use .pushsection/.popsection.
29049 2018-01-14 Martin Sebor <msebor@redhat.com>
29052 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
29054 2018-01-14 Jakub Jelinek <jakub@redhat.com>
29056 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
29057 entry from extra_headers.
29058 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
29059 extra_headers, make the list bitwise identical to the i?86-*-* one.
29061 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
29063 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
29064 -mcmodel=large with -mindirect-branch=thunk,
29065 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
29066 -mfunction-return=thunk-extern.
29067 * doc/invoke.texi: Document -mcmodel=large is incompatible with
29068 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
29069 -mfunction-return=thunk and -mfunction-return=thunk-extern.
29071 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
29073 * config/i386/i386.c (print_reg): Print the name of the full
29074 integer register without '%'.
29075 (ix86_print_operand): Handle 'V'.
29076 * doc/extend.texi: Document 'V' modifier.
29078 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
29080 * config/i386/constraints.md (Bs): Disallow memory operand for
29081 -mindirect-branch-register.
29083 * config/i386/predicates.md (indirect_branch_operand): Likewise.
29084 (GOT_memory_operand): Likewise.
29085 (call_insn_operand): Likewise.
29086 (sibcall_insn_operand): Likewise.
29087 (GOT32_symbol_operand): Likewise.
29088 * config/i386/i386.md (indirect_jump): Call convert_memory_address
29089 for -mindirect-branch-register.
29090 (tablejump): Likewise.
29091 (*sibcall_memory): Likewise.
29092 (*sibcall_value_memory): Likewise.
29093 Disallow peepholes of indirect call and jump via memory for
29094 -mindirect-branch-register.
29095 (*call_pop): Replace m with Bw.
29096 (*call_value_pop): Likewise.
29097 (*sibcall_pop_memory): Replace m with Bs.
29098 * config/i386/i386.opt (mindirect-branch-register): New option.
29099 * doc/invoke.texi: Document -mindirect-branch-register option.
29101 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
29103 * config/i386/i386-protos.h (ix86_output_function_return): New.
29104 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
29105 set function_return_type.
29106 (indirect_thunk_name): Add ret_p to indicate thunk for function
29108 (output_indirect_thunk_function): Pass false to
29109 indirect_thunk_name.
29110 (ix86_output_indirect_branch_via_reg): Likewise.
29111 (ix86_output_indirect_branch_via_push): Likewise.
29112 (output_indirect_thunk_function): Create alias for function
29113 return thunk if regno < 0.
29114 (ix86_output_function_return): New function.
29115 (ix86_handle_fndecl_attribute): Handle function_return.
29116 (ix86_attribute_table): Add function_return.
29117 * config/i386/i386.h (machine_function): Add
29118 function_return_type.
29119 * config/i386/i386.md (simple_return_internal): Use
29120 ix86_output_function_return.
29121 (simple_return_internal_long): Likewise.
29122 * config/i386/i386.opt (mfunction-return=): New option.
29123 (indirect_branch): Mention -mfunction-return=.
29124 * doc/extend.texi: Document function_return function attribute.
29125 * doc/invoke.texi: Document -mfunction-return= option.
29127 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
29129 * config/i386/i386-opts.h (indirect_branch): New.
29130 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
29131 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
29132 with local indirect jump when converting indirect call and jump.
29133 (ix86_set_indirect_branch_type): New.
29134 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
29135 (indirectlabelno): New.
29136 (indirect_thunk_needed): Likewise.
29137 (indirect_thunk_bnd_needed): Likewise.
29138 (indirect_thunks_used): Likewise.
29139 (indirect_thunks_bnd_used): Likewise.
29140 (INDIRECT_LABEL): Likewise.
29141 (indirect_thunk_name): Likewise.
29142 (output_indirect_thunk): Likewise.
29143 (output_indirect_thunk_function): Likewise.
29144 (ix86_output_indirect_branch_via_reg): Likewise.
29145 (ix86_output_indirect_branch_via_push): Likewise.
29146 (ix86_output_indirect_branch): Likewise.
29147 (ix86_output_indirect_jmp): Likewise.
29148 (ix86_code_end): Call output_indirect_thunk_function if needed.
29149 (ix86_output_call_insn): Call ix86_output_indirect_branch if
29151 (ix86_handle_fndecl_attribute): Handle indirect_branch.
29152 (ix86_attribute_table): Add indirect_branch.
29153 * config/i386/i386.h (machine_function): Add indirect_branch_type
29154 and has_local_indirect_jump.
29155 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
29157 (tablejump): Likewise.
29158 (*indirect_jump): Use ix86_output_indirect_jmp.
29159 (*tablejump_1): Likewise.
29160 (simple_return_indirect_internal): Likewise.
29161 * config/i386/i386.opt (mindirect-branch=): New option.
29162 (indirect_branch): New.
29165 (thunk-inline): Likewise.
29166 (thunk-extern): Likewise.
29167 * doc/extend.texi: Document indirect_branch function attribute.
29168 * doc/invoke.texi: Document -mindirect-branch= option.
29170 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
29173 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
29175 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
29177 * ipa-inline.c (want_inline_small_function_p): Return false if
29178 inlining has already failed with CIF_FINAL_ERROR.
29179 (update_caller_keys): Call want_inline_small_function_p before
29181 (update_callee_keys): Likewise.
29183 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29185 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
29187 (rs6000_quadword_masked_address_p): Likewise.
29188 (quad_aligned_load_p): Likewise.
29189 (quad_aligned_store_p): Likewise.
29190 (const_load_sequence_p): Add comment to describe the outer-most loop.
29191 (mimic_memory_attributes_and_flags): New function.
29192 (rs6000_gen_stvx): Likewise.
29193 (replace_swapped_aligned_store): Likewise.
29194 (rs6000_gen_lvx): Likewise.
29195 (replace_swapped_aligned_load): Likewise.
29196 (replace_swapped_load_constant): Capitalize argument name in
29197 comment describing this function.
29198 (rs6000_analyze_swaps): Add a third pass to search for vector loads
29199 and stores that access quad-word aligned addresses and replace
29200 with stvx or lvx instructions when appropriate.
29201 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
29202 New function prototype.
29203 (rs6000_quadword_masked_address_p): Likewise.
29204 (rs6000_gen_lvx): Likewise.
29205 (rs6000_gen_stvx): Likewise.
29206 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
29207 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
29208 when memory address is aligned.
29209 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
29210 this split to select lvx instruction when memory address is aligned.
29211 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
29212 instruction when memory address is aligned.
29213 (*vsx_le_perm_load_v16qi): Likewise.
29214 (four unnamed splitters): Modify to select the stvx instruction
29215 when memory is aligned.
29217 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
29219 * predict.c (determine_unlikely_bbs): Handle correctly BBs
29220 which appears in the queue multiple times.
29222 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29223 Alan Hayward <alan.hayward@arm.com>
29224 David Sherwood <david.sherwood@arm.com>
29226 * tree-vectorizer.h (vec_lower_bound): New structure.
29227 (_loop_vec_info): Add check_nonzero and lower_bounds.
29228 (LOOP_VINFO_CHECK_NONZERO): New macro.
29229 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
29230 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
29231 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
29232 fields. Make seg_len the distance travelled, not including the
29234 (dr_direction_indicator): Declare.
29235 (dr_zero_step_indicator): Likewise.
29236 (dr_known_forward_stride_p): Likewise.
29237 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
29239 (runtime_alias_check_p): Allow runtime alias checks with
29241 (operator ==): Compare access_size and align.
29242 (prune_runtime_alias_test_list): Rework for new distinction between
29243 the access_size and seg_len.
29244 (create_intersect_range_checks_index): Likewise. Cope with polynomial
29246 (get_segment_min_max): New function.
29247 (create_intersect_range_checks): Use it.
29248 (dr_step_indicator): New function.
29249 (dr_direction_indicator): Likewise.
29250 (dr_zero_step_indicator): Likewise.
29251 (dr_known_forward_stride_p): Likewise.
29252 * tree-loop-distribution.c (data_ref_segment_size): Return
29253 DR_STEP * (niters - 1).
29254 (compute_alias_check_pairs): Update call to the dr_with_seg_len
29256 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
29257 (vect_preserves_scalar_order_p): New function, split out from...
29258 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
29259 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
29260 (vect_vfa_access_size): New function.
29261 (vect_vfa_align): Likewise.
29262 (vect_compile_time_alias): Take access_size_a and access_b arguments.
29263 (dump_lower_bound): New function.
29264 (vect_check_lower_bound): Likewise.
29265 (vect_small_gap_p): Likewise.
29266 (vectorizable_with_step_bound_p): Likewise.
29267 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
29268 depencies if the vectorization factor is 1. Convert the checks
29269 for nonzero steps into checks on the bounds of DR_STEP. Try using
29270 a bunds check for variable steps if the minimum required step is
29271 relatively small. Update calls to the dr_with_seg_len
29272 constructor and to vect_compile_time_alias.
29273 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
29275 (vect_loop_versioning): Call it.
29276 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
29278 (vect_estimate_min_profitable_iters): Account for any bounds checks.
29280 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29281 Alan Hayward <alan.hayward@arm.com>
29282 David Sherwood <david.sherwood@arm.com>
29284 * doc/sourcebuild.texi (vect_scatter_store): Document.
29285 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
29287 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
29289 * genopinit.c (main): Add supports_vec_scatter_store and
29290 supports_vec_scatter_store_cached to target_optabs.
29291 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
29292 IFN_MASK_SCATTER_STORE.
29293 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
29295 * internal-fn.h (internal_store_fn_p): Declare.
29296 (internal_fn_stored_value_index): Likewise.
29297 * internal-fn.c (scatter_store_direct): New macro.
29298 (expand_scatter_store_optab_fn): New function.
29299 (direct_scatter_store_optab_supported_p): New macro.
29300 (internal_store_fn_p): New function.
29301 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
29302 IFN_MASK_SCATTER_STORE.
29303 (internal_fn_mask_index): Likewise.
29304 (internal_fn_stored_value_index): New function.
29305 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
29306 for scatter stores.
29307 * optabs-query.h (supports_vec_scatter_store_p): Declare.
29308 * optabs-query.c (supports_vec_scatter_store_p): New function.
29309 * tree-vectorizer.h (vect_get_store_rhs): Declare.
29310 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
29311 true for scatter stores.
29312 (vect_gather_scatter_fn_p): Handle scatter stores too.
29313 (vect_check_gather_scatter): Consider using scatter stores if
29314 supports_vec_scatter_store_p.
29315 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
29316 scatter stores too.
29317 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
29318 internal_fn_stored_value_index.
29319 (check_load_store_masking): Handle scatter stores too.
29320 (vect_get_store_rhs): Make public.
29321 (vectorizable_call): Use internal_store_fn_p.
29322 (vectorizable_store): Handle scatter store internal functions.
29323 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
29324 when deciding whether the end of the group has been reached.
29325 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
29326 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
29327 (mask_scatter_store<mode>): New insns.
29329 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29330 Alan Hayward <alan.hayward@arm.com>
29331 David Sherwood <david.sherwood@arm.com>
29333 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
29334 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
29335 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
29337 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
29338 Use vect_truncate_gather_scatter_offset if we can't treat the
29339 operation as a normal gather load or scatter store.
29340 (get_group_load_store_type): Take the gather_scatter_info
29341 as argument. Try using a gather load or scatter store for
29342 single-element groups.
29343 (get_load_store_type): Update calls to get_group_load_store_type
29344 and vect_use_strided_gather_scatters_p.
29346 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29347 Alan Hayward <alan.hayward@arm.com>
29348 David Sherwood <david.sherwood@arm.com>
29350 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
29351 optional tree argument.
29352 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
29354 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
29355 but continue to use the current value as a fallback.
29356 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
29357 to compare the updates.
29358 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
29359 (get_load_store_type): Use it when handling a strided access.
29360 (vect_get_strided_load_store_ops): New function.
29361 (vect_get_data_ptr_increment): Likewise.
29362 (vectorizable_load): Handle strided gather loads. Always pass
29363 a step to vect_create_data_ref_ptr and bump_vector_ptr.
29365 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29366 Alan Hayward <alan.hayward@arm.com>
29367 David Sherwood <david.sherwood@arm.com>
29369 * doc/md.texi (gather_load@var{m}): Document.
29370 (mask_gather_load@var{m}): Likewise.
29371 * genopinit.c (main): Add supports_vec_gather_load and
29372 supports_vec_gather_load_cached to target_optabs.
29373 * optabs-tree.c (init_tree_optimization_optabs): Use
29374 ggc_cleared_alloc to allocate target_optabs.
29375 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
29376 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
29378 * internal-fn.h (internal_load_fn_p): Declare.
29379 (internal_gather_scatter_fn_p): Likewise.
29380 (internal_fn_mask_index): Likewise.
29381 (internal_gather_scatter_fn_supported_p): Likewise.
29382 * internal-fn.c (gather_load_direct): New macro.
29383 (expand_gather_load_optab_fn): New function.
29384 (direct_gather_load_optab_supported_p): New macro.
29385 (direct_internal_fn_optab): New function.
29386 (internal_load_fn_p): Likewise.
29387 (internal_gather_scatter_fn_p): Likewise.
29388 (internal_fn_mask_index): Likewise.
29389 (internal_gather_scatter_fn_supported_p): Likewise.
29390 * optabs-query.c (supports_at_least_one_mode_p): New function.
29391 (supports_vec_gather_load_p): Likewise.
29392 * optabs-query.h (supports_vec_gather_load_p): Declare.
29393 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
29394 and memory_type field.
29395 (NUM_PATTERNS): Bump to 15.
29396 * tree-vect-data-refs.c: Include internal-fn.h.
29397 (vect_gather_scatter_fn_p): New function.
29398 (vect_describe_gather_scatter_call): Likewise.
29399 (vect_check_gather_scatter): Try using internal functions for
29400 gather loads. Recognize existing calls to a gather load function.
29401 (vect_analyze_data_refs): Consider using gather loads if
29402 supports_vec_gather_load_p.
29403 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
29404 (vect_get_gather_scatter_offset_type): Likewise.
29405 (vect_convert_mask_for_vectype): Likewise.
29406 (vect_add_conversion_to_patterm): Likewise.
29407 (vect_try_gather_scatter_pattern): Likewise.
29408 (vect_recog_gather_scatter_pattern): New pattern recognizer.
29409 (vect_vect_recog_func_ptrs): Add it.
29410 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
29411 internal_fn_mask_index and internal_gather_scatter_fn_p.
29412 (check_load_store_masking): Take the gather_scatter_info as an
29413 argument and handle gather loads.
29414 (vect_get_gather_scatter_ops): New function.
29415 (vectorizable_call): Check internal_load_fn_p.
29416 (vectorizable_load): Likewise. Handle gather load internal
29418 (vectorizable_store): Update call to check_load_store_masking.
29419 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
29420 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
29421 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
29422 (aarch64_gather_scale_operand_d): New predicates.
29423 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
29424 (mask_gather_load<mode>): New insns.
29426 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29427 Alan Hayward <alan.hayward@arm.com>
29428 David Sherwood <david.sherwood@arm.com>
29430 * optabs.def (fold_left_plus_optab): New optab.
29431 * doc/md.texi (fold_left_plus_@var{m}): Document.
29432 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
29433 * internal-fn.c (fold_left_direct): Define.
29434 (expand_fold_left_optab_fn): Likewise.
29435 (direct_fold_left_optab_supported_p): Likewise.
29436 * fold-const-call.c (fold_const_fold_left): New function.
29437 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
29438 * tree-parloops.c (valid_reduction_p): New function.
29439 (gather_scalar_reductions): Use it.
29440 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
29441 (vect_finish_replace_stmt): Declare.
29442 * tree-vect-loop.c (fold_left_reduction_fn): New function.
29443 (needs_fold_left_reduction_p): New function, split out from...
29444 (vect_is_simple_reduction): ...here. Accept reductions that
29445 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
29446 (vect_force_simple_reduction): Also store the reduction type in
29447 the assignment's STMT_VINFO_REDUC_TYPE.
29448 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
29449 (merge_with_identity): New function.
29450 (vect_expand_fold_left): Likewise.
29451 (vectorize_fold_left_reduction): Likewise.
29452 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
29453 scalar phi in place for it. Check for target support and reject
29454 cases that would reassociate the operation. Defer the transform
29455 phase to vectorize_fold_left_reduction.
29456 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
29457 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
29458 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
29460 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29462 * tree-if-conv.c (predicate_mem_writes): Remove redundant
29463 call to ifc_temp_var.
29465 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29466 Alan Hayward <alan.hayward@arm.com>
29467 David Sherwood <david.sherwood@arm.com>
29469 * target.def (legitimize_address_displacement): Take the original
29470 offset as a poly_int.
29471 * targhooks.h (default_legitimize_address_displacement): Update
29473 * targhooks.c (default_legitimize_address_displacement): Likewise.
29474 * doc/tm.texi: Regenerate.
29475 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
29476 as an argument, moving assert of ad->disp == ad->disp_term to...
29477 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
29478 Try calling targetm.legitimize_address_displacement before expanding
29479 the address rather than afterwards, and adjust for the new interface.
29480 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
29481 Match the new hook interface. Handle SVE addresses.
29482 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
29483 new hook interface.
29485 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29487 * Makefile.in (OBJS): Add early-remat.o.
29488 * target.def (select_early_remat_modes): New hook.
29489 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
29490 * doc/tm.texi: Regenerate.
29491 * targhooks.h (default_select_early_remat_modes): Declare.
29492 * targhooks.c (default_select_early_remat_modes): New function.
29493 * timevar.def (TV_EARLY_REMAT): New timevar.
29494 * passes.def (pass_early_remat): New pass.
29495 * tree-pass.h (make_pass_early_remat): Declare.
29496 * early-remat.c: New file.
29497 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
29499 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
29501 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29502 Alan Hayward <alan.hayward@arm.com>
29503 David Sherwood <david.sherwood@arm.com>
29505 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
29506 vfm1 with a bound_epilog parameter.
29507 (vect_do_peeling): Update calls accordingly, and move the prologue
29508 call earlier in the function. Treat the base bound_epilog as 0 for
29509 fully-masked loops and retain vf - 1 for other loops. Add 1 to
29510 this base when peeling for gaps.
29511 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
29512 with fully-masked loops.
29513 (vect_estimate_min_profitable_iters): Handle the single peeled
29514 iteration in that case.
29516 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29517 Alan Hayward <alan.hayward@arm.com>
29518 David Sherwood <david.sherwood@arm.com>
29520 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
29521 single-element interleaving even if the size is not a power of 2.
29522 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
29523 accesses for single-element interleaving if the group size is
29526 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29527 Alan Hayward <alan.hayward@arm.com>
29528 David Sherwood <david.sherwood@arm.com>
29530 * doc/md.texi (fold_extract_last_@var{m}): Document.
29531 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
29532 * optabs.def (fold_extract_last_optab): New optab.
29533 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
29534 * internal-fn.c (fold_extract_direct): New macro.
29535 (expand_fold_extract_optab_fn): Likewise.
29536 (direct_fold_extract_optab_supported_p): Likewise.
29537 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
29538 * tree-vect-loop.c (vect_model_reduction_cost): Handle
29539 EXTRACT_LAST_REDUCTION.
29540 (get_initial_def_for_reduction): Do not create an initial vector
29541 for EXTRACT_LAST_REDUCTION reductions.
29542 (vectorizable_reduction): Leave the scalar phi in place for
29543 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
29544 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
29545 epilogue code for EXTRACT_LAST_REDUCTION and defer the
29546 transform phase to vectorizable_condition.
29547 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
29549 (vect_finish_stmt_generation): ...here.
29550 (vect_finish_replace_stmt): New function.
29551 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
29552 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
29554 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
29556 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29557 Alan Hayward <alan.hayward@arm.com>
29558 David Sherwood <david.sherwood@arm.com>
29560 * doc/md.texi (extract_last_@var{m}): Document.
29561 * optabs.def (extract_last_optab): New optab.
29562 * internal-fn.def (EXTRACT_LAST): New internal function.
29563 * internal-fn.c (cond_unary_direct): New macro.
29564 (expand_cond_unary_optab_fn): Likewise.
29565 (direct_cond_unary_optab_supported_p): Likewise.
29566 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
29567 loops using EXTRACT_LAST.
29568 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
29569 (extract_last_<mode>): ...this optab.
29570 (vec_extract<mode><Vel>): Update accordingly.
29572 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29573 Alan Hayward <alan.hayward@arm.com>
29574 David Sherwood <david.sherwood@arm.com>
29576 * target.def (empty_mask_is_expensive): New hook.
29577 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
29578 * doc/tm.texi: Regenerate.
29579 * targhooks.h (default_empty_mask_is_expensive): Declare.
29580 * targhooks.c (default_empty_mask_is_expensive): New function.
29581 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
29582 if the target says that empty masks are expensive.
29583 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
29585 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
29587 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29588 Alan Hayward <alan.hayward@arm.com>
29589 David Sherwood <david.sherwood@arm.com>
29591 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
29592 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
29593 (vect_use_loop_mask_for_alignment_p): New function.
29594 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
29595 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
29596 niters_skip argument. Make sure that the first niters_skip elements
29597 of the first iteration are inactive.
29598 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
29599 Update call to vect_set_loop_masks_directly.
29600 (get_misalign_in_elems): New function, split out from...
29601 (vect_gen_prolog_loop_niters): ...here.
29602 (vect_update_init_of_dr): Take a code argument that specifies whether
29603 the adjustment should be added or subtracted.
29604 (vect_update_init_of_drs): Likewise.
29605 (vect_prepare_for_masked_peels): New function.
29606 (vect_do_peeling): Skip prologue peeling if we're using a mask
29607 instead. Update call to vect_update_inits_of_drs.
29608 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
29610 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
29611 alignment. Do not include the number of peeled iterations in
29612 the minimum threshold in that case.
29613 (vectorizable_induction): Adjust the start value down by
29614 LOOP_VINFO_MASK_SKIP_NITERS iterations.
29615 (vect_transform_loop): Call vect_prepare_for_masked_peels.
29616 Take the number of skipped iterations into account when calculating
29618 * tree-vect-stmts.c (vect_gen_while_not): New function.
29620 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29621 Alan Hayward <alan.hayward@arm.com>
29622 David Sherwood <david.sherwood@arm.com>
29624 * doc/sourcebuild.texi (vect_fully_masked): Document.
29625 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
29626 default value to 0.
29627 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
29629 (vect_analyze_loop_2): ...here. Don't check the vectorization
29630 factor against the number of loop iterations if the loop is
29633 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29634 Alan Hayward <alan.hayward@arm.com>
29635 David Sherwood <david.sherwood@arm.com>
29637 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
29638 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
29639 (dump_groups): Update accordingly.
29640 (iv_use::mem_type): New member variable.
29641 (address_p): New function.
29642 (record_use): Add a mem_type argument and initialize the new
29644 (record_group_use): Add a mem_type argument. Use address_p.
29645 Remove obsolete null checks of base_object. Update call to record_use.
29646 (find_interesting_uses_op): Update call to record_group_use.
29647 (find_interesting_uses_cond): Likewise.
29648 (find_interesting_uses_address): Likewise.
29649 (get_mem_type_for_internal_fn): New function.
29650 (find_address_like_use): Likewise.
29651 (find_interesting_uses_stmt): Try find_address_like_use before
29652 calling find_interesting_uses_op.
29653 (addr_offset_valid_p): Use the iv mem_type field as the type
29654 of the addressed memory.
29655 (add_autoinc_candidates): Likewise.
29656 (get_address_cost): Likewise.
29657 (split_small_address_groups_p): Use address_p.
29658 (split_address_groups): Likewise.
29659 (add_iv_candidate_for_use): Likewise.
29660 (autoinc_possible_for_pair): Likewise.
29661 (rewrite_groups): Likewise.
29662 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
29663 (determine_group_iv_cost): Update after split of USE_ADDRESS.
29664 (get_alias_ptr_type_for_ptr_address): New function.
29665 (rewrite_use_address): Rewrite address uses in calls that were
29666 identified by find_address_like_use.
29668 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29669 Alan Hayward <alan.hayward@arm.com>
29670 David Sherwood <david.sherwood@arm.com>
29672 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
29674 * gimple-expr.h (is_gimple_addressable: Likewise.
29675 * gimple-expr.c (is_gimple_address): Likewise.
29676 * internal-fn.c (expand_call_mem_ref): New function.
29677 (expand_mask_load_optab_fn): Use it.
29678 (expand_mask_store_optab_fn): Likewise.
29680 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29681 Alan Hayward <alan.hayward@arm.com>
29682 David Sherwood <david.sherwood@arm.com>
29684 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
29685 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
29686 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
29687 (cond_umax@var{mode}): Document.
29688 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
29689 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
29690 (cond_umin_optab, cond_umax_optab): New optabs.
29691 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
29692 (COND_IOR, COND_XOR): New internal functions.
29693 * internal-fn.h (get_conditional_internal_fn): Declare.
29694 * internal-fn.c (cond_binary_direct): New macro.
29695 (expand_cond_binary_optab_fn): Likewise.
29696 (direct_cond_binary_optab_supported_p): Likewise.
29697 (get_conditional_internal_fn): New function.
29698 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
29699 Cope with reduction statements that are vectorized as calls rather
29701 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
29702 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
29703 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
29704 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
29705 (UNSPEC_COND_EOR): New unspecs.
29706 (optab): Add mappings for them.
29707 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
29708 (sve_int_op, sve_fp_op): New int attributes.
29710 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29711 Alan Hayward <alan.hayward@arm.com>
29712 David Sherwood <david.sherwood@arm.com>
29714 * optabs.def (while_ult_optab): New optab.
29715 * doc/md.texi (while_ult@var{m}@var{n}): Document.
29716 * internal-fn.def (WHILE_ULT): New internal function.
29717 * internal-fn.h (direct_internal_fn_supported_p): New override
29718 that takes two types as argument.
29719 * internal-fn.c (while_direct): New macro.
29720 (expand_while_optab_fn): New function.
29721 (convert_optab_supported_p): Likewise.
29722 (direct_while_optab_supported_p): New macro.
29723 * wide-int.h (wi::udiv_ceil): New function.
29724 * tree-vectorizer.h (rgroup_masks): New structure.
29725 (vec_loop_masks): New typedef.
29726 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
29727 and fully_masked_p.
29728 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
29729 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
29730 (vect_max_vf): New function.
29731 (slpeel_make_loop_iterate_ntimes): Delete.
29732 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
29733 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
29734 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
29735 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
29736 internal-fn.h, stor-layout.h and optabs-query.h.
29737 (vect_set_loop_mask): New function.
29738 (add_preheader_seq): Likewise.
29739 (add_header_seq): Likewise.
29740 (interleave_supported_p): Likewise.
29741 (vect_maybe_permute_loop_masks): Likewise.
29742 (vect_set_loop_masks_directly): Likewise.
29743 (vect_set_loop_condition_masked): Likewise.
29744 (vect_set_loop_condition_unmasked): New function, split out from
29745 slpeel_make_loop_iterate_ntimes.
29746 (slpeel_make_loop_iterate_ntimes): Rename to..
29747 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
29748 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
29749 (vect_do_peeling): Update call accordingly.
29750 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
29752 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
29753 mask_compare_type, can_fully_mask_p and fully_masked_p.
29754 (release_vec_loop_masks): New function.
29755 (_loop_vec_info): Use it to free the loop masks.
29756 (can_produce_all_loop_masks_p): New function.
29757 (vect_get_max_nscalars_per_iter): Likewise.
29758 (vect_verify_full_masking): Likewise.
29759 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
29760 retries, and free the mask rgroups before retrying. Check loop-wide
29761 reasons for disallowing fully-masked loops. Make the final decision
29762 about whether use a fully-masked loop or not.
29763 (vect_estimate_min_profitable_iters): Do not assume that peeling
29764 for the number of iterations will be needed for fully-masked loops.
29765 (vectorizable_reduction): Disable fully-masked loops.
29766 (vectorizable_live_operation): Likewise.
29767 (vect_halve_mask_nunits): New function.
29768 (vect_double_mask_nunits): Likewise.
29769 (vect_record_loop_mask): Likewise.
29770 (vect_get_loop_mask): Likewise.
29771 (vect_transform_loop): Handle the case in which the final loop
29772 iteration might handle a partial vector. Call vect_set_loop_condition
29773 instead of slpeel_make_loop_iterate_ntimes.
29774 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
29775 (check_load_store_masking): New function.
29776 (prepare_load_store_mask): Likewise.
29777 (vectorizable_store): Handle fully-masked loops.
29778 (vectorizable_load): Likewise.
29779 (supportable_widening_operation): Use vect_halve_mask_nunits for
29781 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
29782 (vect_gen_while): New function.
29783 * config/aarch64/aarch64.md (umax<mode>3): New expander.
29784 (aarch64_uqdec<mode>): New insn.
29786 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29787 Alan Hayward <alan.hayward@arm.com>
29788 David Sherwood <david.sherwood@arm.com>
29790 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
29791 (reduc_xor_scal_optab): New optabs.
29792 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
29793 (reduc_xor_scal_@var{m}): Document.
29794 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
29795 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
29796 internal functions.
29797 * fold-const-call.c (fold_const_call): Handle them.
29798 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
29799 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
29800 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
29801 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
29802 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
29803 (UNSPEC_XORV): New unspecs.
29804 (optab): Add entries for them.
29805 (BITWISEV): New int iterator.
29806 (bit_reduc_op): New int attributes.
29808 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29809 Alan Hayward <alan.hayward@arm.com>
29810 David Sherwood <david.sherwood@arm.com>
29812 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
29813 * internal-fn.def (VEC_SHL_INSERT): New internal function.
29814 * optabs.def (vec_shl_insert_optab): New optab.
29815 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
29816 (duplicate_and_interleave): Likewise.
29817 * tree-vect-loop.c: Include internal-fn.h.
29818 (neutral_op_for_slp_reduction): New function, split out from
29819 get_initial_defs_for_reduction.
29820 (get_initial_def_for_reduction): Handle option 2 for variable-length
29821 vectors by loading the neutral value into a vector and then shifting
29822 the initial value into element 0.
29823 (get_initial_defs_for_reduction): Replace the code argument with
29824 the neutral value calculated by neutral_op_for_slp_reduction.
29825 Use gimple_build_vector for constant-length vectors.
29826 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
29827 but the first group_size elements have a neutral value.
29828 Use duplicate_and_interleave otherwise.
29829 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
29830 Update call to get_initial_defs_for_reduction. Handle SLP
29831 reductions for variable-length vectors by creating one vector
29832 result for each scalar result, with the elements associated
29833 with other scalar results stubbed out with the neutral value.
29834 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
29835 Require IFN_VEC_SHL_INSERT for double reductions on
29836 variable-length vectors, or SLP reductions that have
29837 a neutral value. Require can_duplicate_and_interleave_p
29838 support for variable-length unchained SLP reductions if there
29839 is no neutral value, such as for MIN/MAX reductions. Also require
29840 the number of vector elements to be a multiple of the number of
29841 SLP statements when doing variable-length unchained SLP reductions.
29842 Update call to vect_create_epilog_for_reduction.
29843 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
29844 and remove initial values.
29845 (duplicate_and_interleave): Make public.
29846 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
29847 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
29849 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29850 Alan Hayward <alan.hayward@arm.com>
29851 David Sherwood <david.sherwood@arm.com>
29853 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
29854 (can_duplicate_and_interleave_p): New function.
29855 (vect_get_and_check_slp_defs): Take the vector of statements
29856 rather than just the current one. Remove excess parentheses.
29857 Restriction rejectinon of vect_constant_def and vect_external_def
29858 for variable-length vectors to boolean types, or types for which
29859 can_duplicate_and_interleave_p is false.
29860 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
29861 (duplicate_and_interleave): New function.
29862 (vect_get_constant_vectors): Use gimple_build_vector for
29863 constant-length vectors and suitable variable-length constant
29864 vectors. Use duplicate_and_interleave for other variable-length
29865 vectors. Don't defer the update when inserting new statements.
29867 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29868 Alan Hayward <alan.hayward@arm.com>
29869 David Sherwood <david.sherwood@arm.com>
29871 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
29872 min_profitable_iters doesn't go negative.
29874 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29875 Alan Hayward <alan.hayward@arm.com>
29876 David Sherwood <david.sherwood@arm.com>
29878 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
29879 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
29880 * optabs.def (vec_mask_load_lanes_optab): New optab.
29881 (vec_mask_store_lanes_optab): Likewise.
29882 * internal-fn.def (MASK_LOAD_LANES): New internal function.
29883 (MASK_STORE_LANES): Likewise.
29884 * internal-fn.c (mask_load_lanes_direct): New macro.
29885 (mask_store_lanes_direct): Likewise.
29886 (expand_mask_load_optab_fn): Handle masked operations.
29887 (expand_mask_load_lanes_optab_fn): New macro.
29888 (expand_mask_store_optab_fn): Handle masked operations.
29889 (expand_mask_store_lanes_optab_fn): New macro.
29890 (direct_mask_load_lanes_optab_supported_p): Likewise.
29891 (direct_mask_store_lanes_optab_supported_p): Likewise.
29892 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
29894 (vect_load_lanes_supported): Likewise.
29895 * tree-vect-data-refs.c (strip_conversion): New function.
29896 (can_group_stmts_p): Likewise.
29897 (vect_analyze_data_ref_accesses): Use it instead of checking
29898 for a pair of assignments.
29899 (vect_store_lanes_supported): Take a masked_p parameter.
29900 (vect_load_lanes_supported): Likewise.
29901 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
29902 vect_store_lanes_supported and vect_load_lanes_supported.
29903 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
29904 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
29905 parameter. Don't allow gaps for masked accesses.
29906 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
29907 and vect_load_lanes_supported.
29908 (get_load_store_type): Take a masked_p parameter and update
29909 call to get_group_load_store_type.
29910 (vectorizable_store): Update call to get_load_store_type.
29911 Handle IFN_MASK_STORE_LANES.
29912 (vectorizable_load): Update call to get_load_store_type.
29913 Handle IFN_MASK_LOAD_LANES.
29915 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29916 Alan Hayward <alan.hayward@arm.com>
29917 David Sherwood <david.sherwood@arm.com>
29919 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
29921 * config/aarch64/aarch64-protos.h
29922 (aarch64_sve_struct_memory_operand_p): Declare.
29923 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
29924 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
29925 (VPRED, vpred): Handle SVE structure modes.
29926 * config/aarch64/constraints.md (Utx): New constraint.
29927 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
29928 (aarch64_sve_struct_nonimmediate_operand): New predicates.
29929 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
29930 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
29931 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
29932 structure modes. Split into pieces after RA.
29933 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
29934 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
29936 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
29937 SVE structure modes.
29938 (aarch64_classify_address): Likewise.
29939 (sizetochar): Move earlier in file.
29940 (aarch64_print_operand): Handle SVE register lists.
29941 (aarch64_array_mode): New function.
29942 (aarch64_sve_struct_memory_operand_p): Likewise.
29943 (TARGET_ARRAY_MODE): Redefine.
29945 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29946 Alan Hayward <alan.hayward@arm.com>
29947 David Sherwood <david.sherwood@arm.com>
29949 * target.def (array_mode): New target hook.
29950 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
29951 * doc/tm.texi: Regenerate.
29952 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
29953 * hooks.c (hook_optmode_mode_uhwi_none): New function.
29954 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
29955 targetm.array_mode.
29956 * stor-layout.c (mode_for_array): Likewise. Support polynomial
29959 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29960 Alan Hayward <alan.hayward@arm.com>
29961 David Sherwood <david.sherwood@arm.com>
29963 * fold-const.c (fold_binary_loc): Check the argument types
29964 rather than the result type when testing for a vector operation.
29966 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29968 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
29969 * doc/tm.texi: Regenerate.
29971 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29972 Alan Hayward <alan.hayward@arm.com>
29973 David Sherwood <david.sherwood@arm.com>
29975 * doc/invoke.texi (-msve-vector-bits=): Document new option.
29976 (sve): Document new AArch64 extension.
29977 * doc/md.texi (w): Extend the description of the AArch64
29978 constraint to include SVE vectors.
29979 (Upl, Upa): Document new AArch64 predicate constraints.
29980 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
29982 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
29983 (msve-vector-bits=): New option.
29984 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
29985 SVE when these are disabled.
29986 (sve): New extension.
29987 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
29988 modes. Adjust their number of units based on aarch64_sve_vg.
29989 (MAX_BITSIZE_MODE_ANY_MODE): Define.
29990 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
29991 aarch64_addr_query_type.
29992 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
29993 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
29994 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
29995 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
29996 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
29997 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
29998 (aarch64_simd_imm_zero_p): Delete.
29999 (aarch64_check_zero_based_sve_index_immediate): Declare.
30000 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
30001 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
30002 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
30003 (aarch64_sve_float_mul_immediate_p): Likewise.
30004 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
30005 rather than an rtx.
30006 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
30007 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
30008 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
30009 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
30010 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
30011 (aarch64_regmode_natural_size): Likewise.
30012 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
30013 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
30015 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
30016 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
30017 for VG and the SVE predicate registers.
30018 (V_ALIASES): Add a "z"-prefixed alias.
30019 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
30020 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
30021 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
30022 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
30023 (REG_CLASS_NAMES): Add entries for them.
30024 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
30025 and the predicate registers.
30026 (aarch64_sve_vg): Declare.
30027 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
30028 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
30029 (REGMODE_NATURAL_SIZE): Define.
30030 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
30032 * config/aarch64/aarch64.c: Include cfgrtl.h.
30033 (simd_immediate_info): Add a constructor for series vectors,
30034 and an associated step field.
30035 (aarch64_sve_vg): New variable.
30036 (aarch64_dbx_register_number): Handle VG and the predicate registers.
30037 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
30038 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
30039 (VEC_ANY_DATA, VEC_STRUCT): New constants.
30040 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
30041 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
30042 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
30043 (aarch64_get_mask_mode): New functions.
30044 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
30045 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
30046 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
30047 predicate modes and predicate registers. Explicitly restrict
30048 GPRs to modes of 16 bytes or smaller. Only allow FP registers
30049 to store a vector mode if it is recognized by
30050 aarch64_classify_vector_mode.
30051 (aarch64_regmode_natural_size): New function.
30052 (aarch64_hard_regno_caller_save_mode): Return the original mode
30054 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
30055 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
30056 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
30057 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
30059 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
30060 does not overlap dest if the function is frame-related. Handle
30062 (aarch64_split_add_offset): New function.
30063 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
30064 them aarch64_add_offset.
30065 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
30066 and update call to aarch64_sub_sp.
30067 (aarch64_add_cfa_expression): New function.
30068 (aarch64_expand_prologue): Pass extra temporary registers to the
30069 functions above. Handle the case in which we need to emit new
30070 DW_CFA_expressions for registers that were originally saved
30071 relative to the stack pointer, but now have to be expressed
30072 relative to the frame pointer.
30073 (aarch64_output_mi_thunk): Pass extra temporary registers to the
30075 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
30076 IP0 and IP1 values for SVE frames.
30077 (aarch64_expand_vec_series): New function.
30078 (aarch64_expand_sve_widened_duplicate): Likewise.
30079 (aarch64_expand_sve_const_vector): Likewise.
30080 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
30081 Handle SVE constants. Use emit_move_insn to move a force_const_mem
30082 into the register, rather than emitting a SET directly.
30083 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
30084 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
30085 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
30086 (offset_9bit_signed_scaled_p): New functions.
30087 (aarch64_replicate_bitmask_imm): New function.
30088 (aarch64_bitmask_imm): Use it.
30089 (aarch64_cannot_force_const_mem): Reject expressions involving
30090 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
30091 (aarch64_classify_index): Handle SVE indices, by requiring
30092 a plain register index with a scale that matches the element size.
30093 (aarch64_classify_address): Handle SVE addresses. Assert that
30094 the mode of the address is VOIDmode or an integer mode.
30095 Update call to aarch64_classify_symbol.
30096 (aarch64_classify_symbolic_expression): Update call to
30097 aarch64_classify_symbol.
30098 (aarch64_const_vec_all_in_range_p): New function.
30099 (aarch64_print_vector_float_operand): Likewise.
30100 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
30101 "vN" for FP registers with SVE modes. Handle (const ...) vectors
30102 and the FP immediates 1.0 and 0.5.
30103 (aarch64_print_address_internal): Handle SVE addresses.
30104 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
30105 (aarch64_regno_regclass): Handle predicate registers.
30106 (aarch64_secondary_reload): Handle big-endian reloads of SVE
30108 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
30109 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
30110 (aarch64_convert_sve_vector_bits): New function.
30111 (aarch64_override_options): Use it to handle -msve-vector-bits=.
30112 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
30113 rather than an rtx.
30114 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
30115 Handle SVE vector and predicate modes. Accept VL-based constants
30116 that need only one temporary register, and VL offsets that require
30117 no temporary registers.
30118 (aarch64_conditional_register_usage): Mark the predicate registers
30119 as fixed if SVE isn't available.
30120 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
30121 Return true for SVE vector and predicate modes.
30122 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
30123 rather than an unsigned int. Handle SVE modes.
30124 (aarch64_preferred_simd_mode): Update call accordingly. Handle
30126 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
30128 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
30129 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
30130 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
30131 (aarch64_sve_float_mul_immediate_p): New functions.
30132 (aarch64_sve_valid_immediate): New function.
30133 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
30134 Explicitly reject structure modes. Check for INDEX constants.
30135 Handle PTRUE and PFALSE constants.
30136 (aarch64_check_zero_based_sve_index_immediate): New function.
30137 (aarch64_simd_imm_zero_p): Delete.
30138 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
30139 vector modes. Accept constants in the range of CNT[BHWD].
30140 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
30141 ask for an Advanced SIMD mode.
30142 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
30143 (aarch64_simd_vector_alignment): Handle SVE predicates.
30144 (aarch64_vectorize_preferred_vector_alignment): New function.
30145 (aarch64_simd_vector_alignment_reachable): Use it instead of
30147 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
30148 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
30150 (MAX_VECT_LEN): Delete.
30151 (expand_vec_perm_d): Add a vec_flags field.
30152 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
30153 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
30154 (aarch64_evpc_ext): Don't apply a big-endian lane correction
30156 (aarch64_evpc_rev): Rename to...
30157 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
30158 (aarch64_evpc_rev_global): New function.
30159 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
30160 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
30162 (aarch64_evpc_sve_tbl): New function.
30163 (aarch64_expand_vec_perm_const_1): Update after rename of
30164 aarch64_evpc_rev. Handle SVE permutes too, trying
30165 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
30166 than aarch64_evpc_tbl.
30167 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
30168 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
30169 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
30170 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
30171 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
30172 (aarch64_expand_sve_vcond): New functions.
30173 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
30174 of aarch64_vector_mode_p.
30175 (aarch64_dwarf_poly_indeterminate_value): New function.
30176 (aarch64_compute_pressure_classes): Likewise.
30177 (aarch64_can_change_mode_class): Likewise.
30178 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
30179 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
30180 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
30181 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
30182 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
30183 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
30184 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
30185 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
30187 (Dn, Dl, Dr): Accept const as well as const_vector.
30188 (Dz): Likewise. Compare against CONST0_RTX.
30189 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
30190 of "vector" where appropriate.
30191 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
30192 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
30193 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
30194 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
30195 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
30196 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
30197 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
30198 (v_int_equiv): Extend to SVE modes.
30199 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
30201 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
30202 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
30203 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
30204 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
30205 (SVE_COND_FP_CMP): New int iterators.
30206 (perm_hilo): Handle the new unpack unspecs.
30207 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
30209 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
30210 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
30211 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
30212 (aarch64_equality_operator, aarch64_constant_vector_operand)
30213 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
30214 (aarch64_sve_nonimmediate_operand): Likewise.
30215 (aarch64_sve_general_operand): Likewise.
30216 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
30217 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
30218 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
30219 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
30220 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
30221 (aarch64_sve_float_arith_immediate): Likewise.
30222 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
30223 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
30224 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
30225 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
30226 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
30227 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
30228 (aarch64_sve_float_arith_operand): Likewise.
30229 (aarch64_sve_float_arith_with_sub_operand): Likewise.
30230 (aarch64_sve_float_mul_operand): Likewise.
30231 (aarch64_sve_vec_perm_operand): Likewise.
30232 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
30233 (aarch64_mov_operand): Accept const_poly_int and const_vector.
30234 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
30235 as well as const_vector.
30236 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
30237 in file. Use CONST0_RTX and CONSTM1_RTX.
30238 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
30239 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
30240 Use aarch64_simd_imm_zero.
30241 * config/aarch64/aarch64-sve.md: New file.
30242 * config/aarch64/aarch64.md: Include it.
30243 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
30244 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
30245 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
30246 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
30247 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
30248 (sve): New attribute.
30249 (enabled): Disable instructions with the sve attribute unless
30251 (movqi, movhi): Pass CONST_POLY_INT operaneds through
30252 aarch64_expand_mov_immediate.
30253 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
30254 CNT[BHSD] immediates.
30255 (movti): Split CONST_POLY_INT moves into two halves.
30256 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
30257 Split additions that need a temporary here if the destination
30258 is the stack pointer.
30259 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
30260 (*add<mode>3_poly_1): New instruction.
30261 (set_clobber_cc): New expander.
30263 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30265 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
30266 parameter and use it instead of GET_MODE_SIZE (innermode). Use
30267 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
30268 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
30269 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
30270 Change innermode from fixed_mode_size to machine_mode.
30271 (simplify_subreg): Update call accordingly. Handle a constant-sized
30272 subreg of a variable-length CONST_VECTOR.
30274 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30275 Alan Hayward <alan.hayward@arm.com>
30276 David Sherwood <david.sherwood@arm.com>
30278 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
30279 (add_offset_to_base): New function, split out from...
30280 (create_mem_ref): ...here. When handling a scale other than 1,
30281 check first whether the address is valid without the offset.
30282 Add it into the base if so, leaving the index and scale as-is.
30284 2018-01-12 Jakub Jelinek <jakub@redhat.com>
30287 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
30288 fold_for_warn before checking if arg2 is INTEGER_CST.
30290 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
30292 * config/rs6000/predicates.md (load_multiple_operation): Delete.
30293 (store_multiple_operation): Delete.
30294 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
30295 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
30296 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
30297 guarded by TARGET_STRING.
30298 (rs6000_output_load_multiple): Delete.
30299 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
30300 OPTION_MASK_STRING / TARGET_STRING handling.
30301 (print_operand) <'N', 'O'>: Add comment that these are unused now.
30302 (const rs6000_opt_masks) <"string">: Change mask to 0.
30303 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
30304 (MASK_STRING): Delete.
30305 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
30307 (load_multiple): Delete.
30314 (store_multiple): Delete.
30321 (movmemsi_8reg): Delete.
30322 (corresponding unnamed define_insn): Delete.
30323 (movmemsi_6reg): Delete.
30324 (corresponding unnamed define_insn): Delete.
30325 (movmemsi_4reg): Delete.
30326 (corresponding unnamed define_insn): Delete.
30327 (movmemsi_2reg): Delete.
30328 (corresponding unnamed define_insn): Delete.
30329 (movmemsi_1reg): Delete.
30330 (corresponding unnamed define_insn): Delete.
30331 * config/rs6000/rs6000.opt (mno-string): New.
30332 (mstring): Replace by deprecation warning stub.
30333 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
30335 2018-01-12 Jakub Jelinek <jakub@redhat.com>
30337 * regrename.c (regrename_do_replace): If replacing the same
30338 reg multiple times, try to reuse last created gen_raw_REG.
30341 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
30342 main to workaround a bug in GDB.
30344 2018-01-12 Tom de Vries <tom@codesourcery.com>
30347 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
30349 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
30351 PR rtl-optimization/80481
30352 * ira-color.c (get_cap_member): New function.
30353 (allocnos_conflict_by_live_ranges_p): Use it.
30354 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
30355 (setup_slot_coalesced_allocno_live_ranges): Ditto.
30357 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
30360 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
30361 (*saddl_se_1): Ditto.
30362 (*ssubsi_1): Ditto.
30363 (*ssubl_se_1): Ditto.
30365 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
30367 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
30368 rather than wi::to_widest for DR_INITs.
30369 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
30370 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
30371 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
30373 (vect_analyze_group_access_1): Note that here.
30375 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
30377 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
30378 polynomial type sizes.
30380 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
30382 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
30383 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
30384 (gimple_add_tmp_var): Likewise.
30386 2018-01-12 Martin Liska <mliska@suse.cz>
30388 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
30389 (gimple_alloc_sizes): Likewise.
30390 (dump_gimple_statistics): Use PRIu64 in printf format.
30391 * gimple.h: Change uint64_t to int.
30393 2018-01-12 Martin Liska <mliska@suse.cz>
30395 * tree-core.h: Use uint64_t instead of int.
30396 * tree.c (tree_node_counts): Likewise.
30397 (tree_node_sizes): Likewise.
30398 (dump_tree_statistics): Use PRIu64 in printf format.
30400 2018-01-12 Martin Liska <mliska@suse.cz>
30402 * Makefile.in: As qsort_chk is implemented in vec.c, add
30403 vec.o to linkage of gencfn-macros.
30404 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
30405 passing the info to record_node_allocation_statistics.
30406 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
30408 * ggc-common.c (struct ggc_usage): Add operator== and use
30409 it in operator< and compare function.
30410 * mem-stats.h (struct mem_usage): Likewise.
30411 * vec.c (struct vec_usage): Remove operator< and compare
30412 function. Can be simply inherited.
30414 2018-01-12 Martin Jambor <mjambor@suse.cz>
30417 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
30418 * tree-ssa-math-opts.c: Include domwalk.h.
30419 (convert_mult_to_fma_1): New function.
30420 (fma_transformation_info): New type.
30421 (fma_deferring_state): Likewise.
30422 (cancel_fma_deferring): New function.
30423 (result_of_phi): Likewise.
30424 (last_fma_candidate_feeds_initial_phi): Likewise.
30425 (convert_mult_to_fma): Added deferring logic, split actual
30426 transformation to convert_mult_to_fma_1.
30427 (math_opts_dom_walker): New type.
30428 (math_opts_dom_walker::after_dom_children): New method, body moved
30429 here from pass_optimize_widening_mul::execute, added deferring logic
30431 (pass_optimize_widening_mul::execute): Moved most of code to
30432 math_opts_dom_walker::after_dom_children.
30433 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
30434 * config/i386/i386.c (ix86_option_override_internal): Added
30435 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
30437 2018-01-12 Richard Biener <rguenther@suse.de>
30440 * dwarf2out.c (gen_variable_die): Do not reset old_die for
30441 inline instance vars.
30443 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
30446 * config/rx/rx.c (rx_is_restricted_memory_address):
30447 Handle SUBREG case.
30449 2018-01-12 Richard Biener <rguenther@suse.de>
30451 PR tree-optimization/80846
30452 * target.def (split_reduction): New target hook.
30453 * targhooks.c (default_split_reduction): New function.
30454 * targhooks.h (default_split_reduction): Declare.
30455 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
30456 target requests first reduce vectors by combining low and high
30458 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
30459 (get_vectype_for_scalar_type_and_size): Export.
30460 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
30461 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
30462 * doc/tm.texi: Regenerate.
30463 * config/i386/i386.c (ix86_split_reduction): Implement
30464 TARGET_VECTORIZE_SPLIT_REDUCTION.
30466 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
30469 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
30470 in PIC mode except for TARGET_VXWORKS_RTP.
30471 * config/sparc/sparc.c: Include cfgrtl.h.
30472 (TARGET_INIT_PIC_REG): Define.
30473 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
30474 (sparc_pic_register_p): New predicate.
30475 (sparc_legitimate_address_p): Use it.
30476 (sparc_legitimize_pic_address): Likewise.
30477 (sparc_delegitimize_address): Likewise.
30478 (sparc_mode_dependent_address_p): Likewise.
30479 (gen_load_pcrel_sym): Remove 4th parameter.
30480 (load_got_register): Adjust call to above. Remove obsolete stuff.
30481 (sparc_expand_prologue): Do not call load_got_register here.
30482 (sparc_flat_expand_prologue): Likewise.
30483 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
30484 (sparc_use_pseudo_pic_reg): New function.
30485 (sparc_init_pic_reg): Likewise.
30486 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
30487 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
30489 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
30491 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
30492 Add item for branch_cost.
30494 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
30496 PR rtl-optimization/83565
30497 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
30498 not extend the result to a larger mode for rotate operations.
30499 (num_sign_bit_copies1): Likewise.
30501 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30504 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
30506 Use values-Xc.o for -pedantic.
30507 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
30509 2018-01-12 Martin Liska <mliska@suse.cz>
30512 * ipa-devirt.c (final_warning_record::grow_type_warnings):
30514 (possible_polymorphic_call_targets): Use it.
30515 (ipa_devirt): Likewise.
30517 2018-01-12 Martin Liska <mliska@suse.cz>
30519 * profile-count.h (enum profile_quality): Use 0 as invalid
30520 enum value of profile_quality.
30522 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
30524 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
30525 -mext-string options.
30527 2018-01-12 Richard Biener <rguenther@suse.de>
30529 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
30530 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
30531 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
30533 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
30535 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30537 * configure.ac (--with-long-double-format): Add support for the
30538 configuration option to change the default long double format on
30540 * config.gcc (powerpc*-linux*-*): Likewise.
30541 * configure: Regenerate.
30542 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
30543 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
30544 used without modification.
30546 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30548 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
30549 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
30550 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
30551 MISC_BUILTIN_SPEC_BARRIER.
30552 (rs6000_init_builtins): Likewise.
30553 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
30555 (speculation_barrier): New define_insn.
30556 * doc/extend.texi: Document __builtin_speculation_barrier.
30558 2018-01-11 Jakub Jelinek <jakub@redhat.com>
30561 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
30562 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
30563 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
30565 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
30566 integral modes instead of "ss" and "sd".
30567 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
30568 vectors with 32-bit and 64-bit elements.
30569 (vecdupssescalarmodesuffix): New mode attribute.
30570 (vec_dup<mode>): Use it.
30572 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
30575 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
30576 frame if argument is passed on stack.
30578 2018-01-11 Jakub Jelinek <jakub@redhat.com>
30581 * ree.c (combine_reaching_defs): Optimize also
30582 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
30583 reg2=any_extend(exp); reg1=reg2;, formatting fix.
30585 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
30587 PR middle-end/83189
30588 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
30590 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
30592 PR middle-end/83718
30593 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
30594 after they are computed.
30596 2018-01-11 Bin Cheng <bin.cheng@arm.com>
30598 PR tree-optimization/83695
30599 * gimple-loop-linterchange.cc
30600 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
30601 reset cached scev information after interchange.
30602 (pass_linterchange::execute): Remove call to scev_reset_htab.
30604 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30606 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
30607 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
30608 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
30609 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
30610 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
30611 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
30612 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
30613 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
30614 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
30615 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
30616 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
30617 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
30618 (V_lane_reg): Likewise.
30619 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
30621 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
30622 (vfmal_lane_low<mode>_intrinsic,
30623 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
30624 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
30625 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
30626 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
30627 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
30628 vfmsl_lane_high<mode>_intrinsic): New define_insns.
30630 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30632 * config/arm/arm-cpus.in (fp16fml): New feature.
30633 (ALL_SIMD): Add fp16fml.
30634 (armv8.2-a): Add fp16fml as an option.
30635 (armv8.3-a): Likewise.
30636 (armv8.4-a): Add fp16fml as part of fp16.
30637 * config/arm/arm.h (TARGET_FP16FML): Define.
30638 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
30640 * config/arm/arm-modes.def (V2HF): Define.
30641 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
30642 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
30643 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
30644 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
30645 vfmsl_low, vfmsl_high): New set of builtins.
30646 * config/arm/iterators.md (PLUSMINUS): New code iterator.
30647 (vfml_op): New code attribute.
30648 (VFMLHALVES): New int iterator.
30649 (VFML, VFMLSEL): New mode attributes.
30650 (V_reg): Define mapping for V2HF.
30651 (V_hi, V_lo): New mode attributes.
30652 (VF_constraint): Likewise.
30653 (vfml_half, vfml_half_selector): New int attributes.
30654 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
30656 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
30657 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
30659 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
30660 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
30661 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
30662 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
30664 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
30665 Document new effective target and option set.
30667 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30669 * config/arm/arm-cpus.in (armv8_4): New feature.
30670 (ARMv8_4a): New fgroup.
30671 (armv8.4-a): New arch.
30672 * config/arm/arm-tables.opt: Regenerate.
30673 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
30674 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
30675 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
30676 Add matching rules for -march=armv8.4-a and extensions.
30677 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
30679 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
30682 * config/rx/rx.md (BW): New mode attribute.
30683 (sync_lock_test_and_setsi): Add mode suffix to insn output.
30685 2018-01-11 Richard Biener <rguenther@suse.de>
30687 PR tree-optimization/83435
30688 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
30689 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
30690 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
30692 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
30693 Alan Hayward <alan.hayward@arm.com>
30694 David Sherwood <david.sherwood@arm.com>
30696 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
30698 (aarch64_classify_address): Initialize it. Track polynomial offsets.
30699 (aarch64_print_address_internal): Use it to check for a zero offset.
30701 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
30702 Alan Hayward <alan.hayward@arm.com>
30703 David Sherwood <david.sherwood@arm.com>
30705 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
30706 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
30707 Return a poly_int64 rather than a HOST_WIDE_INT.
30708 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
30709 rather than a HOST_WIDE_INT.
30710 * config/aarch64/aarch64.h (aarch64_frame): Protect with
30711 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
30712 hard_fp_offset, frame_size, initial_adjust, callee_offset and
30713 final_offset from HOST_WIDE_INT to poly_int64.
30714 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
30715 to_constant when getting the number of units in an Advanced SIMD
30717 (aarch64_builtin_vectorized_function): Check for a constant number
30719 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
30721 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
30722 attribute instead of GET_MODE_NUNITS.
30723 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
30724 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
30725 GET_MODE_SIZE for fixed-size registers.
30726 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
30727 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
30728 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
30729 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
30730 (aarch64_print_operand, aarch64_print_address_internal)
30731 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
30732 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
30733 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
30734 Handle polynomial GET_MODE_SIZE.
30735 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
30736 wider than SImode without modification.
30737 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
30738 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
30739 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
30740 passing and returning SVE modes.
30741 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
30742 rather than GEN_INT.
30743 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
30744 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
30745 (aarch64_allocate_and_probe_stack_space): Likewise.
30746 (aarch64_layout_frame): Cope with polynomial offsets.
30747 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
30748 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
30749 polynomial offsets.
30750 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
30751 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
30752 poly_int64 rather than a HOST_WIDE_INT.
30753 (aarch64_get_separate_components, aarch64_process_components)
30754 (aarch64_expand_prologue, aarch64_expand_epilogue)
30755 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
30756 (aarch64_anchor_offset): New function, split out from...
30757 (aarch64_legitimize_address): ...here.
30758 (aarch64_builtin_vectorization_cost): Handle polynomial
30759 TYPE_VECTOR_SUBPARTS.
30760 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
30762 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
30763 number of elements from the PARALLEL rather than the mode.
30764 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
30765 rather than GET_MODE_BITSIZE.
30766 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
30767 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
30768 (aarch64_expand_vec_perm_const_1): Handle polynomial
30769 d->perm.length () and d->perm elements.
30770 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
30771 Apply to_constant to d->perm elements.
30772 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
30773 polynomial CONST_VECTOR_NUNITS.
30774 (aarch64_move_pointer): Take amount as a poly_int64 rather
30776 (aarch64_progress_pointer): Avoid temporary variable.
30777 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
30778 the mode attribute instead of GET_MODE.
30780 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
30781 Alan Hayward <alan.hayward@arm.com>
30782 David Sherwood <david.sherwood@arm.com>
30784 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
30785 x exists before using it.
30786 (aarch64_add_constant_internal): Rename to...
30787 (aarch64_add_offset_1): ...this. Replace regnum with separate
30788 src and dest rtxes. Handle the case in which they're different,
30789 including when the offset is zero. Replace scratchreg with an rtx.
30790 Use 2 additions if there is no spare register into which we can
30791 move a 16-bit constant.
30792 (aarch64_add_constant): Delete.
30793 (aarch64_add_offset): Replace reg with separate src and dest
30794 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
30795 Use aarch64_add_offset_1.
30796 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
30797 an rtx rather than an int. Take the delta as a poly_int64
30798 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
30799 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
30800 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
30801 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
30802 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
30803 and aarch64_add_sp.
30804 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
30805 aarch64_add_constant.
30807 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
30809 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
30810 Use scalar_float_mode.
30812 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
30814 * config/aarch64/aarch64-simd.md
30815 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
30816 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
30817 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
30818 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
30819 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
30820 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
30821 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
30822 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
30823 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
30824 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
30826 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30829 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
30830 targ_options->x_arm_arch_string is non NULL.
30832 2018-01-11 Tamar Christina <tamar.christina@arm.com>
30834 * config/aarch64/aarch64.h
30835 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
30837 2018-01-11 Sudakshina Das <sudi.das@arm.com>
30840 * expmed.c (emit_store_flag_force): Swap if const op0
30841 and change VOIDmode to mode of op0.
30843 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
30845 PR rtl-optimization/83761
30846 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
30847 than bytes to mode_for_size.
30849 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
30851 PR middle-end/83189
30852 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
30853 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
30856 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
30858 PR middle-end/83575
30859 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
30860 when in layout mode.
30861 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
30862 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
30865 2018-01-10 Michael Collison <michael.collison@arm.com>
30867 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
30868 * config/aarch64/aarch64-option-extension.def: Add
30869 AARCH64_OPT_EXTENSION of 'fp16fml'.
30870 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
30871 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
30872 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
30873 * config/aarch64/constraints.md (Ui7): New constraint.
30874 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
30875 (VFMLA_SEL_W): Ditto.
30878 (VFMLA16_LOW): New int iterator.
30879 (VFMLA16_HIGH): Ditto.
30880 (UNSPEC_FMLAL): New unspec.
30881 (UNSPEC_FMLSL): Ditto.
30882 (UNSPEC_FMLAL2): Ditto.
30883 (UNSPEC_FMLSL2): Ditto.
30884 (f16mac): New code attribute.
30885 * config/aarch64/aarch64-simd-builtins.def
30886 (aarch64_fmlal_lowv2sf): Ditto.
30887 (aarch64_fmlsl_lowv2sf): Ditto.
30888 (aarch64_fmlalq_lowv4sf): Ditto.
30889 (aarch64_fmlslq_lowv4sf): Ditto.
30890 (aarch64_fmlal_highv2sf): Ditto.
30891 (aarch64_fmlsl_highv2sf): Ditto.
30892 (aarch64_fmlalq_highv4sf): Ditto.
30893 (aarch64_fmlslq_highv4sf): Ditto.
30894 (aarch64_fmlal_lane_lowv2sf): Ditto.
30895 (aarch64_fmlsl_lane_lowv2sf): Ditto.
30896 (aarch64_fmlal_laneq_lowv2sf): Ditto.
30897 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
30898 (aarch64_fmlalq_lane_lowv4sf): Ditto.
30899 (aarch64_fmlsl_lane_lowv4sf): Ditto.
30900 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
30901 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
30902 (aarch64_fmlal_lane_highv2sf): Ditto.
30903 (aarch64_fmlsl_lane_highv2sf): Ditto.
30904 (aarch64_fmlal_laneq_highv2sf): Ditto.
30905 (aarch64_fmlsl_laneq_highv2sf): Ditto.
30906 (aarch64_fmlalq_lane_highv4sf): Ditto.
30907 (aarch64_fmlsl_lane_highv4sf): Ditto.
30908 (aarch64_fmlalq_laneq_highv4sf): Ditto.
30909 (aarch64_fmlsl_laneq_highv4sf): Ditto.
30910 * config/aarch64/aarch64-simd.md:
30911 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
30912 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
30913 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
30914 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
30915 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
30916 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
30917 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
30918 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
30919 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
30920 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
30921 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
30922 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
30923 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
30924 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
30925 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
30926 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
30927 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
30928 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
30929 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
30930 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
30931 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
30932 (vfmlsl_low_u32): Ditto.
30933 (vfmlalq_low_u32): Ditto.
30934 (vfmlslq_low_u32): Ditto.
30935 (vfmlal_high_u32): Ditto.
30936 (vfmlsl_high_u32): Ditto.
30937 (vfmlalq_high_u32): Ditto.
30938 (vfmlslq_high_u32): Ditto.
30939 (vfmlal_lane_low_u32): Ditto.
30940 (vfmlsl_lane_low_u32): Ditto.
30941 (vfmlal_laneq_low_u32): Ditto.
30942 (vfmlsl_laneq_low_u32): Ditto.
30943 (vfmlalq_lane_low_u32): Ditto.
30944 (vfmlslq_lane_low_u32): Ditto.
30945 (vfmlalq_laneq_low_u32): Ditto.
30946 (vfmlslq_laneq_low_u32): Ditto.
30947 (vfmlal_lane_high_u32): Ditto.
30948 (vfmlsl_lane_high_u32): Ditto.
30949 (vfmlal_laneq_high_u32): Ditto.
30950 (vfmlsl_laneq_high_u32): Ditto.
30951 (vfmlalq_lane_high_u32): Ditto.
30952 (vfmlslq_lane_high_u32): Ditto.
30953 (vfmlalq_laneq_high_u32): Ditto.
30954 (vfmlslq_laneq_high_u32): Ditto.
30955 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
30956 (AARCH64_FL_FOR_ARCH8_4): New.
30957 (AARCH64_ISA_F16FML): New ISA flag.
30958 (TARGET_F16FML): New feature flag for fp16fml.
30959 (doc/invoke.texi): Document new fp16fml option.
30961 2018-01-10 Michael Collison <michael.collison@arm.com>
30963 * config/aarch64/aarch64-builtins.c:
30964 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
30965 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
30966 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
30967 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
30968 (AARCH64_ISA_SHA3): New ISA flag.
30969 (TARGET_SHA3): New feature flag for sha3.
30970 * config/aarch64/iterators.md (sha512_op): New int attribute.
30971 (CRYPTO_SHA512): New int iterator.
30972 (UNSPEC_SHA512H): New unspec.
30973 (UNSPEC_SHA512H2): Ditto.
30974 (UNSPEC_SHA512SU0): Ditto.
30975 (UNSPEC_SHA512SU1): Ditto.
30976 * config/aarch64/aarch64-simd-builtins.def
30977 (aarch64_crypto_sha512hqv2di): New builtin.
30978 (aarch64_crypto_sha512h2qv2di): Ditto.
30979 (aarch64_crypto_sha512su0qv2di): Ditto.
30980 (aarch64_crypto_sha512su1qv2di): Ditto.
30981 (aarch64_eor3qv8hi): Ditto.
30982 (aarch64_rax1qv2di): Ditto.
30983 (aarch64_xarqv2di): Ditto.
30984 (aarch64_bcaxqv8hi): Ditto.
30985 * config/aarch64/aarch64-simd.md:
30986 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
30987 (aarch64_crypto_sha512su0qv2di): Ditto.
30988 (aarch64_crypto_sha512su1qv2di): Ditto.
30989 (aarch64_eor3qv8hi): Ditto.
30990 (aarch64_rax1qv2di): Ditto.
30991 (aarch64_xarqv2di): Ditto.
30992 (aarch64_bcaxqv8hi): Ditto.
30993 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
30994 (vsha512h2q_u64): Ditto.
30995 (vsha512su0q_u64): Ditto.
30996 (vsha512su1q_u64): Ditto.
30997 (veor3q_u16): Ditto.
30998 (vrax1q_u64): Ditto.
30999 (vxarq_u64): Ditto.
31000 (vbcaxq_u16): Ditto.
31001 * config/arm/types.md (crypto_sha512): New type attribute.
31002 (crypto_sha3): Ditto.
31003 (doc/invoke.texi): Document new sha3 option.
31005 2018-01-10 Michael Collison <michael.collison@arm.com>
31007 * config/aarch64/aarch64-builtins.c:
31008 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
31009 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
31010 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
31011 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
31012 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
31013 (AARCH64_ISA_SM4): New ISA flag.
31014 (TARGET_SM4): New feature flag for sm4.
31015 * config/aarch64/aarch64-simd-builtins.def
31016 (aarch64_sm3ss1qv4si): Ditto.
31017 (aarch64_sm3tt1aq4si): Ditto.
31018 (aarch64_sm3tt1bq4si): Ditto.
31019 (aarch64_sm3tt2aq4si): Ditto.
31020 (aarch64_sm3tt2bq4si): Ditto.
31021 (aarch64_sm3partw1qv4si): Ditto.
31022 (aarch64_sm3partw2qv4si): Ditto.
31023 (aarch64_sm4eqv4si): Ditto.
31024 (aarch64_sm4ekeyqv4si): Ditto.
31025 * config/aarch64/aarch64-simd.md:
31026 (aarch64_sm3ss1qv4si): Ditto.
31027 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
31028 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
31029 (aarch64_sm4eqv4si): Ditto.
31030 (aarch64_sm4ekeyqv4si): Ditto.
31031 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
31032 (sm3part_op): Ditto.
31033 (CRYPTO_SM3TT): Ditto.
31034 (CRYPTO_SM3PART): Ditto.
31035 (UNSPEC_SM3SS1): New unspec.
31036 (UNSPEC_SM3TT1A): Ditto.
31037 (UNSPEC_SM3TT1B): Ditto.
31038 (UNSPEC_SM3TT2A): Ditto.
31039 (UNSPEC_SM3TT2B): Ditto.
31040 (UNSPEC_SM3PARTW1): Ditto.
31041 (UNSPEC_SM3PARTW2): Ditto.
31042 (UNSPEC_SM4E): Ditto.
31043 (UNSPEC_SM4EKEY): Ditto.
31044 * config/aarch64/constraints.md (Ui2): New constraint.
31045 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
31046 * config/arm/types.md (crypto_sm3): New type attribute.
31047 (crypto_sm4): Ditto.
31048 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
31049 (vsm3tt1aq_u32): Ditto.
31050 (vsm3tt1bq_u32): Ditto.
31051 (vsm3tt2aq_u32): Ditto.
31052 (vsm3tt2bq_u32): Ditto.
31053 (vsm3partw1q_u32): Ditto.
31054 (vsm3partw2q_u32): Ditto.
31055 (vsm4eq_u32): Ditto.
31056 (vsm4ekeyq_u32): Ditto.
31057 (doc/invoke.texi): Document new sm4 option.
31059 2018-01-10 Michael Collison <michael.collison@arm.com>
31061 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
31062 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
31063 (AARCH64_FL_FOR_ARCH8_4): New.
31064 (AARCH64_FL_V8_4): New flag.
31065 (doc/invoke.texi): Document new armv8.4-a option.
31067 2018-01-10 Michael Collison <michael.collison@arm.com>
31069 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
31070 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
31071 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
31072 * config/aarch64/aarch64-option-extension.def: Add
31073 AARCH64_OPT_EXTENSION of 'sha2'.
31074 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
31075 (crypto): Disable sha2 and aes if crypto disabled.
31076 (crypto): Enable aes and sha2 if enabled.
31077 (simd): Disable sha2 and aes if simd disabled.
31078 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
31080 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
31081 (TARGET_SHA2): New feature flag for sha2.
31082 (TARGET_AES): New feature flag for aes.
31083 * config/aarch64/aarch64-simd.md:
31084 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
31085 conditional on TARGET_AES.
31086 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
31087 (aarch64_crypto_sha1hsi): Make pattern conditional
31089 (aarch64_crypto_sha1hv4si): Ditto.
31090 (aarch64_be_crypto_sha1hv4si): Ditto.
31091 (aarch64_crypto_sha1su1v4si): Ditto.
31092 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
31093 (aarch64_crypto_sha1su0v4si): Ditto.
31094 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
31095 (aarch64_crypto_sha256su0v4si): Ditto.
31096 (aarch64_crypto_sha256su1v4si): Ditto.
31097 (doc/invoke.texi): Document new aes and sha2 options.
31099 2018-01-10 Martin Sebor <msebor@redhat.com>
31101 PR tree-optimization/83781
31102 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
31105 2018-01-11 Martin Sebor <msebor@gmail.com>
31106 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31108 PR tree-optimization/83501
31109 PR tree-optimization/81703
31111 * tree-ssa-strlen.c (get_string_cst): Rename...
31112 (get_string_len): ...to this. Handle global constants.
31113 (handle_char_store): Adjust.
31115 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
31116 Jim Wilson <jimw@sifive.com>
31118 * config/riscv/riscv-protos.h (riscv_output_return): New.
31119 * config/riscv/riscv.c (struct machine_function): New naked_p field.
31120 (riscv_attribute_table, riscv_output_return),
31121 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
31122 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
31123 (riscv_compute_frame_info): Only compute frame->mask if not a naked
31125 (riscv_expand_prologue): Add early return for naked function.
31126 (riscv_expand_epilogue): Likewise.
31127 (riscv_function_ok_for_sibcall): Return false for naked function.
31128 (riscv_set_current_function): New.
31129 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
31130 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
31131 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
31132 * doc/extend.texi (RISC-V Function Attributes): New.
31134 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
31136 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
31137 check for 128-bit long double before checking TCmode.
31138 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
31139 128-bit long doubles before checking TFmode or TCmode.
31140 (FLOAT128_IBM_P): Likewise.
31142 2018-01-10 Martin Sebor <msebor@redhat.com>
31144 PR tree-optimization/83671
31145 * builtins.c (c_strlen): Unconditionally return zero for the empty
31147 Use -Warray-bounds for warnings.
31148 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
31149 for non-constant array indices with COMPONENT_REF, arrays of
31150 arrays, and pointers to arrays.
31151 (gimple_fold_builtin_strlen): Determine and set length range for
31152 non-constant character arrays.
31154 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
31156 PR middle-end/81897
31157 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
31160 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
31162 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
31164 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
31167 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
31168 VECTOR_MEM_ALTIVEC_OR_VSX_P.
31169 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
31170 indexed_or_indirect_operand predicate.
31171 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
31172 (*vsx_le_perm_load_v8hi): Likewise.
31173 (*vsx_le_perm_load_v16qi): Likewise.
31174 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
31175 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
31176 (*vsx_le_perm_store_v8hi): Likewise.
31177 (*vsx_le_perm_store_v16qi): Likewise.
31178 (eight unnamed splitters): Likewise.
31180 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
31182 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
31183 * config/rs6000/emmintrin.h: Likewise.
31184 * config/rs6000/mmintrin.h: Likewise.
31185 * config/rs6000/xmmintrin.h: Likewise.
31187 2018-01-10 David Malcolm <dmalcolm@redhat.com>
31190 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
31192 * tree.c (tree_nop_conversion): Return true for location wrapper
31194 (maybe_wrap_with_location): New function.
31195 (selftest::check_strip_nops): New function.
31196 (selftest::test_location_wrappers): New function.
31197 (selftest::tree_c_tests): Call it.
31198 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
31199 (maybe_wrap_with_location): New decl.
31200 (EXPR_LOCATION_WRAPPER_P): New macro.
31201 (location_wrapper_p): New inline function.
31202 (tree_strip_any_location_wrapper): New inline function.
31204 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
31207 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
31208 stack_realign_offset for the largest alignment of stack slot
31210 (ix86_find_max_used_stack_alignment): New function.
31211 (ix86_finalize_stack_frame_flags): Use it. Set
31212 max_used_stack_alignment if we don't realign stack.
31213 * config/i386/i386.h (machine_function): Add
31214 max_used_stack_alignment.
31216 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
31218 * config/arm/arm.opt (-mbranch-cost): New option.
31219 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
31222 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
31225 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
31226 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
31228 2018-01-10 Richard Biener <rguenther@suse.de>
31231 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
31232 early out so it also covers the case where we have a non-NULL
31235 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
31237 PR tree-optimization/83753
31238 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
31239 for non-strided grouped accesses if the number of elements is 1.
31241 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
31244 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
31245 * i386.h (TARGET_USE_GATHER): Define.
31246 * x86-tune.def (X86_TUNE_USE_GATHER): New.
31248 2018-01-10 Martin Liska <mliska@suse.cz>
31251 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
31252 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
31254 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
31255 CLEANUP_NO_PARTITIONING is not set.
31257 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
31259 * doc/rtl.texi: Remove documentation of (const ...) wrappers
31260 for vectors, as a partial revert of r254296.
31261 * rtl.h (const_vec_p): Delete.
31262 (const_vec_duplicate_p): Don't test for vector CONSTs.
31263 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
31264 * expmed.c (make_tree): Likewise.
31267 * common.md (E, F): Use CONSTANT_P instead of checking for
31269 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
31270 checking for CONST_VECTOR.
31272 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
31274 PR middle-end/83575
31275 * predict.c (force_edge_cold): Handle in more sane way edges
31276 with no prediction.
31278 2018-01-09 Carl Love <cel@us.ibm.com>
31280 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
31282 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
31283 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
31284 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
31285 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
31286 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
31287 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
31288 * config/rs6000/rs6000-protos.h: Add extern defition for
31289 rs6000_generate_float2_double_code.
31290 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
31292 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
31293 (float2_v2df): Add define_expand.
31295 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
31298 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
31299 op_mode in the force_to_mode call.
31301 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
31303 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
31304 instead of checking each element individually.
31305 (aarch64_evpc_uzp): Likewise.
31306 (aarch64_evpc_zip): Likewise.
31307 (aarch64_evpc_ext): Likewise.
31308 (aarch64_evpc_rev): Likewise.
31309 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
31310 instead of checking each element individually. Return true without
31312 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
31313 whether all selected elements come from the same input, instead of
31314 checking each element individually. Remove calls to gen_rtx_REG,
31315 start_sequence and end_sequence and instead assert that no rtl is
31318 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
31320 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
31321 order of HIGH and CONST checks.
31323 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
31325 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
31326 if the destination isn't an SSA_NAME.
31328 2018-01-09 Richard Biener <rguenther@suse.de>
31330 PR tree-optimization/83668
31331 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
31333 (canonicalize_loop_form): ... here, renamed from ...
31334 (canonicalize_loop_closed_ssa_form): ... this and amended to
31335 swap successor edges for loop exit blocks to make us use
31336 the RPO order we need for initial schedule generation.
31338 2018-01-09 Joseph Myers <joseph@codesourcery.com>
31340 PR tree-optimization/64811
31341 * match.pd: When optimizing comparisons with Inf, avoid
31342 introducing or losing exceptions from comparisons with NaN.
31344 2018-01-09 Martin Liska <mliska@suse.cz>
31347 * asan.c (shadow_mem_size): Add gcc_assert.
31349 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
31351 Don't save registers in main().
31354 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
31355 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
31356 * config/avr/avr.c (avr_set_current_function): Don't error if
31357 naked, OS_task or OS_main are specified at the same time.
31358 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
31360 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
31362 * common/config/avr/avr-common.c (avr_option_optimization_table):
31363 Switch on -mmain-is-OS_task for optimizing compilations.
31365 2018-01-09 Richard Biener <rguenther@suse.de>
31367 PR tree-optimization/83572
31368 * graphite.c: Include cfganal.h.
31369 (graphite_transform_loops): Connect infinite loops to exit
31370 and remove fake edges at the end.
31372 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
31374 * ipa-inline.c (edge_badness): Revert accidental checkin.
31376 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
31379 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
31380 symbols; not inline clones.
31382 2018-01-09 Jakub Jelinek <jakub@redhat.com>
31385 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
31386 hard registers. Formatting fixes.
31388 PR preprocessor/83722
31389 * gcc.c (try_generate_repro): Pass
31390 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
31391 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
31394 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
31395 Kito Cheng <kito.cheng@gmail.com>
31397 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
31398 (riscv_leaf_function_p): Delete.
31399 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
31401 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31403 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
31405 (do_ifelse): New function.
31406 (do_isel): New function.
31407 (do_sub3): New function.
31408 (do_add3): New function.
31409 (do_load_mask_compare): New function.
31410 (do_overlap_load_compare): New function.
31411 (expand_compare_loop): New function.
31412 (expand_block_compare): Call expand_compare_loop() when appropriate.
31413 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
31414 option description.
31415 (-mblock-compare-inline-loop-limit): New option.
31417 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31420 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
31421 Reverse order of second and third operands in first alternative.
31422 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
31423 of first and second elements in UNSPEC_VPERMR vector.
31424 (altivec_expand_vec_perm_le): Likewise.
31426 2018-01-08 Jeff Law <law@redhat.com>
31428 PR rtl-optimizatin/81308
31429 * tree-switch-conversion.c (cfg_altered): New file scoped static.
31430 (process_switch): If group_case_labels makes a change, then set
31432 (pass_convert_switch::execute): If a switch is converted, then
31433 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
31435 PR rtl-optimization/81308
31436 * recog.c (split_all_insns): Conditionally cleanup the CFG after
31439 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
31441 PR target/83663 - Revert r255946
31442 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
31443 generation for cases where splatting a value is not useful.
31444 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
31445 across a vec_duplicate and a paradoxical subreg forming a vector
31446 mode to a vec_concat.
31448 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31450 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
31451 -march=armv8.3-a variants.
31452 * config/arm/t-multilib: Likewise.
31453 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
31455 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31457 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
31459 (cceq_ior_compare_complement): Give it a name so I can use it, and
31460 change boolean_or_operator predicate to boolean_operator so it can
31461 be used to generate a crand.
31462 (eqne): New code iterator.
31463 (bd/bd_neg): New code_attrs.
31464 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
31465 a single define_insn.
31466 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
31467 decrement (bdnzt/bdnzf/bdzt/bdzf).
31468 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
31469 with the new names of the branch decrement patterns, and added the
31470 names of the branch decrement conditional patterns.
31472 2018-01-08 Richard Biener <rguenther@suse.de>
31474 PR tree-optimization/83563
31475 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
31478 2018-01-08 Richard Biener <rguenther@suse.de>
31480 PR middle-end/83713
31481 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
31483 2018-01-08 Richard Biener <rguenther@suse.de>
31485 PR tree-optimization/83685
31486 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
31487 references to abnormals.
31489 2018-01-08 Richard Biener <rguenther@suse.de>
31492 * dwarf2out.c (output_indirect_strings): Handle empty
31493 skeleton_debug_str_hash.
31494 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
31496 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
31498 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
31499 (emit_store_direct): Likewise.
31500 (arc_trampoline_adjust_address): Likewise.
31501 (arc_asm_trampoline_template): New function.
31502 (arc_initialize_trampoline): Use asm_trampoline_template.
31503 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
31504 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
31505 * config/arc/arc.md (flush_icache): Delete pattern.
31507 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
31509 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
31510 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
31513 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
31516 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
31517 by not USED_FOR_TARGET.
31518 (make_pass_resolve_sw_modes): Likewise.
31520 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
31522 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
31525 2018-01-08 Richard Biener <rguenther@suse.de>
31527 PR middle-end/83580
31528 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
31530 2018-01-08 Richard Biener <rguenther@suse.de>
31532 PR middle-end/83517
31533 * match.pd ((t * 2) / 2) -> t): Add missing :c.
31535 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
31537 PR middle-end/81897
31538 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
31539 basic blocks with a small number of successors.
31540 (convert_control_dep_chain_into_preds): Improve handling of
31542 (dump_predicates): Split apart into...
31543 (dump_pred_chain): ...here...
31544 (dump_pred_info): ...and here.
31545 (can_one_predicate_be_invalidated_p): Add debugging printfs.
31546 (can_chain_union_be_invalidated_p): Improve check for invalidation
31548 (uninit_uses_cannot_happen): Avoid unnecessary if
31549 convert_control_dep_chain_into_preds yielded nothing.
31551 2018-01-06 Martin Sebor <msebor@redhat.com>
31553 PR tree-optimization/83640
31554 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
31555 subtracting negative offset from size.
31556 (builtin_access::overlap): Adjust offset bounds of the access to fall
31557 within the size of the object if possible.
31559 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
31561 PR rtl-optimization/83699
31562 * expmed.c (extract_bit_field_1): Restrict the vector usage of
31563 extract_bit_field_as_subreg to cases in which the extracted
31564 value is also a vector.
31566 * lra-constraints.c (process_alt_operands): Test for the equivalence
31567 substitutions when detecting a possible reload cycle.
31569 2018-01-06 Jakub Jelinek <jakub@redhat.com>
31572 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
31573 by default if flag_selective_schedling{,2}. Formatting fixes.
31575 PR rtl-optimization/83682
31576 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
31577 if it has non-VECTOR_MODE element mode.
31578 (vec_duplicate_p): Likewise.
31580 PR middle-end/83694
31581 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
31582 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
31584 2018-01-05 Jakub Jelinek <jakub@redhat.com>
31587 * config/i386/i386-builtin.def
31588 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
31589 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
31590 Require also OPTION_MASK_ISA_AVX512F in addition to
31591 OPTION_MASK_ISA_GFNI.
31592 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
31593 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
31594 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
31595 to OPTION_MASK_ISA_GFNI.
31596 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
31597 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
31598 OPTION_MASK_ISA_AVX512BW.
31599 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
31600 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
31601 addition to OPTION_MASK_ISA_GFNI.
31602 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
31603 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
31604 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
31605 to OPTION_MASK_ISA_GFNI.
31606 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
31607 a requirement for all ISAs rather than any of them with a few
31609 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
31611 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
31612 bitmasks to be enabled with 3 exceptions, instead of requiring any
31613 enabled ISA with lots of exceptions.
31614 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
31615 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
31616 Change avx512bw in isa attribute to avx512f.
31617 * config/i386/sgxintrin.h: Add license boilerplate.
31618 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
31619 to __AVX512F__ and __AVX512VL to __AVX512VL__.
31620 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
31621 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
31623 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
31624 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
31625 temporarily sse2 rather than sse if not enabled already.
31628 * config/i386/sse.md (VI248_VLBW): Rename to ...
31629 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
31630 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
31631 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
31632 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
31633 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
31634 mode iterator instead of VI248_VLBW.
31636 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
31638 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
31639 (record_modified): Skip clobbers; add debug output.
31640 (param_change_prob): Use sreal frequencies.
31642 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
31644 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
31645 punt for user-aligned variables.
31647 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
31649 * tree-chrec.c (chrec_contains_symbols): Return true for
31652 2018-01-05 Sudakshina Das <sudi.das@arm.com>
31655 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
31656 of (x|y) == x for BICS pattern.
31658 2018-01-05 Jakub Jelinek <jakub@redhat.com>
31660 PR tree-optimization/83605
31661 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
31662 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
31665 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
31667 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
31668 * config/epiphany/rtems.h: New file.
31670 2018-01-04 Jakub Jelinek <jakub@redhat.com>
31671 Uros Bizjak <ubizjak@gmail.com>
31674 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
31675 QIreg_operand instead of register_operand predicate.
31676 * config/i386/i386.c (ix86_rop_should_change_byte_p,
31677 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
31678 comments instead of -fmitigate[-_]rop.
31680 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31683 * cgraphunit.c (symbol_table::compile): Switch to text_section
31684 before calling assembly_start debug hook.
31685 * run-rtl-passes.c (run_rtl_passes): Likewise.
31688 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
31690 * tree-vrp.c (extract_range_from_binary_expr_1): Check
31691 range_int_cst_p rather than !symbolic_range_p before calling
31692 extract_range_from_multiplicative_op_1.
31694 2018-01-04 Jeff Law <law@redhat.com>
31696 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
31697 redundant test in assertion.
31699 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
31701 * doc/rtl.texi: Document machine_mode wrapper classes.
31703 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
31705 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
31706 using tree_to_uhwi.
31708 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
31710 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
31711 the VEC_PERM_EXPR fold to fail.
31713 2018-01-04 Jakub Jelinek <jakub@redhat.com>
31716 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
31717 to switched_sections.
31719 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
31722 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
31723 test for d.testing.
31725 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
31728 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
31729 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
31731 2018-01-04 Jakub Jelinek <jakub@redhat.com>
31734 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
31735 is BLKmode and bitpos not zero or mode change is needed.
31737 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
31740 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
31743 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
31746 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
31747 instead of MULT rtx. Update all corresponding splitters.
31748 (*saddl_se): Ditto.
31749 (*ssub<modesuffix>): Ditto.
31750 (*ssubl_se): Ditto.
31751 (*cmp_sadd_di): Update split patterns.
31752 (*cmp_sadd_si): Ditto.
31753 (*cmp_sadd_sidi): Ditto.
31754 (*cmp_ssub_di): Ditto.
31755 (*cmp_ssub_si): Ditto.
31756 (*cmp_ssub_sidi): Ditto.
31757 * config/alpha/predicates.md (const23_operand): New predicate.
31758 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
31759 Look for ASHIFT, not MULT inner operand.
31760 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
31762 2018-01-04 Martin Liska <mliska@suse.cz>
31764 PR gcov-profile/83669
31765 * gcov.c (output_intermediate_file): Add version to intermediate
31767 * doc/gcov.texi: Document new field 'version' in intermediate
31768 file format. Fix location of '-k' option of gcov command.
31770 2018-01-04 Martin Liska <mliska@suse.cz>
31773 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
31775 2018-01-04 Jakub Jelinek <jakub@redhat.com>
31777 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
31779 2018-01-03 Martin Sebor <msebor@redhat.com>
31781 PR tree-optimization/83655
31782 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
31783 checking calls with invalid arguments.
31785 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31787 * tree-vect-stmts.c (vect_get_store_rhs): New function.
31788 (vectorizable_mask_load_store): Delete.
31789 (vectorizable_call): Return false for masked loads and stores.
31790 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
31791 instead of gimple_assign_rhs1.
31792 (vectorizable_load): Handle IFN_MASK_LOAD.
31793 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
31795 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31797 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
31799 (vectorizable_mask_load_store): ...here.
31800 (vectorizable_load): ...and here.
31802 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31804 * tree-vect-stmts.c (vect_build_all_ones_mask)
31805 (vect_build_zero_merge_argument): New functions, split out from...
31806 (vectorizable_load): ...here.
31808 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31810 * tree-vect-stmts.c (vect_check_store_rhs): New function,
31812 (vectorizable_mask_load_store): ...here.
31813 (vectorizable_store): ...and here.
31815 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31817 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
31819 (vectorizable_mask_load_store): ...here.
31821 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31823 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
31824 (vect_model_store_cost): Take a vec_load_store_type instead of a
31826 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
31827 (vect_model_store_cost): Take a vec_load_store_type instead of a
31829 (vectorizable_mask_load_store): Update accordingly.
31830 (vectorizable_store): Likewise.
31831 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
31833 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31835 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
31836 IFN_MASK_LOAD calls here rather than...
31837 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
31839 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31840 Alan Hayward <alan.hayward@arm.com>
31841 David Sherwood <david.sherwood@arm.com>
31843 * expmed.c (extract_bit_field_1): For vector extracts,
31844 fall back to extract_bit_field_as_subreg if vec_extract
31847 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31848 Alan Hayward <alan.hayward@arm.com>
31849 David Sherwood <david.sherwood@arm.com>
31851 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
31852 they are variable or constant sized.
31853 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
31854 slots for constant-sized data.
31856 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31857 Alan Hayward <alan.hayward@arm.com>
31858 David Sherwood <david.sherwood@arm.com>
31860 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
31861 handling COND_EXPRs with boolean comparisons, try to find a better
31862 basis for the mask type than the boolean itself.
31864 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31866 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
31867 is calculated and how it can be overridden.
31868 * genmodes.c (max_bitsize_mode_any_mode): New variable.
31869 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
31871 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
31874 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31875 Alan Hayward <alan.hayward@arm.com>
31876 David Sherwood <david.sherwood@arm.com>
31878 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
31879 Remove the mode argument.
31880 (aarch64_simd_valid_immediate): Remove the mode and inverse
31882 * config/aarch64/iterators.md (bitsize): New iterator.
31883 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
31884 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
31885 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
31886 aarch64_simd_valid_immediate.
31887 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
31888 (aarch64_reg_or_bic_imm): Likewise.
31889 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
31890 with an insn_type enum and msl with a modifier_type enum.
31891 Replace element_width with a scalar_mode. Change the shift
31892 to unsigned int. Add constructors for scalar_float_mode and
31893 scalar_int_mode elements.
31894 (aarch64_vect_float_const_representable_p): Delete.
31895 (aarch64_can_const_movi_rtx_p)
31896 (aarch64_simd_scalar_immediate_valid_for_move)
31897 (aarch64_simd_make_constant): Update call to
31898 aarch64_simd_valid_immediate.
31899 (aarch64_advsimd_valid_immediate_hs): New function.
31900 (aarch64_advsimd_valid_immediate): Likewise.
31901 (aarch64_simd_valid_immediate): Remove mode and inverse
31902 arguments. Rewrite to use the above. Use const_vec_duplicate_p
31903 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
31904 and aarch64_float_const_representable_p on the result.
31905 (aarch64_output_simd_mov_immediate): Remove mode argument.
31906 Update call to aarch64_simd_valid_immediate and use of
31907 simd_immediate_info.
31908 (aarch64_output_scalar_simd_mov_immediate): Update call
31911 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31912 Alan Hayward <alan.hayward@arm.com>
31913 David Sherwood <david.sherwood@arm.com>
31915 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
31916 (mode_nunits): Likewise CONST_MODE_NUNITS.
31917 * machmode.def (ADJUST_NUNITS): Document.
31918 * genmodes.c (mode_data::need_nunits_adj): New field.
31919 (blank_mode): Update accordingly.
31920 (adj_nunits): New variable.
31921 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
31923 (emit_mode_size_inline): Set need_bytesize_adj for all modes
31924 listed in adj_nunits.
31925 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
31926 listed in adj_nunits. Don't emit case statements for such modes.
31927 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
31928 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
31929 nothing if adj_nunits is nonnull.
31930 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
31931 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
31932 (emit_mode_fbit): Update use of print_maybe_const_decl.
31933 (emit_move_size): Likewise. Treat the array as non-const
31935 (emit_mode_adjustments): Handle adj_nunits.
31937 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31939 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
31940 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
31941 (VECTOR_MODES): Use it.
31942 (make_vector_modes): Take the prefix as an argument.
31944 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31945 Alan Hayward <alan.hayward@arm.com>
31946 David Sherwood <david.sherwood@arm.com>
31948 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
31949 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
31950 for MODE_VECTOR_BOOL.
31951 * machmode.def (VECTOR_BOOL_MODE): Document.
31952 * genmodes.c (VECTOR_BOOL_MODE): New macro.
31953 (make_vector_bool_mode): New function.
31954 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
31956 * lto-streamer-in.c (lto_input_mode_table): Likewise.
31957 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
31959 * stor-layout.c (int_mode_for_mode): Likewise.
31960 * tree.c (build_vector_type_for_mode): Likewise.
31961 * varasm.c (output_constant_pool_2): Likewise.
31962 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
31963 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
31964 for MODE_VECTOR_BOOL.
31965 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
31966 of mode class checks.
31967 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
31968 instead of a list of mode class checks.
31969 (expand_vector_scalar_condition): Likewise.
31970 (type_for_widest_vector_mode): Handle BImode as an inner mode.
31972 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31973 Alan Hayward <alan.hayward@arm.com>
31974 David Sherwood <david.sherwood@arm.com>
31976 * machmode.h (mode_size): Change from unsigned short to
31978 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
31979 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
31980 or if measurement_type is not polynomial.
31981 (fixed_size_mode::includes_p): Check for constant-sized modes.
31982 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
31983 return a poly_uint16 rather than an unsigned short.
31984 (emit_mode_size): Change the type of mode_size from unsigned short
31985 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
31986 (emit_mode_adjustments): Cope with polynomial vector sizes.
31987 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
31989 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
31991 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
31992 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
31993 * caller-save.c (setup_save_areas): Likewise.
31994 (replace_reg_with_saved_mem): Likewise.
31995 * calls.c (emit_library_call_value_1): Likewise.
31996 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
31997 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
31998 (gen_lowpart_for_combine): Likewise.
31999 * convert.c (convert_to_integer_1): Likewise.
32000 * cse.c (equiv_constant, cse_insn): Likewise.
32001 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
32002 (cselib_subst_to_values): Likewise.
32003 * dce.c (word_dce_process_block): Likewise.
32004 * df-problems.c (df_word_lr_mark_ref): Likewise.
32005 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
32006 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
32007 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
32008 (rtl_for_decl_location): Likewise.
32009 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
32010 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
32011 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
32012 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
32013 (expand_expr_real_1): Likewise.
32014 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
32015 (pad_below): Likewise.
32016 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
32017 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
32018 * ira.c (get_subreg_tracking_sizes): Likewise.
32019 * ira-build.c (ira_create_allocno_objects): Likewise.
32020 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
32021 (ira_sort_regnos_for_alter_reg): Likewise.
32022 * ira-costs.c (record_operand_costs): Likewise.
32023 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
32024 (resolve_simple_move): Likewise.
32025 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
32026 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
32027 (lra_constraints): Likewise.
32028 (CONST_POOL_OK_P): Reject variable-sized modes.
32029 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
32030 (add_pseudo_to_slot, lra_spill): Likewise.
32031 * omp-low.c (omp_clause_aligned_alignment): Likewise.
32032 * optabs-query.c (get_best_extraction_insn): Likewise.
32033 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
32034 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
32035 (expand_mult_highpart, valid_multiword_target_p): Likewise.
32036 * recog.c (offsettable_address_addr_space_p): Likewise.
32037 * regcprop.c (maybe_mode_change): Likewise.
32038 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
32039 * regrename.c (build_def_use): Likewise.
32040 * regstat.c (dump_reg_info): Likewise.
32041 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
32042 (find_reloads, find_reloads_subreg_address): Likewise.
32043 * reload1.c (eliminate_regs_1): Likewise.
32044 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
32045 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
32046 (simplify_binary_operation_1, simplify_subreg): Likewise.
32047 * targhooks.c (default_function_arg_padding): Likewise.
32048 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
32049 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
32050 (verify_gimple_assign_ternary): Likewise.
32051 * tree-inline.c (estimate_move_cost): Likewise.
32052 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
32053 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
32054 (get_address_cost_ainc): Likewise.
32055 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
32056 (vect_supportable_dr_alignment): Likewise.
32057 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32058 (vectorizable_reduction): Likewise.
32059 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
32060 (vectorizable_operation, vectorizable_load): Likewise.
32061 * tree.c (build_same_sized_truth_vector_type): Likewise.
32062 * valtrack.c (cleanup_auto_inc_dec): Likewise.
32063 * var-tracking.c (emit_note_insn_var_location): Likewise.
32064 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
32065 (ADDR_VEC_ALIGN): Likewise.
32067 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32068 Alan Hayward <alan.hayward@arm.com>
32069 David Sherwood <david.sherwood@arm.com>
32071 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
32073 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
32074 or if measurement_type is polynomial.
32075 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
32076 * combine.c (make_extraction): Likewise.
32077 * dse.c (find_shift_sequence): Likewise.
32078 * dwarf2out.c (mem_loc_descriptor): Likewise.
32079 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
32080 (extract_bit_field, extract_low_bits): Likewise.
32081 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
32082 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
32083 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
32084 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
32085 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
32086 * reload.c (find_reloads): Likewise.
32087 * reload1.c (alter_reg): Likewise.
32088 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
32089 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
32090 * tree-if-conv.c (predicate_mem_writes): Likewise.
32091 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
32092 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
32093 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
32094 * valtrack.c (dead_debug_insert_temp): Likewise.
32095 * varasm.c (mergeable_constant_section): Likewise.
32096 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
32098 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32099 Alan Hayward <alan.hayward@arm.com>
32100 David Sherwood <david.sherwood@arm.com>
32102 * expr.c (expand_assignment): Cope with polynomial mode sizes
32103 when assigning to a CONCAT.
32105 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32106 Alan Hayward <alan.hayward@arm.com>
32107 David Sherwood <david.sherwood@arm.com>
32109 * machmode.h (mode_precision): Change from unsigned short to
32111 (mode_to_precision): Return a poly_uint16 rather than an unsigned
32113 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
32114 or if measurement_type is not polynomial.
32115 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
32116 in which the mode is already known to be a scalar_int_mode.
32117 * genmodes.c (emit_mode_precision): Change the type of mode_precision
32118 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
32120 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
32121 for GET_MODE_PRECISION.
32122 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
32123 for GET_MODE_PRECISION.
32124 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
32126 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
32127 (expand_field_assignment, make_extraction): Likewise.
32128 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
32129 (get_last_value): Likewise.
32130 * convert.c (convert_to_integer_1): Likewise.
32131 * cse.c (cse_insn): Likewise.
32132 * expr.c (expand_expr_real_1): Likewise.
32133 * lra-constraints.c (simplify_operand_subreg): Likewise.
32134 * optabs-query.c (can_atomic_load_p): Likewise.
32135 * optabs.c (expand_atomic_load): Likewise.
32136 (expand_atomic_store): Likewise.
32137 * ree.c (combine_reaching_defs): Likewise.
32138 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
32139 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
32140 * tree.h (type_has_mode_precision_p): Likewise.
32141 * ubsan.c (instrument_si_overflow): Likewise.
32143 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32144 Alan Hayward <alan.hayward@arm.com>
32145 David Sherwood <david.sherwood@arm.com>
32147 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
32148 polynomial numbers of units.
32149 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
32150 (valid_vector_subparts_p): New function.
32151 (build_vector_type): Remove temporary shim and take the number
32152 of units as a poly_uint64 rather than an int.
32153 (build_opaque_vector_type): Take the number of units as a
32154 poly_uint64 rather than an int.
32155 * tree.c (build_vector_from_ctor): Handle polynomial
32156 TYPE_VECTOR_SUBPARTS.
32157 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
32158 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
32159 (build_vector_from_val): If the number of units is variable,
32160 use build_vec_duplicate_cst for constant operands and
32161 VEC_DUPLICATE_EXPR otherwise.
32162 (make_vector_type): Remove temporary is_constant ().
32163 (build_vector_type, build_opaque_vector_type): Take the number of
32164 units as a poly_uint64 rather than an int.
32165 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
32167 * cfgexpand.c (expand_debug_expr): Likewise.
32168 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
32169 (store_constructor, expand_expr_real_1): Likewise.
32170 (const_scalar_mask_from_tree): Likewise.
32171 * fold-const-call.c (fold_const_reduction): Likewise.
32172 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
32173 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
32174 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
32175 (fold_relational_const): Likewise.
32176 (native_interpret_vector): Likewise. Change the size from an
32177 int to an unsigned int.
32178 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
32179 TYPE_VECTOR_SUBPARTS.
32180 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
32181 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
32182 duplicating a non-constant operand into a variable-length vector.
32183 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
32184 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
32185 * ipa-icf.c (sem_variable::equals): Likewise.
32186 * match.pd: Likewise.
32187 * omp-simd-clone.c (simd_clone_subparts): Likewise.
32188 * print-tree.c (print_node): Likewise.
32189 * stor-layout.c (layout_type): Likewise.
32190 * targhooks.c (default_builtin_vectorization_cost): Likewise.
32191 * tree-cfg.c (verify_gimple_comparison): Likewise.
32192 (verify_gimple_assign_binary): Likewise.
32193 (verify_gimple_assign_ternary): Likewise.
32194 (verify_gimple_assign_single): Likewise.
32195 * tree-pretty-print.c (dump_generic_node): Likewise.
32196 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
32197 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
32198 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
32199 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
32200 (vect_shift_permute_load_chain): Likewise.
32201 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
32202 (expand_vector_condition, optimize_vector_constructor): Likewise.
32203 (lower_vec_perm, get_compute_type): Likewise.
32204 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32205 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
32206 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
32207 (vect_recog_mask_conversion_pattern): Likewise.
32208 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
32209 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
32210 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
32211 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
32212 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
32213 (vectorizable_shift, vectorizable_operation, vectorizable_store)
32214 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
32215 (supportable_widening_operation): Likewise.
32216 (supportable_narrowing_operation): Likewise.
32217 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
32219 * varasm.c (output_constant): Likewise.
32221 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32222 Alan Hayward <alan.hayward@arm.com>
32223 David Sherwood <david.sherwood@arm.com>
32225 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
32226 so that both the length == 3 and length != 3 cases set up their
32227 own permute vectors. Add comments explaining why we know the
32228 number of elements is constant.
32229 (vect_permute_load_chain): Likewise.
32231 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32232 Alan Hayward <alan.hayward@arm.com>
32233 David Sherwood <david.sherwood@arm.com>
32235 * machmode.h (mode_nunits): Change from unsigned char to
32237 (ONLY_FIXED_SIZE_MODES): New macro.
32238 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
32239 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
32240 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
32242 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
32243 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
32244 or if measurement_type is not polynomial.
32245 * genmodes.c (ZERO_COEFFS): New macro.
32246 (emit_mode_nunits_inline): Make mode_nunits_inline return a
32248 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
32249 Use ZERO_COEFFS when emitting initializers.
32250 * data-streamer.h (bp_pack_poly_value): New function.
32251 (bp_unpack_poly_value): Likewise.
32252 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
32253 for GET_MODE_NUNITS.
32254 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
32255 for GET_MODE_NUNITS.
32256 * tree.c (make_vector_type): Remove temporary shim and make
32257 the real function take the number of units as a poly_uint64
32258 rather than an int.
32259 (build_vector_type_for_mode): Handle polynomial nunits.
32260 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
32261 * emit-rtl.c (const_vec_series_p_1): Likewise.
32262 (gen_rtx_CONST_VECTOR): Likewise.
32263 * fold-const.c (test_vec_duplicate_folding): Likewise.
32264 * genrecog.c (validate_pattern): Likewise.
32265 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
32266 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
32267 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
32268 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
32269 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
32270 * rtlanal.c (subreg_get_info): Likewise.
32271 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
32272 (vect_grouped_load_supported): Likewise.
32273 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
32274 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
32275 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
32276 (simplify_const_unary_operation, simplify_binary_operation_1)
32277 (simplify_const_binary_operation, simplify_ternary_operation)
32278 (test_vector_ops_duplicate, test_vector_ops): Likewise.
32279 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
32280 instead of CONST_VECTOR_NUNITS.
32281 * varasm.c (output_constant_pool_2): Likewise.
32282 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
32283 explicit-encoded elements in the XVEC for variable-length vectors.
32285 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32287 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
32289 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32290 Alan Hayward <alan.hayward@arm.com>
32291 David Sherwood <david.sherwood@arm.com>
32293 * coretypes.h (fixed_size_mode): Declare.
32294 (fixed_size_mode_pod): New typedef.
32295 * builtins.h (target_builtins::x_apply_args_mode)
32296 (target_builtins::x_apply_result_mode): Change type to
32297 fixed_size_mode_pod.
32298 * builtins.c (apply_args_size, apply_result_size, result_vector)
32299 (expand_builtin_apply_args_1, expand_builtin_apply)
32300 (expand_builtin_return): Update accordingly.
32302 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32304 * cse.c (hash_rtx_cb): Hash only the encoded elements.
32305 * cselib.c (cselib_hash_rtx): Likewise.
32306 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
32307 CONST_VECTOR encoding.
32309 2018-01-03 Jakub Jelinek <jakub@redhat.com>
32310 Jeff Law <law@redhat.com>
32313 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
32314 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
32315 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
32316 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
32319 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
32320 explicitly probe *sp in a noreturn function if there were any callee
32321 register saves or frame pointer is needed.
32323 2018-01-03 Jakub Jelinek <jakub@redhat.com>
32326 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
32327 BLKmode for ternary, binary or unary expressions.
32330 * var-tracking.c (delete_vta_debug_insn): New inline function.
32331 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
32332 insns from get_insns () to NULL instead of each bb separately.
32333 Use delete_vta_debug_insn. No longer static.
32334 (vt_debug_insns_local, variable_tracking_main_1): Adjust
32335 delete_vta_debug_insns callers.
32336 * rtl.h (delete_vta_debug_insns): Declare.
32337 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
32338 instead of variable_tracking_main.
32340 2018-01-03 Martin Sebor <msebor@redhat.com>
32342 PR tree-optimization/83603
32343 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
32344 arguments past the endof the argument list in functions declared
32345 without a prototype.
32346 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
32347 Avoid checking when arguments are null.
32349 2018-01-03 Martin Sebor <msebor@redhat.com>
32352 * doc/extend.texi (attribute const): Fix a typo.
32353 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
32354 issuing -Wsuggest-attribute for void functions.
32356 2018-01-03 Martin Sebor <msebor@redhat.com>
32358 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
32359 offset_int::from instead of wide_int::to_shwi.
32360 (maybe_diag_overlap): Remove assertion.
32361 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
32362 * gimple-ssa-sprintf.c (format_directive): Same.
32363 (parse_directive): Same.
32364 (sprintf_dom_walker::compute_format_length): Same.
32365 (try_substitute_return_value): Same.
32367 2018-01-03 Jeff Law <law@redhat.com>
32369 PR middle-end/83654
32370 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
32371 non-constant residual for zero at runtime and avoid probing in
32372 that case. Reorganize code for trailing problem to mirror handling
32375 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32377 PR tree-optimization/83501
32378 * tree-ssa-strlen.c (get_string_cst): New.
32379 (handle_char_store): Call get_string_cst.
32381 2018-01-03 Martin Liska <mliska@suse.cz>
32383 PR tree-optimization/83593
32384 * tree-ssa-strlen.c: Include tree-cfg.h.
32385 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
32386 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
32387 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
32389 (strlen_dom_walker::before_dom_children): Call
32390 gimple_purge_dead_eh_edges. Dump tranformation with details
32392 (strlen_dom_walker::before_dom_children): Update call by adding
32393 new argument cleanup_eh.
32394 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
32396 2018-01-03 Martin Liska <mliska@suse.cz>
32399 * cif-code.def (VARIADIC_THUNK): New enum value.
32400 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
32403 2018-01-03 Jan Beulich <jbeulich@suse.com>
32405 * sse.md (mov<mode>_internal): Tighten condition for when to use
32406 vmovdqu<ssescalarsize> for TI and OI modes.
32408 2018-01-03 Jakub Jelinek <jakub@redhat.com>
32410 Update copyright years.
32412 2018-01-03 Martin Liska <mliska@suse.cz>
32415 * ipa-visibility.c (function_and_variable_visibility): Skip
32416 functions with noipa attribure.
32418 2018-01-03 Jakub Jelinek <jakub@redhat.com>
32420 * gcc.c (process_command): Update copyright notice dates.
32421 * gcov-dump.c (print_version): Ditto.
32422 * gcov.c (print_version): Ditto.
32423 * gcov-tool.c (print_version): Ditto.
32424 * gengtype.c (create_file): Ditto.
32425 * doc/cpp.texi: Bump @copying's copyright year.
32426 * doc/cppinternals.texi: Ditto.
32427 * doc/gcc.texi: Ditto.
32428 * doc/gccint.texi: Ditto.
32429 * doc/gcov.texi: Ditto.
32430 * doc/install.texi: Ditto.
32431 * doc/invoke.texi: Ditto.
32433 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32435 * vector-builder.h (vector_builder::m_full_nelts): Change from
32436 unsigned int to poly_uint64.
32437 (vector_builder::full_nelts): Update prototype accordingly.
32438 (vector_builder::new_vector): Likewise.
32439 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
32440 (vector_builder::operator ==): Likewise.
32441 (vector_builder::finalize): Likewise.
32442 * int-vector-builder.h (int_vector_builder::int_vector_builder):
32443 Take the number of elements as a poly_uint64 rather than an
32445 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
32446 from unsigned int to poly_uint64.
32447 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
32448 (vec_perm_indices::new_vector): Likewise.
32449 (vec_perm_indices::length): Likewise.
32450 (vec_perm_indices::nelts_per_input): Likewise.
32451 (vec_perm_indices::input_nelts): Likewise.
32452 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
32453 number of elements per input as a poly_uint64 rather than an
32454 unsigned int. Use the original encoding for variable-length
32455 vectors, rather than clamping each individual element.
32456 For the second and subsequent elements in each pattern,
32457 clamp the step and base before clamping their sum.
32458 (vec_perm_indices::series_p): Handle polynomial element counts.
32459 (vec_perm_indices::all_in_range_p): Likewise.
32460 (vec_perm_indices_to_tree): Likewise.
32461 (vec_perm_indices_to_rtx): Likewise.
32462 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
32463 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
32464 (tree_vector_builder::new_binary_operation): Handle polynomial
32465 element counts. Return false if we need to know the number
32466 of elements at compile time.
32467 * fold-const.c (fold_vec_perm): Punt if the number of elements
32468 isn't known at compile time.
32470 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32472 * vec-perm-indices.h (vec_perm_builder): Change element type
32473 from HOST_WIDE_INT to poly_int64.
32474 (vec_perm_indices::element_type): Update accordingly.
32475 (vec_perm_indices::clamp): Handle polynomial element_types.
32476 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
32477 (vec_perm_indices::all_in_range_p): Likewise.
32478 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
32480 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
32481 polynomial vec_perm_indices element types.
32482 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
32483 * fold-const.c (fold_vec_perm): Likewise.
32484 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
32485 * tree-vect-generic.c (lower_vec_perm): Likewise.
32486 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
32487 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
32488 element type to HOST_WIDE_INT.
32490 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32491 Alan Hayward <alan.hayward@arm.com>
32492 David Sherwood <david.sherwood@arm.com>
32494 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
32495 rather than an int. Use plus_constant.
32496 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
32497 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
32499 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32500 Alan Hayward <alan.hayward@arm.com>
32501 David Sherwood <david.sherwood@arm.com>
32503 * calls.c (emit_call_1, expand_call): Change struct_value_size from
32504 a HOST_WIDE_INT to a poly_int64.
32506 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32507 Alan Hayward <alan.hayward@arm.com>
32508 David Sherwood <david.sherwood@arm.com>
32510 * calls.c (load_register_parameters): Cope with polynomial
32511 mode sizes. Require a constant size for BLKmode parameters
32512 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
32513 forces a parameter to be padded at the lsb end in order to
32514 fill a complete number of words, require the parameter size
32515 to be ordered wrt UNITS_PER_WORD.
32517 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32518 Alan Hayward <alan.hayward@arm.com>
32519 David Sherwood <david.sherwood@arm.com>
32521 * reload1.c (spill_stack_slot_width): Change element type
32522 from unsigned int to poly_uint64_pod.
32523 (alter_reg): Treat mode sizes as polynomial.
32525 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32526 Alan Hayward <alan.hayward@arm.com>
32527 David Sherwood <david.sherwood@arm.com>
32529 * reload.c (complex_word_subreg_p): New function.
32530 (reload_inner_reg_of_subreg, push_reload): Use it.
32532 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32533 Alan Hayward <alan.hayward@arm.com>
32534 David Sherwood <david.sherwood@arm.com>
32536 * lra-constraints.c (process_alt_operands): Reject matched
32537 operands whose sizes aren't ordered.
32538 (match_reload): Refer to this check here.
32540 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32541 Alan Hayward <alan.hayward@arm.com>
32542 David Sherwood <david.sherwood@arm.com>
32544 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
32545 that the mode size is in the set {1, 2, 4, 8, 16}.
32547 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32548 Alan Hayward <alan.hayward@arm.com>
32549 David Sherwood <david.sherwood@arm.com>
32551 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
32552 Use plus_constant instead of gen_rtx_PLUS.
32554 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32555 Alan Hayward <alan.hayward@arm.com>
32556 David Sherwood <david.sherwood@arm.com>
32558 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
32559 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
32560 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
32561 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
32562 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
32563 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
32564 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
32565 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
32566 * config/i386/i386.c (ix86_push_rounding): ...this new function.
32567 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
32569 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
32570 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
32571 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
32572 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
32573 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
32574 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
32575 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
32576 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
32577 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
32578 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
32580 * expr.c (emit_move_resolve_push): Treat the input and result
32581 of PUSH_ROUNDING as a poly_int64.
32582 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
32583 (emit_push_insn): Likewise.
32584 * lra-eliminations.c (mark_not_eliminable): Likewise.
32585 * recog.c (push_operand): Likewise.
32586 * reload1.c (elimination_effects): Likewise.
32587 * rtlanal.c (nonzero_bits1): Likewise.
32588 * calls.c (store_one_arg): Likewise. Require the padding to be
32589 known at compile time.
32591 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32592 Alan Hayward <alan.hayward@arm.com>
32593 David Sherwood <david.sherwood@arm.com>
32595 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
32596 Use plus_constant instead of gen_rtx_PLUS.
32598 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32599 Alan Hayward <alan.hayward@arm.com>
32600 David Sherwood <david.sherwood@arm.com>
32602 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
32603 rather than an int.
32605 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32606 Alan Hayward <alan.hayward@arm.com>
32607 David Sherwood <david.sherwood@arm.com>
32609 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
32610 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
32611 via stack temporaries. Treat the mode size as polynomial too.
32613 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32614 Alan Hayward <alan.hayward@arm.com>
32615 David Sherwood <david.sherwood@arm.com>
32617 * expr.c (expand_expr_real_2): When handling conversions involving
32618 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
32619 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
32620 as a poly_uint64 too.
32622 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32623 Alan Hayward <alan.hayward@arm.com>
32624 David Sherwood <david.sherwood@arm.com>
32626 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
32628 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32629 Alan Hayward <alan.hayward@arm.com>
32630 David Sherwood <david.sherwood@arm.com>
32632 * combine.c (can_change_dest_mode): Handle polynomial
32633 REGMODE_NATURAL_SIZE.
32634 * expmed.c (store_bit_field_1): Likewise.
32635 * expr.c (store_constructor): Likewise.
32636 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
32637 and polynomial REGMODE_NATURAL_SIZE.
32638 (gen_lowpart_common): Likewise.
32639 * reginfo.c (record_subregs_of_mode): Likewise.
32640 * rtlanal.c (read_modify_subreg_p): Likewise.
32642 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32643 Alan Hayward <alan.hayward@arm.com>
32644 David Sherwood <david.sherwood@arm.com>
32646 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
32647 numbers of elements.
32649 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32650 Alan Hayward <alan.hayward@arm.com>
32651 David Sherwood <david.sherwood@arm.com>
32653 * match.pd: Cope with polynomial numbers of vector elements.
32655 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32656 Alan Hayward <alan.hayward@arm.com>
32657 David Sherwood <david.sherwood@arm.com>
32659 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
32660 in a POINTER_PLUS_EXPR.
32662 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32663 Alan Hayward <alan.hayward@arm.com>
32664 David Sherwood <david.sherwood@arm.com>
32666 * omp-simd-clone.c (simd_clone_subparts): New function.
32667 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
32668 (ipa_simd_modify_function_body): Likewise.
32670 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32671 Alan Hayward <alan.hayward@arm.com>
32672 David Sherwood <david.sherwood@arm.com>
32674 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
32675 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
32676 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
32677 (expand_vector_condition, vector_element): Likewise.
32678 (subparts_gt): New function.
32679 (get_compute_type): Use subparts_gt.
32680 (count_type_subparts): Delete.
32681 (expand_vector_operations_1): Use subparts_gt instead of
32682 count_type_subparts.
32684 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32685 Alan Hayward <alan.hayward@arm.com>
32686 David Sherwood <david.sherwood@arm.com>
32688 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
32689 (vect_compile_time_alias): ...this new function. Do the calculation
32690 on poly_ints rather than trees.
32691 (vect_prune_runtime_alias_test_list): Update call accordingly.
32693 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32694 Alan Hayward <alan.hayward@arm.com>
32695 David Sherwood <david.sherwood@arm.com>
32697 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
32699 (vect_schedule_slp_instance): Likewise.
32701 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32702 Alan Hayward <alan.hayward@arm.com>
32703 David Sherwood <david.sherwood@arm.com>
32705 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
32706 constant and extern definitions for variable-length vectors.
32707 (vect_get_constant_vectors): Note that the number of units
32708 is known to be constant.
32710 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32711 Alan Hayward <alan.hayward@arm.com>
32712 David Sherwood <david.sherwood@arm.com>
32714 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
32715 of units as polynomial. Choose between WIDE and NARROW based
32718 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32719 Alan Hayward <alan.hayward@arm.com>
32720 David Sherwood <david.sherwood@arm.com>
32722 * tree-vect-stmts.c (simd_clone_subparts): New function.
32723 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
32725 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32726 Alan Hayward <alan.hayward@arm.com>
32727 David Sherwood <david.sherwood@arm.com>
32729 * tree-vect-stmts.c (vectorizable_call): Treat the number of
32730 vectors as polynomial. Use build_index_vector for
32731 IFN_GOMP_SIMD_LANE.
32733 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32734 Alan Hayward <alan.hayward@arm.com>
32735 David Sherwood <david.sherwood@arm.com>
32737 * tree-vect-stmts.c (get_load_store_type): Treat the number of
32738 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
32739 for variable-length vectors.
32740 (vectorizable_mask_load_store): Treat the number of units as
32741 polynomial, asserting that it is constant if the condition has
32742 already been enforced.
32743 (vectorizable_store, vectorizable_load): Likewise.
32745 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32746 Alan Hayward <alan.hayward@arm.com>
32747 David Sherwood <david.sherwood@arm.com>
32749 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
32750 of units as polynomial. Punt if we can't tell at compile time
32751 which vector contains the final result.
32753 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32754 Alan Hayward <alan.hayward@arm.com>
32755 David Sherwood <david.sherwood@arm.com>
32757 * tree-vect-loop.c (vectorizable_induction): Treat the number
32758 of units as polynomial. Punt on SLP inductions. Use an integer
32759 VEC_SERIES_EXPR for variable-length integer reductions. Use a
32760 cast of such a series for variable-length floating-point
32763 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32764 Alan Hayward <alan.hayward@arm.com>
32765 David Sherwood <david.sherwood@arm.com>
32767 * tree.h (build_index_vector): Declare.
32768 * tree.c (build_index_vector): New function.
32769 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
32770 of units as polynomial, forcibly converting it to a constant if
32771 vectorizable_reduction has already enforced the condition.
32772 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
32773 to create a {1,2,3,...} vector.
32774 (vectorizable_reduction): Treat the number of units as polynomial.
32775 Choose vectype_in based on the largest scalar element size rather
32776 than the smallest number of units. Enforce the restrictions
32779 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32780 Alan Hayward <alan.hayward@arm.com>
32781 David Sherwood <david.sherwood@arm.com>
32783 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
32784 number of units as polynomial.
32786 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32787 Alan Hayward <alan.hayward@arm.com>
32788 David Sherwood <david.sherwood@arm.com>
32790 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
32791 * target.def (autovectorize_vector_sizes): Return the vector sizes
32792 by pointer, using vector_sizes rather than a bitmask.
32793 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
32794 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
32795 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
32797 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
32798 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
32799 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
32800 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
32801 * omp-general.c (omp_max_vf): Likewise.
32802 * omp-low.c (omp_clause_aligned_alignment): Likewise.
32803 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
32804 * tree-vect-loop.c (vect_analyze_loop): Likewise.
32805 * tree-vect-slp.c (vect_slp_bb): Likewise.
32806 * doc/tm.texi: Regenerate.
32807 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
32809 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
32810 the vector size as a poly_uint64 rather than an unsigned int.
32811 (current_vector_size): Change from an unsigned int to a poly_uint64.
32812 (get_vectype_for_scalar_type): Update accordingly.
32813 * tree.h (build_truth_vector_type): Take the size and number of
32814 units as a poly_uint64 rather than an unsigned int.
32815 (build_vector_type): Add a temporary overload that takes
32816 the number of units as a poly_uint64 rather than an unsigned int.
32817 * tree.c (make_vector_type): Likewise.
32818 (build_truth_vector_type): Take the number of units as a poly_uint64
32819 rather than an unsigned int.
32821 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32822 Alan Hayward <alan.hayward@arm.com>
32823 David Sherwood <david.sherwood@arm.com>
32825 * target.def (get_mask_mode): Take the number of units and length
32826 as poly_uint64s rather than unsigned ints.
32827 * targhooks.h (default_get_mask_mode): Update accordingly.
32828 * targhooks.c (default_get_mask_mode): Likewise.
32829 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
32830 * doc/tm.texi: Regenerate.
32832 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32833 Alan Hayward <alan.hayward@arm.com>
32834 David Sherwood <david.sherwood@arm.com>
32836 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
32837 * omp-general.c (omp_max_vf): Likewise.
32838 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
32839 (expand_omp_simd): Handle polynomial safelen.
32840 * omp-low.c (omplow_simd_context): Add a default constructor.
32841 (omplow_simd_context::max_vf): Change from int to poly_uint64.
32842 (lower_rec_simd_input_clauses): Update accordingly.
32843 (lower_rec_input_clauses): Likewise.
32845 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32846 Alan Hayward <alan.hayward@arm.com>
32847 David Sherwood <david.sherwood@arm.com>
32849 * tree-vectorizer.h (vect_nunits_for_cost): New function.
32850 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
32851 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
32852 (vect_analyze_slp_cost): Likewise.
32853 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
32854 (vect_model_load_cost): Likewise.
32856 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32857 Alan Hayward <alan.hayward@arm.com>
32858 David Sherwood <david.sherwood@arm.com>
32860 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
32861 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
32862 from an unsigned int * to a poly_uint64_pod *.
32863 (calculate_unrolling_factor): New function.
32864 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
32866 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32867 Alan Hayward <alan.hayward@arm.com>
32868 David Sherwood <david.sherwood@arm.com>
32870 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
32871 from an unsigned int to a poly_uint64.
32872 (_loop_vec_info::slp_unrolling_factor): Likewise.
32873 (_loop_vec_info::vectorization_factor): Change from an int
32875 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
32876 (vect_get_num_vectors): New function.
32877 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
32878 (vect_get_num_copies): Use vect_get_num_vectors.
32879 (vect_analyze_data_ref_dependences): Change max_vf from an int *
32880 to an unsigned int *.
32881 (vect_analyze_data_refs): Change min_vf from an int * to a
32883 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
32884 than an unsigned HOST_WIDE_INT.
32885 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
32886 (vect_analyze_data_ref_dependence): Change max_vf from an int *
32887 to an unsigned int *.
32888 (vect_analyze_data_ref_dependences): Likewise.
32889 (vect_compute_data_ref_alignment): Handle polynomial vf.
32890 (vect_enhance_data_refs_alignment): Likewise.
32891 (vect_prune_runtime_alias_test_list): Likewise.
32892 (vect_shift_permute_load_chain): Likewise.
32893 (vect_supportable_dr_alignment): Likewise.
32894 (dependence_distance_ge_vf): Take the vectorization factor as a
32895 poly_uint64 rather than an unsigned HOST_WIDE_INT.
32896 (vect_analyze_data_refs): Change min_vf from an int * to a
32898 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
32899 vfm1 as a poly_uint64 rather than an int. Make the same change
32900 for the returned bound_scalar.
32901 (vect_gen_vector_loop_niters): Handle polynomial vf.
32902 (vect_do_peeling): Likewise. Update call to
32903 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
32904 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
32906 * tree-vect-loop.c (vect_determine_vectorization_factor)
32907 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
32908 (vect_get_known_peeling_cost): Likewise.
32909 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
32910 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
32911 (vect_transform_loop): Likewise. Use the lowest possible VF when
32912 updating the upper bounds of the loop.
32913 (vect_min_worthwhile_factor): Make static. Return an unsigned int
32914 rather than an int.
32915 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
32916 polynomial unroll factors.
32917 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
32918 (vect_make_slp_decision): Likewise.
32919 (vect_supported_load_permutation_p): Likewise, and polynomial
32921 (vect_analyze_slp_cost): Handle polynomial vf.
32922 (vect_slp_analyze_node_operations): Likewise.
32923 (vect_slp_analyze_bb_1): Likewise.
32924 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
32925 than an unsigned HOST_WIDE_INT.
32926 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
32927 (vectorizable_load): Handle polynomial vf.
32928 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
32930 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
32932 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32933 Alan Hayward <alan.hayward@arm.com>
32934 David Sherwood <david.sherwood@arm.com>
32936 * match.pd: Handle bit operations involving three constants
32937 and try to fold one pair.
32939 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32941 * tree-vect-loop-manip.c: Include gimple-fold.h.
32942 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
32943 niters_maybe_zero parameters. Handle other cases besides a step of 1.
32944 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
32945 Add a path that uses a step of VF instead of 1, but disable it
32947 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
32948 and niters_no_overflow parameters. Update calls to
32949 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
32950 Create a new SSA name if the latter choses to use a ste other
32951 than zero, and return it via niters_vector_mult_vf_var.
32952 * tree-vect-loop.c (vect_transform_loop): Update calls to
32953 vect_do_peeling, vect_gen_vector_loop_niters and
32954 slpeel_make_loop_iterate_ntimes.
32955 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
32956 (vect_gen_vector_loop_niters): Update declarations after above changes.
32958 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
32960 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
32961 128-bit round to integer instructions.
32962 (ceil<mode>2): Likewise.
32963 (btrunc<mode>2): Likewise.
32964 (round<mode>2): Likewise.
32966 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32968 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
32969 unaligned VSX load/store on P8/P9.
32970 (expand_block_clear): Allow the use of unaligned VSX
32971 load/store on P8/P9.
32973 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32975 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
32977 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
32978 swap associated with both a load and a store.
32980 2018-01-02 Andrew Waterman <andrew@sifive.com>
32982 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
32983 * config/riscv/riscv.md (clear_cache): Use it.
32985 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
32987 * web.c: Remove out-of-date comment.
32989 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32991 * expr.c (fixup_args_size_notes): Check that any existing
32992 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
32993 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
32994 (emit_single_push_insn): ...here.
32996 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32998 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
32999 (const_vector_encoded_nelts): New function.
33000 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
33001 (const_vector_int_elt, const_vector_elt): Declare.
33002 * emit-rtl.c (const_vector_int_elt_1): New function.
33003 (const_vector_elt): Likewise.
33004 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
33005 of CONST_VECTOR_ELT.
33007 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33009 * expr.c: Include rtx-vector-builder.h.
33010 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
33011 directly on the tree encoding.
33012 (const_vector_from_tree): Likewise.
33013 * optabs.c: Include rtx-vector-builder.h.
33014 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
33015 sequence of "u" values.
33016 * vec-perm-indices.c: Include rtx-vector-builder.h.
33017 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
33018 directly on the vec_perm_indices encoding.
33020 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33022 * doc/rtl.texi (const_vector): Describe new encoding scheme.
33023 * Makefile.in (OBJS): Add rtx-vector-builder.o.
33024 * rtx-vector-builder.h: New file.
33025 * rtx-vector-builder.c: Likewise.
33026 * rtl.h (rtx_def::u2): Add a const_vector field.
33027 (CONST_VECTOR_NPATTERNS): New macro.
33028 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
33029 (CONST_VECTOR_DUPLICATE_P): Likewise.
33030 (CONST_VECTOR_STEPPED_P): Likewise.
33031 (CONST_VECTOR_ENCODED_ELT): Likewise.
33032 (const_vec_duplicate_p): Check for a duplicated vector encoding.
33033 (unwrap_const_vec_duplicate): Likewise.
33034 (const_vec_series_p): Check for a non-duplicated vector encoding.
33035 Say that the function only returns true for integer vectors.
33036 * emit-rtl.c: Include rtx-vector-builder.h.
33037 (gen_const_vec_duplicate_1): Delete.
33038 (gen_const_vector): Call gen_const_vec_duplicate instead of
33039 gen_const_vec_duplicate_1.
33040 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
33041 (gen_const_vec_duplicate): Use rtx_vector_builder.
33042 (gen_const_vec_series): Likewise.
33043 (gen_rtx_CONST_VECTOR): Likewise.
33044 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
33045 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
33046 Build a new vector rather than modifying a CONST_VECTOR in-place.
33047 (handle_special_swappables): Update call accordingly.
33048 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
33049 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
33050 Build a new vector rather than modifying a CONST_VECTOR in-place.
33051 (handle_special_swappables): Update call accordingly.
33053 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33055 * simplify-rtx.c (simplify_const_binary_operation): Use
33056 CONST_VECTOR_ELT instead of XVECEXP.
33058 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33060 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
33061 the selector elements to be different from the data elements
33062 if the selector is a VECTOR_CST.
33063 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
33064 ssizetype for the selector.
33066 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33068 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
33069 before testing each element individually.
33070 * tree-vect-generic.c (lower_vec_perm): Likewise.
33072 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33074 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
33075 * selftest-run-tests.c (selftest::run_tests): Call it.
33076 * vector-builder.h (vector_builder::operator ==): New function.
33077 (vector_builder::operator !=): Likewise.
33078 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
33079 (vec_perm_indices::all_from_input_p): New function.
33080 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
33081 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
33082 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
33083 instead of reading the VECTOR_CST directly. Detect whether both
33084 vector inputs are the same before constructing the vec_perm_indices,
33085 and update the number of inputs argument accordingly. Use the
33086 utility functions added above. Only construct sel2 if we need to.
33088 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33090 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
33091 the broadcast of the low byte.
33092 (expand_mult_highpart): Use an explicit encoding for the permutes.
33093 * optabs-query.c (can_mult_highpart_p): Likewise.
33094 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
33095 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
33096 (vectorizable_bswap): Likewise.
33097 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
33098 explicit encoding for the power-of-2 permutes.
33099 (vect_permute_store_chain): Likewise.
33100 (vect_grouped_load_supported): Likewise.
33101 (vect_permute_load_chain): Likewise.
33103 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33105 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
33106 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
33107 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
33108 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
33109 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
33110 (vect_gen_perm_mask_any): Likewise.
33112 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33114 * int-vector-builder.h: New file.
33115 * vec-perm-indices.h: Include int-vector-builder.h.
33116 (vec_perm_indices): Redefine as an int_vector_builder.
33117 (auto_vec_perm_indices): Delete.
33118 (vec_perm_builder): Redefine as a stand-alone class.
33119 (vec_perm_indices::vec_perm_indices): New function.
33120 (vec_perm_indices::clamp): Likewise.
33121 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
33122 (vec_perm_indices::new_vector): New function.
33123 (vec_perm_indices::new_expanded_vector): Update for new
33124 vec_perm_indices class.
33125 (vec_perm_indices::rotate_inputs): New function.
33126 (vec_perm_indices::all_in_range_p): Operate directly on the
33127 encoded form, without computing elided elements.
33128 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
33129 encoding. Update for new vec_perm_indices class.
33130 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
33131 the given vec_perm_builder.
33132 (expand_vec_perm_var): Update vec_perm_builder constructor.
33133 (expand_mult_highpart): Use vec_perm_builder instead of
33134 auto_vec_perm_indices.
33135 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
33136 vec_perm_indices instead of auto_vec_perm_indices. Use a single
33137 or double series encoding as appropriate.
33138 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
33139 vec_perm_indices instead of auto_vec_perm_indices.
33140 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
33141 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
33142 (vect_permute_store_chain): Likewise.
33143 (vect_grouped_load_supported): Likewise.
33144 (vect_permute_load_chain): Likewise.
33145 (vect_shift_permute_load_chain): Likewise.
33146 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
33147 (vect_transform_slp_perm_load): Likewise.
33148 (vect_schedule_slp_instance): Likewise.
33149 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
33150 (vectorizable_mask_load_store): Likewise.
33151 (vectorizable_bswap): Likewise.
33152 (vectorizable_store): Likewise.
33153 (vectorizable_load): Likewise.
33154 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
33155 vec_perm_indices instead of auto_vec_perm_indices. Use
33156 tree_to_vec_perm_builder to read the vector from a tree.
33157 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
33158 vec_perm_builder instead of a vec_perm_indices.
33159 (have_whole_vector_shift): Use vec_perm_builder and
33160 vec_perm_indices instead of auto_vec_perm_indices. Leave the
33161 truncation to calc_vec_perm_mask_for_shift.
33162 (vect_create_epilog_for_reduction): Likewise.
33163 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
33164 from auto_vec_perm_indices to vec_perm_indices.
33165 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
33166 instead of changing individual elements.
33167 (aarch64_vectorize_vec_perm_const): Use new_vector to install
33168 the vector in d.perm.
33169 * config/arm/arm.c (expand_vec_perm_d::perm): Change
33170 from auto_vec_perm_indices to vec_perm_indices.
33171 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
33172 instead of changing individual elements.
33173 (arm_vectorize_vec_perm_const): Use new_vector to install
33174 the vector in d.perm.
33175 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
33176 Update vec_perm_builder constructor.
33177 (rs6000_expand_interleave): Likewise.
33178 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
33179 (rs6000_expand_interleave): Likewise.
33181 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33183 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
33184 to qimode could truncate the indices.
33185 * optabs.c (expand_vec_perm_var): Likewise.
33187 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33189 * Makefile.in (OBJS): Add vec-perm-indices.o.
33190 * vec-perm-indices.h: New file.
33191 * vec-perm-indices.c: Likewise.
33192 * target.h (vec_perm_indices): Replace with a forward class
33194 (auto_vec_perm_indices): Move to vec-perm-indices.h.
33195 * optabs.h: Include vec-perm-indices.h.
33196 (expand_vec_perm): Delete.
33197 (selector_fits_mode_p, expand_vec_perm_var): Declare.
33198 (expand_vec_perm_const): Declare.
33199 * target.def (vec_perm_const_ok): Replace with...
33200 (vec_perm_const): ...this new hook.
33201 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
33202 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
33203 * doc/tm.texi: Regenerate.
33204 * optabs.def (vec_perm_const): Delete.
33205 * doc/md.texi (vec_perm_const): Likewise.
33206 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
33207 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
33208 expand_vec_perm for constant permutation vectors. Assert that
33209 the mode of variable permutation vectors is the integer equivalent
33210 of the mode that is being permuted.
33211 * optabs-query.h (selector_fits_mode_p): Declare.
33212 * optabs-query.c: Include vec-perm-indices.h.
33213 (selector_fits_mode_p): New function.
33214 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
33215 is defined, instead of checking whether the vec_perm_const_optab
33216 exists. Use targetm.vectorize.vec_perm_const instead of
33217 targetm.vectorize.vec_perm_const_ok. Check whether the indices
33218 fit in the vector mode before using a variable permute.
33219 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
33220 vec_perm_indices instead of an rtx.
33221 (expand_vec_perm): Replace with...
33222 (expand_vec_perm_const): ...this new function. Take the selector
33223 as a vec_perm_indices rather than an rtx. Also take the mode of
33224 the selector. Update call to shift_amt_for_vec_perm_mask.
33225 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
33226 Use vec_perm_indices::new_expanded_vector to expand the original
33227 selector into bytes. Check whether the indices fit in the vector
33228 mode before using a variable permute.
33229 (expand_vec_perm_var): Make global.
33230 (expand_mult_highpart): Use expand_vec_perm_const.
33231 * fold-const.c: Includes vec-perm-indices.h.
33232 * tree-ssa-forwprop.c: Likewise.
33233 * tree-vect-data-refs.c: Likewise.
33234 * tree-vect-generic.c: Likewise.
33235 * tree-vect-loop.c: Likewise.
33236 * tree-vect-slp.c: Likewise.
33237 * tree-vect-stmts.c: Likewise.
33238 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
33240 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
33241 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
33242 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
33243 (aarch64_vectorize_vec_perm_const): ...this new function.
33244 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
33245 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
33246 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
33247 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
33248 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
33249 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
33250 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
33252 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
33253 check for NEON modes.
33254 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
33255 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
33256 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
33257 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
33259 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
33260 the old VEC_PERM_CONST conditions.
33261 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
33262 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
33263 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
33264 (ia64_vectorize_vec_perm_const_ok): Merge into...
33265 (ia64_vectorize_vec_perm_const): ...this new function.
33266 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
33267 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
33268 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
33269 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
33270 * config/mips/mips.c (mips_expand_vec_perm_const)
33271 (mips_vectorize_vec_perm_const_ok): Merge into...
33272 (mips_vectorize_vec_perm_const): ...this new function.
33273 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
33274 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
33275 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
33276 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
33277 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
33278 (rs6000_expand_vec_perm_const): Delete.
33279 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
33281 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
33282 (altivec_expand_vec_perm_const_le): Take each operand individually.
33283 Operate on constant selectors rather than rtxes.
33284 (altivec_expand_vec_perm_const): Likewise. Update call to
33285 altivec_expand_vec_perm_const_le.
33286 (rs6000_expand_vec_perm_const): Delete.
33287 (rs6000_vectorize_vec_perm_const_ok): Delete.
33288 (rs6000_vectorize_vec_perm_const): New function.
33289 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
33290 an element count and rtx array.
33291 (rs6000_expand_extract_even): Update call accordingly.
33292 (rs6000_expand_interleave): Likewise.
33293 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
33294 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
33295 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
33296 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
33297 (rs6000_expand_vec_perm_const): Delete.
33298 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
33299 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
33300 (altivec_expand_vec_perm_const_le): Take each operand individually.
33301 Operate on constant selectors rather than rtxes.
33302 (altivec_expand_vec_perm_const): Likewise. Update call to
33303 altivec_expand_vec_perm_const_le.
33304 (rs6000_expand_vec_perm_const): Delete.
33305 (rs6000_vectorize_vec_perm_const_ok): Delete.
33306 (rs6000_vectorize_vec_perm_const): New function. Remove stray
33307 reference to the SPE evmerge intructions.
33308 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
33309 an element count and rtx array.
33310 (rs6000_expand_extract_even): Update call accordingly.
33311 (rs6000_expand_interleave): Likewise.
33312 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
33313 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
33315 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
33317 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33319 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
33320 vector mode and that that mode matches the mode of the data
33322 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
33323 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
33324 directly using expand_vec_perm_1 when forcing selectors into
33326 (expand_vec_perm_var): New function, split out from expand_vec_perm.
33328 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33330 * optabs-query.h (can_vec_perm_p): Delete.
33331 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
33332 * optabs-query.c (can_vec_perm_p): Split into...
33333 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
33334 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
33335 particular selector is valid.
33336 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
33337 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
33338 (vect_grouped_load_supported): Likewise.
33339 (vect_shift_permute_load_chain): Likewise.
33340 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
33341 (vect_transform_slp_perm_load): Likewise.
33342 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
33343 (vectorizable_bswap): Likewise.
33344 (vect_gen_perm_mask_checked): Likewise.
33345 * fold-const.c (fold_ternary_loc): Likewise. Don't take
33346 implementations of variable permutation vectors into account
33347 when deciding which selector to use.
33348 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
33349 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
33350 with a false third argument.
33351 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
33352 to test whether the constant selector is valid and can_vec_perm_var_p
33353 to test whether a variable selector is valid.
33355 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33357 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
33358 * optabs-query.c (can_vec_perm_p): Likewise.
33359 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
33360 instead of vec_perm_indices.
33361 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
33362 (vect_gen_perm_mask_checked): Likewise,
33363 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
33364 (vect_gen_perm_mask_checked): Likewise,
33366 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33368 * optabs-query.h (qimode_for_vec_perm): Declare.
33369 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
33370 (qimode_for_vec_perm): ...this new function.
33371 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
33373 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33375 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
33376 does not have a conditional at the top.
33378 2018-01-02 Richard Biener <rguenther@suse.de>
33380 * ipa-inline.c (big_speedup_p): Fix expression.
33382 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
33385 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
33388 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
33392 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
33393 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
33394 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
33395 cond_taken_branch_cost 3->4.
33397 2018-01-01 Jakub Jelinek <jakub@redhat.com>
33399 PR tree-optimization/83581
33400 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
33401 TODO_cleanup_cfg if any changes have been made.
33403 PR middle-end/83608
33404 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
33405 convert_modes if target mode has the right side, but different mode
33408 PR middle-end/83609
33409 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
33410 last argument when extracting from CONCAT. If either from_real or
33411 from_imag is NULL, use expansion through memory. If result is not
33412 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
33413 the parts directly to inner mode, if even that fails, use expansion
33416 PR middle-end/83623
33417 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
33418 check for bswap in mode rather than HImode and use that in expand_unop
33421 Copyright (C) 2018 Free Software Foundation, Inc.
33423 Copying and distribution of this file, with or without modification,
33424 are permitted in any medium without royalty provided the copyright
33425 notice and this notice are preserved.